1 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3 * except.c (setjmp_fn): New global variable.
4 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6 if DONT_USE_BUILTIN_SETJMP is defined.
8 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10 * target.def (insn_cost): New hook.
11 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12 * doc/tm.texi: Regenerate.
13 * rtlanal.c (insn_cost): Use the new hook.
15 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
17 * combine.c (combine_validate_cost): Compute the new insn_cost,
18 not just pattern_cost.
19 (try_combine): Adjust comment.
21 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
23 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
25 * combine.c (uid_insn_cost): Adjust comment.
26 (combine_validate_cost): Adjust comment. Use pattern_cost instead
28 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
29 * dse.c (find_shift_sequence): Ditto.
30 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
31 (bb_valid_for_noce_process_p): Use pattern_cost.
32 * rtl.h (insn_rtx_cost): Delete.
33 (pattern_cost): New prototype.
34 (insn_cost): New prototype.
35 * rtlanal.c (insn_rtx_cost): Rename to...
36 (pattern_cost): ... this.
39 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
41 * config/i386/i386.md (*jcc_2): Remove insn pattern.
42 (*jcc<mode>_0_r_i387): Ditto.
43 (*jccxf_r_i387): Ditto.
44 (*jcc<mode>_r_i387): Ditto.
45 (*jccu<mode>_r_i387): Ditto.
46 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
47 (*jcc): Rename from *jcc_1.
49 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
51 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
52 deferred rescans after the lvx/stvx recombination pre-pass.
54 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
56 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
57 memory operation instruction support.
58 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
59 (rs6000-ibm-aix[789]*): Likewise.
60 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
61 Document new functions.
63 2017-10-09 Richard Biener <rguenther@suse.de>
65 PR tree-optimization/82397
66 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
67 equality only for semantically equal trees.
69 2017-10-09 Richard Biener <rguenther@suse.de>
71 PR tree-optimization/82449
72 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
73 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
74 allow constant addresses.
75 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
78 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
80 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
83 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
86 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
89 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
92 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
94 2017-10-09 Jakub Jelinek <jakub@redhat.com>
97 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
98 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
100 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
102 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
103 (WI_BINARY_PREDICATE_RESULT): Likewise.
104 (wi::binary_traits::operator_result): New type.
105 (wi::binary_traits::predicate_result): Likewise.
106 (generic_wide_int::operator~, unary generic_wide_int::operator-)
107 (generic_wide_int::operator==, generic_wide_int::operator!=)
108 (generic_wide_int::operator&, generic_wide_int::and_not)
109 (generic_wide_int::operator|, generic_wide_int::or_not)
110 (generic_wide_int::operator^, generic_wide_int::operator+
111 (binary generic_wide_int::operator-, generic_wide_int::operator*):
113 (operator~, unary operator-, operator==, operator!=, operator&)
114 (operator|, operator^, operator+, binary operator-, operator*): New
116 * expr.c (get_inner_reference): Use wi::bit_and_not.
117 * fold-const.c (fold_binary_loc): Likewise.
118 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
119 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
120 (bit_value_binop): Likewise.
121 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
122 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
123 (extract_range_from_binary_expr_1): Likewise.
124 (masked_increment): Likewise.
125 (simplify_bit_ops_using_ranges): Likewise.
127 2017-10-09 Martin Jambor <mjambor@suse.cz>
130 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
131 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
132 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
134 (hsa_fixup_mov_insn_type): New function.
135 (hsa_op_with_type::get_in_type): Use it.
136 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
137 immediates in an assert.
138 (hsa_op_with_type::extend_int_to_32bit): New method.
139 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
140 types. Convert to dest type if necessary.
141 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
142 (reg_for_gimple_ssa): Pass false as min32int to
143 hsa_type_for_scalar_tree_type.
144 (gen_hsa_addr): Fixup type when creating addresable temporary.
145 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
146 (gen_hsa_unary_operation): Extend operands and convert to dest type if
147 necessary. Call hsa_fixup_mov_insn_type.
148 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
149 extend operands and convert to dest type if necessary.
150 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
151 to dest type if necessary.
152 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
153 if conversion nt necessary and size matches.
154 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
155 to dest type if necessary.
156 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
157 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
159 (gen_hsa_clrsb): Likewise.
160 (gen_hsa_ffs): Likewise.
161 (gen_hsa_divmod): Extend operands and convert to dest type if
163 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
165 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
167 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
168 Remove empty default arguments. Use a brace block as output
170 (conditional return): Ditto.
172 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
173 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
175 (group_ending_nop): Ditto.
177 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
178 (splitters for those): Ditto.
180 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
182 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
183 a conditional jump (and the compare for it) so that pc_rtx is the
185 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
186 for the deleted and renamed ctr<mode>_internal[234] patterns.
187 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
188 Delete second conditional return pattern.
189 (ctr<mode>_internal2): Delete this second bdnz pattern.
190 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
191 (ctr<mode>_internal4): Delete this second bdz pattern.
193 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
195 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
196 (always_initialized_rtx_for_ssa_name_p): New predicate.
197 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
198 (finish_out_of_ssa): Free new field of SA.
199 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
200 * tree-ssa-coalesce.c: Include tree-ssa.h.
201 (get_parm_default_def_partitions): Remove extern keyword.
202 (get_undefined_value_partitions): New function.
203 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
204 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
206 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
208 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
210 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
212 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
214 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
217 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
219 * i386.c (has_dispatch): Disable for Ryzen.
221 2017-10-08 Olivier Hainque <hainque@adacore.com>
223 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
224 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
226 (thumb_set_return_address): Likewise.
228 2017-10-08 Olivier Hainque <hainque@adacore.com>
230 * common/config/arm/arm-common.c (arm_except_unwind_info):
231 Handle DWARF2_UNWIND_INFO.
233 2017-10-07 Michael Collison <michael.collison@arm.com>
235 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
238 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
240 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
241 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
242 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
243 defined, force the creation of a new block for a dispatch label.
245 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
247 * invoke.texi (Wsuggest-attribute=cold): Document.
248 * common.opt (Wsuggest-attribute=cold): New
249 * ipa-pure-const.c (warn_function_cold): New function.
250 * predict.c (compute_function_frequency): Use it.
251 * predict.h (warn_function_cold): Declare.
253 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
255 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
258 2017-10-06 Martin Liska <mliska@suse.cz>
260 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
261 keyword for member functions.
262 (struct sanopt_tree_couple): New struct.
263 (struct sanopt_tree_couple_hash): New function.
264 (struct sanopt_ctx): Add new hash_map.
265 (has_dominating_ubsan_ptr_check): New function.
266 (record_ubsan_ptr_check_stmt): Likewise.
267 (maybe_optimize_ubsan_ptr_ifn): Likewise.
268 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
269 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
271 2017-10-06 Sudakshina Das <sudi.das@arm.com>
274 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
275 aarch64_simd_valid_immediate on CONST_VECTORs.
276 (aarch64_reg_or_bic_imm): Likewise.
278 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
280 PR rtl-optimization/82396
281 * haifa-sched.c (ready_sort_real): Disable qsort checking.
283 2017-10-06 Sebastian Pop <sebpop@gmail.com>
285 * graphite-dependences.c (scop_get_reads): Move code to...
286 (scop_get_must_writes): Move code to...
287 (scop_get_may_writes): Move code to...
288 (scop_get_reads_and_writes): ... here.
289 (scop_get_dependences): Call scop_get_reads_and_writes.
291 2017-10-06 Jakub Jelinek <jakub@redhat.com>
293 PR tree-optimization/82434
294 * fold-const.h (can_native_encode_type_p,
295 can_native_encode_string_p): Remove.
296 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
297 don't encode anything, just return what would be otherwise returned.
298 (native_encode_fixed, native_encode_complex, native_encode_vector):
300 (native_encode_string): Likewise. Inline by hand
301 can_native_encode_string_p.
302 (can_native_encode_type_p): Remove.
303 (can_native_encode_string_p): Remove.
304 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
305 STRING_CSTs using can_native_encode_string_p, test all
306 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
307 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
308 argument from native_encode_expr.
309 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
310 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
313 2017-10-06 Richard Biener <rguenther@suse.de>
315 PR tree-optimization/82397
316 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
317 operand_equal_p but rely on data_ref_compare_tree for detecting
319 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
320 to match up with dr_group_sort_cmp.
322 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
325 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
327 * config/s390/s390-builtin-types.def: Regenerate.
329 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
332 * config/s390/s390-builtin-types.def: Regenerate.
333 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
334 Change flag from B_VXE to B_VX.
335 (s390_vec_min_dbl): Remove B_VXE flag.
337 2017-10-06 Richard Biener <rguenther@suse.de>
339 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
340 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
341 translate_isl_ast_to_gimple::is_valid_rename,
342 translate_isl_ast_to_gimple::get_rename,
343 translate_isl_ast_to_gimple::get_def_bb_for_const,
344 translate_isl_ast_to_gimple::get_new_name,
345 translate_isl_ast_to_gimple::collect_all_ssa_names,
346 translate_isl_ast_to_gimple::copy_loop_phi_args,
347 translate_isl_ast_to_gimple::collect_all_ssa_names,
348 translate_isl_ast_to_gimple::copy_loop_phi_args,
349 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
350 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
351 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
352 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
353 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
354 translate_isl_ast_to_gimple::copy_cond_phi_args,
355 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
356 translate_isl_ast_to_gimple::edge_for_new_close_phis,
357 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
358 translate_isl_ast_to_gimple::rename_uses,
359 translate_isl_ast_to_gimple::rename_all_uses): Remove.
360 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
361 (set_rename_for_each_def): Likewise.
362 (graphite_copy_stmts_from_block): Handle debug stmt resetting
363 here. Handle rewriting SCEV analyzable uses here.
364 (copy_bb_and_scalar_dependences): Generate code for PHI
366 (graphite_regenerate_ast_isl): Adjust.
367 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
368 (add_write, add_read): New functions.
369 (build_cross_bb_scalars_def): Use it and simplify.
370 (build_cross_bb_scalars_use): Likewise.
371 (graphite_find_cross_bb_scalar_vars): Inline into...
372 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
373 simulating out-of-SSA. Compute liveout and add dependencies.
374 (build_scops): Force an empty entry block.
375 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
377 (sese_build_liveouts): Declare.
378 (sese_trivially_empty_bb_p): Likewise.
379 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
380 compute liveout and debug_liveout.
381 (sese_bad_liveouts_use): Remove.
382 (sese_reset_debug_liveouts_bb): Likewise.
383 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
384 (sese_build_liveouts): Build liveout and debug_liveout and store
386 (new_sese_info): Adjust.
387 (free_sese_info): Likewise.
388 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
389 do not build liveout here.
390 (move_sese_in_condition): Adjust region entry.
391 (scev_analyzable_p): Match up with chrec_apply requirements.
392 (sese_trivially_empty_bb_p): New.
393 * tree-into-ssa.c (get_reaching_def): Properly support generating
394 default-defs for incremental rewrite of anonymous names.
396 2017-10-06 Richard Biener <rguenther@suse.de>
398 * graphite-sese-to-poly.c (extract_affine): For casts increasing
399 precision do not perform modulo reduction.
401 2017-10-06 Richard Biener <rguenther@suse.de>
403 PR tree-optimization/82436
404 * tree-vect-slp.c (vect_supported_load_permutation_p): More
405 conservatively choose the vectorization factor when checking
406 whether we can perform the required load permutation.
407 (vect_transform_slp_perm_load): Assert when we may not fail.
409 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
411 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
412 message for incompatible -msdata=* and -mcall-* options.
414 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
416 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
417 for post-reload scheduling.
419 2017-10-05 Tamar Christina <tamar.christina@arm.com>
421 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
423 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
425 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
426 to improve monte carlo in scimark.
428 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
430 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
431 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
432 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
433 pentium4_cost, nocona_cost): Set reassociation width to 1.
434 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
435 width to 2 for fp operations and 1 otherwise.
436 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
438 (atom_cost): Set reassociation width to 2.
439 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
440 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
441 (core_cost): Set fp reassociation width to 4 and vector to 2.
442 (ix86_reassociation_width): Rewrite using cost table; special case
443 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
444 and TARGET_AVX128_OPTIMAL.
445 * i386.h (processor_costs): Add
446 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
447 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
448 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
449 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
450 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
451 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
453 2017-10-05 Nathan Sidwell <nathan@acm.org>
455 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
457 2017-10-05 Tamar Christina <tamar.christina@arm.com>
459 * config/arm/arm.c (arm_test_fpu_data): New.
460 (arm_run_selftests): Call arm_test_fpu_data.
462 2017-10-04 Nathan Sidwell <nathan@acm.org>
464 * toplev.c (toplev::main): Remove excess parens on pretty_printer
466 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
468 2017-10-04 Sudakshina Das <sudi.das@arm.com>
470 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
471 check type for aarch64_simd_valid_immediate.
472 (aarch64_output_simd_mov_immediate): Update prototype.
473 (aarch64_simd_valid_immediate): Update prototype.
474 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
475 support for ORR-immediate.
476 (and<mode>3): modified pattern to add support for BIC-immediate.
477 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
478 now checks for valid immediate for BIC and ORR based on new enum
480 (aarch64_output_simd_mov_immediate): Function now used to output
481 BIC/ORR imm as well based on new enum argument.
482 * config/aarch64/constraints.md (Do): New vector immediate constraint.
484 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
485 (aarch64_reg_or_bic_imm): Likewise.
487 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
489 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
490 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
492 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
496 PR rtl-optimization/82396
497 * haifa-sched.c (autopref_multipass_init): Simplify
499 (autopref_rank_data): Simplify sort order.
500 * sched-int.h (autopref_multipass_data_): Remove
501 multi_mem_insn_p, min_offset and max_offset.
503 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
505 * doc/sourcebuild.texi: Document vect_peeling_profitable.
507 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
509 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
512 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
514 * doc/sourcebuild.texi: Document vect_long_mult.
516 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
518 PR tree-optimization/82413
519 * fold-const.c (build_range_check): Use widest_int when comparing
520 the maximum ETYPE value with HIGH.
522 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
524 PR rtl-optimization/82396
525 * haifa-sched.c (autopref_multipass_init): Simplify
527 (autopref_rank_data): Simplify sort order.
528 * sched-int.h (autopref_multipass_data_): Remove
529 multi_mem_insn_p, min_offset and max_offset.
531 2017-10-04 Jakub Jelinek <jakub@redhat.com>
533 PR tree-optimization/82381
534 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
535 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
538 PR tree-optimization/82374
539 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
540 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
541 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
542 current_function_decl to the new decl.
544 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
546 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
547 helper macro for IEEE float128 hardware built-in functions.
548 (SQRTF128_ODD): Add built-in functions with the round-to-odd
550 (TRUNCF128_ODD): Likewise.
551 (ADDF128_ODD): Likewise.
552 (SUBF128_ODD): Likewise.
553 (MULF128_ODD): Likewise.
554 (DIVF128_ODD): Likewise.
555 (FMAF128_ODD): Likewise.
556 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
557 UNSPEC_TRUNC_ROUND_TO_ODD.
558 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
559 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
560 floating point round to odd instructions.
561 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
562 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
563 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
564 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
565 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
566 (trunc<mode>sf2_hw): Change the truncate with round to odd
567 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
568 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
569 to odd hardware instructions.
570 (sub<mode>3_odd): Likewise.
571 (mul<mode>3_odd): Likewise.
572 (div<mode>3_odd): Likewise.
573 (sqrt<mode>2_odd): Likewise.
574 (fma<mode>4_odd): Likewise.
575 (fms<mode>4_odd): Likewise.
576 (nfma<mode>4_odd): Likewise.
577 (nfms<mode>4_odd): Likewise.
578 (trunc<mode>df2_odd): Change the truncate with round to odd
579 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
581 * doc/extend.texi (PowerPC built-in functions): Update documentation
582 for existing IEEE float128-bit built-in functions. Add built-in
583 functions that generate the IEEE 128-bit floating point round to
586 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
588 PR rtl-optimization/77729
589 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
590 to (X&(C1&~C2))|C2 transformations.
592 2017-10-03 Martin Jambor <mjambor@suse.cz>
594 PR tree-optimization/82363
595 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
596 mismatch, mark lacc written regardless of racc.
598 2017-10-03 Jakub Jelinek <jakub@redhat.com>
600 PR tree-optimization/82381
601 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
602 stmt_to_insert nor wheather SSA_NAMEs are default defs.
603 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
604 fallthrough into reassoc_stmt_dominates_stmt_p.
607 * combine.c (combine_instructions): Don't combine in unreachable
610 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
613 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
614 function to not use the have_cpu variable. Do not set cpu_index,
615 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
617 (rs6000_valid_attribute_p): Remove duplicate initializations of
618 old_optimize and func_optimize.
619 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
620 (rs6000_activate_target_options): Make global.
621 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
624 2017-10-02 Jakub Jelinek <jakub@redhat.com>
626 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
627 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
628 Set *poffset to 0 and *psize and *pmax_size to -1 if
629 *poffset + *psize overflows in HOST_WIDE_INT.
631 PR tree-optimization/82387
632 PR tree-optimization/82388
633 PR tree-optimization/82389
634 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
635 instead of live_bytes non-NULL.
637 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
640 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
643 2017-10-02 Richard Biener <rguenther@suse.de>
645 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
646 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
647 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
649 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
651 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
652 requested precision matches the type's.
653 * calls.c (alloc_max_size): Calculate the new candidate size as
654 a widest_int and use wi::to_widest when comparing it with the
655 current candidate size.
656 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
657 zero rather than integer_zero_node.
658 * match.pd: Check for a no-op conversion before using wi::add
659 rather than after. Use tree_to_uhwi when summing small shift
660 counts into an unsigned int.
662 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
663 Alan Hayward <alan.hayward@arm.com>
664 David Sherwood <david.sherwood@arm.com>
667 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
668 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
669 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
672 2017-10-02 Richard Biener <rguenther@suse.de>
674 PR tree-optimization/82355
675 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
676 a mapping for the enclosing loop but avoid generating one for
678 (copy_bb_and_scalar_dependences): Remove premature codegen
679 error on PHIs in blocks duplicated into multiple places.
680 * graphite-scop-detection.c
681 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
682 in the region use it as loop and nest to analyze the DR in.
683 (try_generate_gimple_bb): Likewise.
684 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
685 (add_loop_constraints): For blocks in a loop not in the region
686 create a dimension with a single iteration.
687 * sese.h (gbb_loop_at_index): Remove assert.
689 2017-10-01 Kevin Buettner <kevinb@redhat.com>
691 * omp-expand.c (adjust_context_scope): New function.
692 (expand_parallel_call): Call adjust_context_scope.
694 2017-10-01 Jeff Law <law@redhat.com>
696 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
697 dom_opt_dom_walker class with direct access to private members.
698 Add comments. Call test_for_singularity.
699 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
700 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
701 m_dummy_cond anymore.
702 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
704 (pass_dominator:execute): Build the dummy_cond here and pass it
705 to the dom_opt_dom_walker ctor.
706 (test_for_singularity): New function.
708 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
709 Maya Rashish <coypu@sdf.org>
711 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
712 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
713 (alpha*-*-netbsd*) Use nbsd_tm_file.
714 (arm*-*-netbsdelf*) Likewise.
715 (i[34567]86-*-netbsdelf*) Likewise.
716 (x86_64-*-netbsd*) Likewise.
717 (mips*-*-netbsd*) Likewise.
718 (powerpc-*-netbsd*) Likewise.
719 (sh*-*-netbsd*) Likewise.
720 (sparc-*-netbsdelf*) Likewise.
721 (sparc64-*-netbsd*) Likewise.
722 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
724 (vax-*-netbsdelf*) Likewise.
725 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
726 (UINT_FAST8_TYPE) Likewise.
727 (INT_FAST16_TYPE) Check CHAR_FAST16.
728 (UINT_FAST16_TYPE) Likewise.
730 2017-09-30 Jakub Jelinek <jakub@redhat.com>
733 * config/i386/i386.md
734 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
735 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
736 *divmodsi4_zext_2): New define_insn_and_split.
737 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
738 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
739 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
740 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
741 New define_insn_and_split.
742 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
743 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
744 operands[1] having DImode when mode is SImode.
746 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
747 always SImode for DIV and MOD in REG_EQUAL notes.
749 2017-09-29 Yury Gribov <tetra2005@gmail.com>
752 * match.pd: Fix handling of NaNs in pattern.
754 2017-09-29 Jeff Law <law@redhat.com>
756 * sbitmap.c (bitmap_bit_in_range_p): New function.
757 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
758 * tree-ssa-dse.c (live_bytes_read): New function.
759 (dse_classify_store): Ignore reads of dead bytes.
761 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
762 typos and whitespace errors.
763 * config/i386/predicates.md (address_no_seg_operand): Likewise.
764 * config/s390/s390.c (s390_emit_prologue): Likewise.
766 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
769 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
770 with a symbol for LRA.
772 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
774 PR rtl-optimization/82338
775 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
777 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
779 * genmodes.c (calc_wider_mode): Suppress qsort macro.
780 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
781 (qsort_chk): Declare.
782 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
783 (qsort_chk): New function.
785 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
787 PR tree-optimization/82337
788 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
789 phi definition if the PHI result appears in an abnormal PHI.
790 (find_basis_for_base_expr): Don't record a basis if the LHS of the
791 basis appears in an abnormal PHI.
793 2017-09-29 Richard Biener <rguenther@suse.de>
795 * graphite-isl-ast-to-gimple.c
796 (translate_isl_ast_to_gimple::set_codegen_error): New function.
797 (binary_op_to_tree): Use it.
798 (get_rename_from_scev): Likewise.
799 (copy_loop_phi_nodes): Likewise.
800 (copy_bb_and_scalar_dependences): Likewise.
801 (translate_pending_phi_nodes): Likewise.
803 2017-09-29 Jakub Jelinek <jakub@redhat.com>
806 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
807 for movabsq $(i32 << shift), r64.
809 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
811 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
812 index when encoding %esp as %rsp to avoid 0x67 prefix.
814 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
816 * config/i386/i386.md (*movsf_internal, *movdf_internal):
817 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
819 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
821 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
822 Extensions with more than 16 double VFP registers.
823 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
824 to_clear_mask and all code related to it. Replace the remaining
825 entry by a sbitmap and adapt code accordingly.
827 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
829 * brig-builtins.def: Change pure attributes to const.
831 2017-09-28 Joseph Myers <joseph@codesourcery.com>
833 * config.gcc (default_gnu_indirect_function): Default to yes for
834 sparc*-*-linux* with glibc.
836 2017-09-28 Joseph Myers <joseph@codesourcery.com>
838 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
839 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
840 when creating .init_array and .fini_array sections with priority
843 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
846 * config/aarch64/aarch64.c
847 (aarch64_builtin_support_vector_misalignment): Always return false
848 when misalignment is unknown.
850 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
852 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
853 this function to return false if the definition used by the swap
854 instruction is artificial, or if the memory address from which the
855 constant value is loaded is not represented by a base address held
856 in a register or if the base address register is a frame or stack
857 pointer. Additionally, return false if the base address of the
858 loaded constant is a SYMBOL_REF but is not considered to be a
860 (replace_swapped_load_constant): New function.
861 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
862 loaded constant vector with a load of a swapped constant vector.
864 2017-09-27 Carl Love <cel@us.ibm.com>
866 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
867 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
868 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
871 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
873 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
874 first, always call autopref_rank_data otherwise.
876 2017-09-27 Richard Biener <rguenther@suse.de>
878 * graphite-scop-detection.c (find_scop_parameters): Move
879 loop bound handling ...
880 (gather_bbs::before_dom_children): ... here, avoiding the need
881 to build scop_info->loop_nest.
882 (record_loop_in_sese): Remove.
883 * sese.h (sese_info_t::loop_nest): Remove.
884 * sese.c (new_sese_info): Do not allocate loop_nest.
885 (free_sese_info): Do not free loop_nest.
887 2017-09-27 Jakub Jelinek <jakub@redhat.com>
890 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
891 lhs from calls if the lhs has addressable type.
893 2017-09-27 Richard Biener <rguenther@suse.de>
895 * graphite.h (scop::max_alias_set): New member.
896 * graphite-scop-detection.c: Remove references to non-existing
898 (build_alias_sets): Record the maximum alias set used for drs.
899 (build_scops): Support zero as unlimited for
900 --param graphite-max-arrays-per-scop.
901 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
903 (build_poly_sr_1): ... here. Compute alias set based on the
904 maximum alias set used for drs rather than
905 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
907 2017-09-27 Richard Biener <rguenther@suse.de>
909 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
910 --param loop-block-tile-size=0 to disable tiling.
912 2017-09-27 Richard Biener <rguenther@suse.de>
914 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
915 (graphite-max-nb-scop-params): Document special value zero.
916 * domwalk.h (dom_walker::STOP): New symbolical constant.
917 (dom_walker::dom_walker): Add optional parameter for bb to
919 (dom_walker::~dom_walker): Declare.
920 (dom_walker::before_dom_children): Document STOP return value.
921 (dom_walker::m_user_bb_to_rpo): New member.
922 (dom_walker::m_bb_to_rpo): Likewise.
923 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
924 mapping here if not provided by the user.
925 (dom_walker::~dom_walker): Free bb to RPO mapping if not
926 provided by the user.
927 (dom_walker::STOP): Define.
928 (dom_walker::walk): Do not compute bb to RPO mapping here.
929 Support STOP return value from before_dom_children to stop
931 * graphite-optimize-isl.c (optimize_isl): If the schedule
932 is the same still generate code if -fgraphite-identity
933 or -floop-parallelize-all are given.
934 * graphite-scop-detection.c: Include cfganal.h.
935 (gather_bbs::gather_bbs): Get and pass through bb to RPO
937 (gather_bbs::before_dom_children): Return STOP for BBs
939 (build_scops): Compute bb to RPO mapping and pass it to
940 the domwalk. Treat --param graphite-max-nb-scop-params=0
941 as not limiting the number of params.
942 * graphite.c (graphite_initialize): Remove limit on the
943 number of basic-blocks in a function.
944 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
945 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
948 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
950 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
951 Adjust code to eliminate needing to do the shift right 32-bits
952 operation after XSCVDPSPN.
954 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
956 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
957 ((X / Y) != 0 -> X >= Y): Likewise.
959 2017-09-26 Carl Love <cel@us.ibm.com>
961 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
962 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
963 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
964 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
965 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
966 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
967 definitions and overloading.
968 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
969 statement for P9V_BUILTIN_XST_LEN_R.
970 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
971 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
972 define_expand and define_insn for the instructions and builtins.
973 * doc/extend.texi: Update the built-in documentation file for the new
975 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
976 define_insn for the instructions
978 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
981 * gcc/config/netbsd-protos.h: New file.
982 * gcc/config/netbsd.c: New file.
983 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
984 * gcc/config/t-netbsd: New file.
985 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
986 (tmake_file) Add t-netbsd.
987 (extra_objs) Add netbsd.o.
989 2017-09-26 Janus Weil <janus@gcc.gnu.org>
993 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
995 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
997 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
998 sign extension from a vector register to a GPR by doing a 32-bit
999 direct move and then an EXTSW.
1000 (extendsi<mode>2 splitter): Likewise.
1001 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1002 right or vector extract after doing XSCVDPSPN. Use
1003 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1005 (movdi_from_sf_zero_ext): Likewise.
1006 (reload_gpr_from_vsxsf): Likewise.
1007 (p8_mfvsrd_4_disf): Delete, no longer used.
1008 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1009 then needing to move the SFmode to a GPR to use the XSCVDPSP
1010 instruction instead of FRSP and XSCVDPSPN.
1011 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1012 it is adjacent to the other XSCVSPDP insns.
1013 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1014 SFmode to be in traditional Altivec registers.
1015 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1016 (vsx_xscvspdpn): Likewise.
1017 (vsx_xscvspdpn_scalar): Likewise.
1019 2017-09-26 Martin Jambor <mjambor@suse.cz>
1021 * tree-sra.c (compare_access_positions): Put integral types first,
1022 stabilize sorting of integral types, remove conditions putting
1023 non-full-precision integers last.
1024 (sort_and_splice_var_accesses): Disable scalarization if a
1025 non-integert would be represented by a non-full-precision integer.
1027 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1029 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1030 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1031 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1032 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1033 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1034 conditionals inside the function instead of around it. Call
1035 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1036 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1038 2017-09-26 Richard Biener <rguenther@suse.de>
1040 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1042 (scop_detection::build_scop_breadth): ... this. Removed.
1043 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1044 (scop_detection::harmful_stmt_in_bb): Likewise.
1045 (scop_detection::graphite_can_represent_stmt): Likewise.
1046 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1047 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1048 (scop_detection::can_represent_loop_1): ... this. Removed.
1049 (scop_detection::harmful_loop_in_region): Simplify after inlining
1050 the above and remove more quadraticness.
1051 (build_scops): Adjust.
1052 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1055 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1058 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1059 REGNO (base) == SP_REG if base is a REG.
1062 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1063 if it is different SSA_NAME.
1064 (optimize_range_tests_cmp_bitwise): New function.
1065 (optimize_range_tests): Call it.
1067 2017-09-26 Richard Biener <rguenther@suse.de>
1069 PR tree-optimization/82321
1070 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1071 for the def being inside the loop.
1073 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1075 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1077 * config/s390/s390-builtins.def: Fix constraint on op4.
1079 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1081 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1082 independent expanders.
1083 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1084 ("vec_ordered", "vec_unordered"): New expanders.
1086 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1088 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1091 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1093 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1094 vec_unpacks_lo_v16qi.
1095 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1097 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1099 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1100 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1101 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1103 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1105 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1107 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1109 ("*vec_slb<mode>"): New insn pattern.
1110 ("vec_shr_<mode>"): New expander.
1111 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1112 and force the shift count operand to V16QImode.
1113 ("vec_srb<mode>"): Set shift count mode to V16QI.
1115 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1117 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1118 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1119 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1121 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1124 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1126 2017-09-26 Richard Biener <rguenther@suse.de>
1128 PR tree-optimization/82320
1129 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1132 2017-09-25 Jeff Law <law@redhat.com>
1134 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1135 prototype for new argument.
1136 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1137 mostly extracted from rs6000_emit_allocate_stack.
1138 (rs6000_emit_probe_stack_range_stack_clash): New function.
1139 (rs6000_emit_allocate_stack): Call
1140 rs6000_emit_probe_stack_range_stack_clash as needed.
1141 (rs6000_emit_probe_stack_range): Add additional argument
1142 to call to gen_probe_stack_range{si,di}.
1143 (output_probe_stack_range): New.
1144 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1145 (output_probe_stack_range_stack_clash): New.
1146 (rs6000_emit_prologue): Emit notes into dump file as requested.
1147 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1148 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1149 Add additional operand and pass it to output_probe_stack_range.
1151 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1153 PR tree-optimization/82163
1154 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1155 (checking_verify_loop_closed_ssa): New parameter.
1156 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1157 (check_loop_closed_ssa_stmt): Delete.
1158 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1159 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1160 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1163 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1165 * brig-builtins.def: Treat HSAIL barrier builtins as
1166 setjmp/longjump style functions.
1168 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1170 * target.def (constant_alignment): New hook.
1171 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1172 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1173 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1174 * doc/tm.texi: Regenerate.
1175 * targhooks.h (default_constant_alignment): Declare.
1176 (constant_alignment_word_strings): Likewise.
1177 * targhooks.c (default_constant_alignment): New function.
1178 (constant_alignment_word_strings): Likewise.
1179 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1180 instead of CONSTANT_ALIGNMENT.
1181 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1182 (force_const_mem): Likewise.
1183 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1184 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1185 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1186 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1187 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1189 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1190 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1191 constant_alignment_word_strings.
1192 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1193 (CONSTANT_ALIGNMENT): Likewise.
1194 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1195 (arm_constant_alignment): New function.
1196 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1197 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1198 constant_alignment_word_strings.
1199 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1200 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1201 constant_alignment_word_strings.
1202 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1203 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1204 (cris_constant_alignment): New function.
1205 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1206 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1207 (epiphany_constant_alignment): New function.
1208 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1209 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1210 constant_alignment_word_strings.
1211 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1212 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1213 constant_alignment_word_strings.
1214 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1215 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1216 constant_alignment_word_strings.
1217 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1218 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1219 * config/i386/i386.c (ix86_constant_alignment): Make static.
1220 Use the same interface as the target hook.
1221 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1222 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1223 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1224 constant_alignment_word_strings.
1225 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1226 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1227 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1228 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1229 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1230 constant_alignment_word_strings.
1231 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1232 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1233 constant_alignment_word_strings.
1234 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1235 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1236 constant_alignment_word_strings.
1237 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1238 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1240 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1241 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1242 * config/mips/mips.c (mips_constant_alignment): New function.
1243 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1244 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1245 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1246 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1247 (mmix_constant_alignment): Make static. Use the same interface
1249 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1250 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1251 constant_alignment_word_strings.
1252 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1253 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1254 constant_alignment_word_strings.
1255 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1256 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1257 constant_alignment_word_strings.
1258 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1259 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1260 (rs6000_constant_alignment): New function.
1261 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1262 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1263 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1264 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1265 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1266 (rs6000_constant_alignment): New function.
1267 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1268 * config/s390/s390.c (s390_constant_alignment): New function.
1269 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1270 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1271 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1272 constant_alignment_word_strings.
1273 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1274 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1275 (sparc_constant_alignment): New function.
1276 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1277 * config/spu/spu.c (spu_constant_alignment): New function.
1278 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1279 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1280 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1281 constant_alignment_word_strings.
1282 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1283 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1284 constant_alignment_word_strings.
1285 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1286 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1287 constant_alignment_word_strings.
1288 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1289 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1290 (visium_constant_alignment): New function.
1291 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1292 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1293 (xtensa_constant_alignment): New function.
1294 * system.h (CONSTANT_ALIGNMENT): Poison.
1296 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1298 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1299 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1300 (rs6000_builtin_valid_without_lhs): New helper function.
1301 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1302 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1304 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1306 * target.h (vec_perm_indices): Use unsigned short rather than
1308 (auto_vec_perm_indices): Likewise.
1309 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1310 Use unsigned int rather than unsigned char.
1311 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1313 2017-09-25 Richard Biener <rguenther@suse.de>
1315 * cfgloop.h (sort_sibling_loops): Declare.
1316 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1317 (sort_sibling_loops): New function sorting the sibling loop list
1319 * graphite.c (graphite_transform_loops): Sort sibling loops.
1321 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1323 * target.def (vec_perm_const_ok): Change sel parameter to
1325 * optabs-query.c (can_vec_perm_p): Update accordingly.
1326 * doc/tm.texi: Regenerate.
1327 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1328 auto_vec_perm_indices and remove separate nelt field.
1329 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1330 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1331 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1332 (aarch64_expand_vec_perm_const): Update accordingly.
1333 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1334 to vec_perm_indices.
1335 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1336 auto_vec_perm_indices and remove separate nelt field.
1337 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1338 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1339 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1341 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1342 to vec_perm_indices.
1343 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1344 sel to vec_perm_indices.
1345 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1346 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1347 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1349 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1352 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1355 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1356 on the FUNCTION_DECL function context if it has a DIE that is a
1359 2017-09-25 Richard Biener <rguenther@suse.de>
1361 PR tree-optimization/82285
1362 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1365 2017-09-25 Tom de Vries <tom@codesourcery.com>
1369 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1372 2017-09-25 Richard Biener <rguenther@suse.de>
1374 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1375 ISL errors other than isl_error_quota happen. Dump if the
1376 schedule is the same.
1377 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1378 errors instead of aborting inside ISL.
1380 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1383 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1384 of libgcc_eh for m64.
1385 * config/i386/darwin64.h: Likewise.
1387 2017-09-25 Richard Biener <rguenther@suse.de>
1390 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1391 attribute for incomplete types nor twice for complete ones.
1393 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1396 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1397 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1400 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1403 * config/i386/i386.opt (mprefer-avx256): Use
1404 ix86_target_flags variable.
1405 * config/i386/i386.c (ix86_target_string): Move
1406 -mprefer-avx256 to flag2_opts.
1408 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1411 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1412 and x != -1 | y != -1 into (x & y) != -1.
1414 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1416 * config.gcc: Add new case statement to set
1417 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1418 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1419 s390x-*-linux* case statements. Added aarch64 to the list of
1420 supported architectures.
1422 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1424 PR tree-optimization/82289
1425 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1426 STMT_VINFO_RELEVANT_P.
1428 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1429 Alan Hayward <alan.hayward@arm.com>
1430 David Sherwood <david.sherwood@arm.com>
1432 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1433 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1434 (extract_range_from_binary_expr_1): Don't call
1435 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1437 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1438 Alan Hayward <alan.hayward@arm.com>
1439 David Sherwood <david.sherwood@arm.com>
1441 * target.def (preferred_vector_alignment): New hook.
1442 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1444 * doc/tm.texi: Regenerate.
1445 * targhooks.h (default_preferred_vector_alignment): Declare.
1446 * targhooks.c (default_preferred_vector_alignment): New function.
1447 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1449 (DR_TARGET_ALIGNMENT): New macro.
1450 (aligned_access_p): Update commentary.
1451 (vect_known_alignment_in_bytes): New function.
1452 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1454 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1455 Calculate the misalignment based on the target alignment rather than
1457 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1458 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1459 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1460 the target alignment, rather than masking the element misalignment
1461 with the number of elements in a vector. Also use the target
1462 alignment when calculating the maximum number of peels.
1463 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1464 instead of TYPE_ALIGN_UNIT.
1465 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1466 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1467 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1468 (vect_create_data_ref_ptr): Likewise.
1469 (vect_setup_realignment): Realign by ANDing with
1470 -DR_TARGET_MISALIGNMENT.
1471 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1472 the number of peels based on DR_TARGET_ALIGNMENT.
1473 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1474 with the guaranteed alignment boundary when deciding whether
1476 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1477 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1478 (ensure_base_align): Remove stmt_info parameter. Get the
1479 target base alignment from DR_TARGET_ALIGNMENT.
1480 (vectorizable_store): Update call accordingly. Interpret
1481 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1483 (vectorizable_load): Likewise.
1485 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1486 Alan Hayward <alan.hayward@arm.com>
1487 David Sherwood <david.sherwood@arm.com>
1489 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1490 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1491 (vect_enhance_data_refs_alignment): Likewise.
1493 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1495 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1496 error. Only quit immediately if parsing is complete.
1497 (BEGIN): Initialize fatal_err and parse_done.
1498 (begin fpu, end fpu): Check number of arguments.
1499 (begin arch, end arch): Likewise.
1500 (begin cpu, end cpu): Likewise.
1501 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1502 (optalias): Likewise.
1504 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1506 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1507 * config/arm/arm-isa.h: Delete. Move definitions to ...
1508 * arm-cpus.in: ... here. Use new feature and fgroup values.
1509 * config/arm/arm.c (arm_option_override): Use lower case for feature
1511 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1512 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1513 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1515 (print_isa_bits_for): New function.
1516 (gen_isa): New function.
1517 (gen_comm_data): Use print_isa_bits_for.
1518 (define feature): New keyword.
1519 (define fgroup): New keyword.
1520 * config/arm/t-arm (TM_H): Remove.
1521 (GTM_H): Add arm-isa.h.
1522 (arm-isa.h): Add rule to generate file.
1523 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1524 case for feature bit names.
1526 2017-09-22 Richard Biener <rguenther@suse.de>
1528 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1530 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1531 print of no dependency loops ...
1532 * graphite.c (graphite_transform_loops): ... here.
1533 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1535 (same_close_phi_node, remove_duplicate_close_phi,
1536 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1537 (canonicalize_loop_closed_ssa): ... here and simplify.
1538 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1539 (optimize_isl): Use dump_printf_loc to tell when we stopped
1540 optimizing because of an ISL timeout.
1542 2017-09-22 Richard Biener <rguenther@suse.de>
1544 PR tree-optimization/82291
1545 * tree-if-conv.c (predicate_mem_writes): Make sure to
1546 remove writes in blocks predicated with false.
1548 2017-09-22 Richard Biener <rguenther@suse.de>
1550 * sese.c: Include cfganal.h.
1551 (if_region_set_false_region): Remove.
1552 (create_if_region_on_edge): Likewise.
1553 (move_sese_in_condition): Re-implement without destroying
1556 2017-09-22 Richard Biener <rguenther@suse.de>
1558 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1559 Verify both BBs contain loop PHI nodes before dispatching to
1561 (graphite_regenerate_ast_isl): Do not recompute dominators,
1562 do not verify three times. Restructure for clarity.
1563 * graphite-scop-detection.c (same_close_phi_node,
1564 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1565 defined_in_loop_p, canonicalize_loop_closed_ssa,
1566 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1567 checking and SSA rewrite, move to ...
1568 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1569 (graphite_initialize): Do not pass in ctx, do not reset the
1570 SCEV cache, compute only dominators.
1571 (graphite_transform_loops): Allocate ISL ctx after
1572 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1573 Maintain post-dominators only around build_scops.
1574 * sese.c (if_region_set_false_region): Make static. Free
1575 and recompute dominators.
1576 (move_sese_in_condition): Assert we don't get called with
1577 post-dominators computed.
1578 * sese.h (if_region_set_false_region): Remove.
1580 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1582 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1583 mode attribute for TARGET_AVX512VL.
1585 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1587 * config/i386/i386.opt (mprefer-avx256): New option.
1588 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1590 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1591 for TARGET_PREFER_AVX256.
1592 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1594 2017-09-21 Jeff Law <law@redhat.com>
1596 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1597 Fix dump output if the only stack space is for pushed registers.
1599 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1601 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1604 2017-09-21 Martin Sebor <msebor@redhat.com>
1607 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1608 code (in C++) or code that triggers warnings.
1610 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1612 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1614 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1616 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1617 * sched-rgn.c: Ditto.
1618 * sel-sched-ir.c: Ditto.
1620 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1622 * toplev.h (set_random_seed): Adjust return type.
1623 * toplev.c (init_local_tick): Move eager initialization of random_seed
1624 to get_random_seed. Adjust comment.
1625 (init_random_seed): Inline to get_random_seed, delete.
1626 (get_random_seed): Initialize random_seed lazily.
1627 (set_random_seed): Do not return previous value.
1628 (print_switch_value): Do not call get_random_seed.
1630 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1632 * cgraph.c (delete_function_version): New, broken out from...
1633 (cgraph_node::delete_function_version): ...here. Rename to
1634 cgraph_node::delete_function_version_by_decl. Update all uses.
1635 (cgraph_node::remove): Call delete_function_version.
1637 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1640 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1641 VAR_DECLs to which addressable non-volatile parameters are mapped
1642 and for id->retvar after the return value assignment. Clear
1643 id->retval and id->retbnd after inlining.
1645 2017-09-21 Richard Biener <rguenther@suse.de>
1647 PR tree-optimization/82276
1648 PR tree-optimization/82244
1649 * tree-vrp.c (build_assert_expr_for): Set
1650 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1652 (remove_range_assertions): Revert earlier change.
1654 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1657 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1659 2017-09-21 Richard Biener <rguenther@suse.de>
1661 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1662 Restore valid IL after code generation errors.
1663 * graphite.c (graphite_transform_loops): Diagnose code
1664 generation issues as MSG_MISSED_OPTIMIZATION and continue
1665 with processing SCOPs.
1667 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1668 Alan Hayward <alan.hayward@arm.com>
1669 David Sherwood <david.sherwood@arm.com>
1671 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1672 rather than choosing between plus_constant and gen_rtx_<CODE>.
1673 * expr.c (emit_push_insn): Likewise.
1674 (expand_expr_real_2): Likewise.
1676 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1677 Alan Hayward <alan.hayward@arm.com>
1678 David Sherwood <david.sherwood@arm.com>
1680 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1682 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1683 Alan Hayward <alan.hayward@arm.com>
1684 David Sherwood <david.sherwood@arm.com>
1686 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1687 calling tree_to_uhwi.
1689 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1690 Alan Hayward <alan.hayward@arm.com>
1691 David Sherwood <david.sherwood@arm.com>
1693 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1694 INTEGER_CST rather than a negative test for ADDR_EXPR.
1696 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1697 Alan Hayward <alan.hayward@arm.com>
1698 David Sherwood <david.sherwood@arm.com>
1700 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1701 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1703 2017-09-21 Richard Biener <rguenther@suse.de>
1705 PR tree-optimization/71351
1706 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1707 graphite_create_new_loop_guard): Remove, fold remaining parts
1709 (translate_isl_ast_node_for): ... here and simplify.
1711 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1714 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1715 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1716 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1717 alternative always use QI mode, for -Os imov (=R,R) alternative
1718 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1721 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1722 Jeff Law <law@redhat.com>
1724 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1725 (allocate_stack_space): New function, partially extracted from
1727 (s390_emit_prologue): Track offset to most recent stack probe.
1728 Code to allocate space moved into allocate_stack_space.
1729 Dump actions when no stack is allocated.
1730 (s390_prologue_plus_offset): New function.
1731 (s390_emit_stack_probe): Likewise.
1733 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1735 * common.opt (Wa, Wl, Wp, g, gz=): Add
1737 (gno-column-info): Remove.
1738 (gcolumn-info): Drop RejectNegative.
1740 (gno-record-gcc-switches): Remove.
1741 (grecord-gcc-switches): Drop RejectNegative.
1742 (gno-split-dwarf): Remove.
1743 (gsplit-dwarf): Drop RejectNegative.
1744 (gno-strict-dwarf): Remove.
1745 (gstrict-dwarf): Drop RejectNegative.
1746 * config/darwin.opt (gfull, gused): Add RejectNegative.
1747 * dwarf2out.c (gen_producer_string): Drop
1748 gno-record-gcc-switches handler.
1749 * optc-gen.awk: Add g to prefixes with negative forms.
1750 * opts-common.c (remapping_prefix_p): New.
1751 (find_opt): Check it.
1752 (generate_canonical_option): Test g prefix.
1753 (option_map): Add -gno- mapping.
1754 (add_misspelling_candidates): Check remapping_prefix_p.
1756 2017-09-20 Jeff Law <law@redhat.com>
1758 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1759 thinko in stack clash protection support.
1761 * explow.c (compute_stack_clash_protection_loop_data): Use
1762 CONST_INT_P instead of explicit test. Verify object is a
1763 CONST_INT_P before looking at INTVAL.
1764 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1765 instead of explicit test.
1767 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1770 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1771 address instead of to r1 and r11.
1773 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1775 * config.gcc: Support "knm".
1776 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1777 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1779 * config/i386/i386.c (m_KNM): Define.
1780 (processor_target_table): Add "knm".
1782 (ix86_option_override_internal): Add "knm".
1783 (ix86_issue_rate): Add PROCESSOR_KNM.
1784 (ix86_adjust_cost): Ditto.
1785 (ia32_multipass_dfa_lookahead): Ditto.
1786 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1787 (fold_builtin_cpu): Add M_INTEL_KNM.
1788 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1789 (processor_type): Add PROCESSOR_KNM.
1790 * config/i386/x86-tune.def: Add m_KNM.
1791 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1793 2017-09-20 Richard Biener <rguenther@suse.de>
1795 PR tree-optimization/80213
1796 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1797 are allowed in empty BBs as well.
1798 (canonicalize_loop_closed_ssa): Also look for other complex
1800 (scop_detection::get_sese): Include the loop-closed PHI block
1802 (scop_detection::merge_sese): Remove code adding extra blocks.
1803 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1804 (build_scops): Assert the final returned scop is invalid.
1806 2017-09-20 Richard Biener <rguenther@suse.de>
1808 PR tree-optimization/82264
1809 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1811 (vn_phi_lookup): Likewise.
1812 (vn_phi_insert): Likewise.
1814 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1816 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1817 that fits into uhwi or shwi, add DW_AT_const_value regardless
1818 of early_dwarf without going through RTL, using add_AT_unsigned
1821 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1822 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1823 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1824 (DEBUG_MACRO_SECTION): Likewise.
1825 (DEBUG_LTO_MACRO_SECTION): Likewise.
1826 (DEBUG_STR_DWO_SECTION): Likewise.
1827 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1828 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1829 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1830 (DEBUG_LTO_LINE_STR_SECTION): Define.
1831 (init_sections_and_labels): Initialize debug_line_str_section
1832 variable. Initialize debug_loc_section for -gdwarf-5 to
1833 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1835 2017-09-20 Richard Biener <rguenther@suse.de>
1837 * graphite-sese-to-poly.c (extract_affine): Properly handle
1838 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1840 2017-09-20 Richard Biener <rguenther@suse.de>
1842 PR tree-optimization/81373
1843 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1844 Force SESE live-out defs to be handled even if they are
1847 2017-09-19 Jeff Law <law@redhat.com>
1849 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1850 nothing for stack adjustments with REG_STACK_CHECK.
1851 * sched-deps.c (parse_add_or_inc): Reject insns with
1852 REG_STACK_CHECK from dependency breaking.
1853 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1854 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1855 * reg-notes.def (STACK_CHECK): New note.
1857 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1858 (ix86_expand_prologue): Dump stack clash info as needed.
1859 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1861 * function.c (dump_stack_clash_frame_info): New function.
1862 * function.h (dump_stack_clash_frame_info): Prototype.
1863 (enum stack_clash_probes): New enum.
1865 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1866 flag_stack_clash_protection.
1867 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1868 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1869 (arm_frame_pointer_required): Likewise.
1870 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1871 (ia64_expand_prologue): Likewise.
1872 * config/mips/mips.c (mips_expand_prologue): Likewise.
1873 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1874 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1875 (sparc_flat_expand_prologue): Likewise.
1876 * config/spu/spu.c (spu_expand_prologue): Likewise.
1878 * explow.c: Include "params.h".
1879 (anti_adjust_stack_and_probe_stack_clash): New function.
1880 (get_stack_check_protect): Likewise.
1881 (compute_stack_clash_protection_loop_data): Likewise.
1882 (emit_stack_clash_protection_loop_start): Likewise.
1883 (emit_stack_clash_protection_loop_end): Likewise.
1884 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1885 Use anti_adjust_stack_and_probe_stack_clash.
1886 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1887 (emit_stack_clash_protection_loop_start): Likewise.
1888 (emit_stack_clash_protection_loop_end): Likewise.
1889 * rtl.h (get_stack_check_protect): Prototype.
1890 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1891 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1892 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1894 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1896 * doc/tm.texi: Rebuilt.
1897 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1898 get_stack_check_protect.
1899 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1900 * config/arm/arm.c (arm_expand_prologue): Likewise.
1901 (arm_frame_pointer_required): Likewise.
1902 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1903 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1904 * config/mips/mips.c (mips_expand_prologue): Likewise.
1905 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1906 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1907 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1908 (sparc_flat_expand_prologue): Likewise.
1910 * common.opt (-fstack-clash-protection): New option.
1911 * flag-types.h (enum stack_check_type): Note difference between
1912 -fstack-check= and -fstack-clash-protection.
1913 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1914 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1915 * toplev.c (process_options): Issue warnings/errors for cases
1916 not handled with -fstack-clash-protection.
1917 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1918 (-fstack-check): Note additional problem with -fstack-check=generic.
1919 Note that -fstack-check is primarily for Ada and refer users
1920 to -fstack-clash-protection for stack-clash-protection.
1921 Document new params for stack clash protection.
1923 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1925 * config/i386/i386.c (ix86_split_long_move): Do not handle
1926 address used for LEA in a special way.
1928 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1930 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1932 2017-09-19 Martin Sebor <msebor@redhat.com>
1935 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1936 of incompatible types.
1938 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1940 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1941 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1942 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1943 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1945 2017-09-19 Richard Biener <rguenther@suse.de>
1947 PR tree-optimization/82244
1948 * tree-vrp.c (remove_range_assertions): Do not propagate
1949 a constant to abnormals but replace the assert with a copy.
1951 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1953 PR rtl-optimization/57878
1954 PR rtl-optimization/68988
1955 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1956 avoidance test involving non_reload_pseudos. Move frequency test
1957 below the general fragmentation avoidance test.
1959 2017-09-19 Richard Biener <rguenther@suse.de>
1961 PR tree-optimization/69728
1962 * graphite-sese-to-poly.c (schedule_error): New global.
1963 (add_loop_schedule): Handle empty domain by failing the
1965 (build_original_schedule): Handle schedule_error.
1967 2017-09-19 Richard Biener <rguenther@suse.de>
1969 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1970 Do not iterate to sibling loops but only to siblings of inner
1973 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1976 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1977 that the registers are different.
1979 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1981 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1982 to processor_model and "amdfam17h" to arch_names_table.
1983 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1985 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1988 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1990 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1991 Alan Hayward <alan.hayward@arm.com>
1992 David Sherwood <david.sherwood@arm.com>
1994 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1996 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1998 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1999 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2000 vect_schedule_slp_instance.
2001 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2002 Update call to vect_slp_analyze_node_operations. Simplify return
2004 (vect_slp_analyze_bb_1): Update call accordingly.
2005 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2006 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2007 (vect_schedule_slp): Update call accordingly.
2009 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2010 Alan Hayward <alan.hayward@arm.com>
2011 David Sherwood <david.sherwood@arm.com>
2013 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2014 with types that aren't in fact scalar.
2016 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2018 * tree-vect-slp.c (vect_record_max_nunits): New function,
2020 (vect_build_slp_tree_1): ...here.
2021 (vect_build_slp_tree_2): Call it for phis too.
2023 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2025 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2026 to vect_get_vec_def_for_operand when getting the mask operand.
2028 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2029 Alan Hayward <alan.hayward@arm.com>
2030 David Sherwood <david.sherwood@arm.com>
2032 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2035 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2036 Alan Hayward <alan.hayward@arm.com>
2037 David Sherwood <david.sherwood@arm.com>
2039 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2040 calculation for vector booleans.
2042 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2043 Alan Hayward <alan.hayward@arm.com>
2044 David Sherwood <david.sherwood@arm.com>
2046 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2048 (vect_transform_stmt): ...here.
2049 (vect_analyze_stmt): Use it instead of calling
2050 vectorizable_live_operation directly.
2052 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2054 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2055 non-SIMT targets in acc vector loops.
2057 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2059 * configure.ac: Add arc and check if assembler supports gdwarf2.
2060 * configure: Regenerate.
2062 2017-09-18 Richard Biener <rguenther@suse.de>
2064 PR tree-optimization/82220
2065 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2066 epilogue niters from the min_profitable_iters compute.
2068 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2071 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2072 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2073 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2075 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2078 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2079 switching to a new text section.
2081 2017-09-18 Richard Biener <rguenther@suse.de>
2083 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2085 (build_alias_set): Reject aliases with no access function.
2087 2017-09-18 Richard Biener <rguenther@suse.de>
2089 PR tree-optimization/79622
2090 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2092 (build_cross_bb_scalars_use): Likewise.
2094 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2096 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2098 2017-09-18 Alan Modra <amodra@gmail.com>
2101 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2102 stack_pointer_rtx for count 0. Update comments. Break up
2103 large rtl expression.
2105 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2107 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2108 Increase to 20 bytes.
2109 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2110 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2111 or avx version of the stub.
2113 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2116 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2117 compute the minimum stack alignment. Also update preferred stack
2118 boundary for leaf functions.
2120 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2122 PR tree-optimization/82228
2123 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2126 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2128 * common/config/nds32/nds32-common.c
2129 (nds32_option_optimization_table): Refine formatting.
2130 (nds32_option_optimization_table): Use -fsched-pressure and
2131 -fomit-frame-pointer for specific optimization level.
2133 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2135 * config/nds32/nds32.c: Refine formatting and comments.
2136 * config/nds32/nds32.h: Likewise.
2137 * config/nds32/nds32.md: Likewise.
2138 * config/nds32/nds32-cost.c: Likewise.
2139 * config/nds32/nds32-isr.c: Likewise.
2140 * config/nds32/nds32-md-auxiliary.c: Likewise.
2141 * config/nds32/nds32-multiple.md: Likewise.
2142 * config/nds32/nds32-predicates.c: Likewise.
2144 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2145 Jakub Jelinek <jakub@redhat.com>
2147 Add support for -std=c++2a.
2148 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2150 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2152 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2155 * doc/extend.texi (Common Function Attributes): Add
2156 references to ARM, AArch64, and S/390 specific attributes.
2157 (Function Specific Option Pragmas): Add AArch64 and S/390
2158 to list of back ends that support the target pragma.
2160 2017-09-15 Nathan Sidwell <nathan@acm.org>
2162 * doc/standards.texi: Fix C++17 description. Update URLs for
2165 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2167 * common.opt (Wcast-align=strict): New warning option.
2168 * doc/invoke.texi: Document -Wcast-align=strict.
2170 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2172 * cgraph.h (cgraph_thunk_info): Add comments.
2173 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2174 assert for VIRTUAL_* arguments stricter.
2176 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2178 PR tree-optimization/71026
2179 * match.pd: Move RDIV patterns from fold-const.c
2180 * fold-const.c (distribute_real_division): Removed.
2181 (fold_binary_loc): Remove calls to distribute_real_divison.
2183 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2185 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2186 c++1z and gnu++1z as deprecated. Change other references to
2187 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2188 Change -Wc++1z-compat to -Wc++17-compat.
2189 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2190 * dwarf2out.c (highest_c_language): Handle C++17.
2191 (gen_compile_unit_die): Likewise.
2193 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2195 PR rtl-optimization/82192
2196 * combine.c (make_extraction): Don't look through non-paradoxical
2197 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2200 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2201 Alan Hayward <alan.hayward@arm.com>
2202 David Sherwood <david.sherwood@arm.com>
2204 * target.def (function_arg_offset): New hook.
2205 * targhooks.h (default_function_arg_offset): Declare.
2206 * targhooks.c (default_function_arg_offset): New function.
2207 * function.c (locate_and_pad_parm): Use
2208 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2209 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2210 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2211 * doc/tm.texi: Regenerate.
2212 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2213 * config/spu/spu.c (spu_function_arg_offset): New function.
2214 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2215 * system.h (FUNCTION_ARG_OFFSET): Poison.
2217 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2218 Alan Hayard <alan.hayward@arm.com>
2219 David Sherwood <david.sherwood@arm.com>
2221 * target.def (truly_noop_truncation): New hook.
2222 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2223 than TRULY_NOOP_TRUNCATION.
2224 * hooks.h (hook_bool_uint_uint_true): Declare.
2225 * hooks.c (hook_bool_uint_uint_true): New function.
2226 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2227 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2228 * doc/tm.texi: Regenerate.
2229 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2230 rather than TRULY_NOOP_TRUNCATION in comments.
2231 (simplify_comparison): Likewise.
2232 (record_truncated_value): Likewise.
2233 * expmed.c (extract_bit_field_1): Likewise.
2234 (extract_split_bit_field): Likewise.
2235 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2236 instead of TRULY_NOOP_TRUNCATION.
2237 * function.c (assign_parm_setup_block): Likewise.
2238 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2239 * rtlhooks.c: Include target.h.
2240 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2241 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2242 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2243 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2244 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2245 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2246 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2247 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2248 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2249 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2250 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2251 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2252 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2253 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2254 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2255 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2256 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2257 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2258 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2259 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2260 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2261 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2262 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2263 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2264 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2265 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2266 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2267 rather than TRULY_NOOP_TRUNCATION in comments.
2268 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2269 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2270 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2271 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2272 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2273 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2274 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2275 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2276 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2277 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2278 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2279 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2280 rather than TRULY_NOOP_TRUNCATION in comments.
2281 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2282 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2283 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2284 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2285 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2286 TRULY_NOOP_TRUNCATION condition.
2287 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2288 (TRULY_NOOP_TRUNCATION): Delete.
2289 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2290 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2291 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2292 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2293 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2294 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2295 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2296 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2297 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2298 rather than TRULY_NOOP_TRUNCATION in comments.
2299 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2300 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2301 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2302 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2303 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2304 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2306 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2309 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2310 (*cmp_ior): Likewise.
2311 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2312 (*ior_scc_scc_cmp): Likewise.
2313 (*and_scc_scc): Likewise.
2314 (*and_scc_scc_cmp): Likewise.
2316 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2317 Alan Hayard <alan.hayward@arm.com>
2318 David Sherwood <david.sherwood@arm.com>
2320 * target.def (can_change_mode_class): New hook.
2321 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2322 (hard_regno_nregs): Likewise.
2323 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2324 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2325 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2326 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2327 (LOAD_EXTEND_OP): Update accordingly.
2328 * doc/tm.texi: Regenerate.
2329 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2330 CANNOT_CHANGE_MODE_CLASS.
2331 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2332 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2333 * combine.c (simplify_set): Update accordingly.
2334 * emit-rtl.c (validate_subreg): Likewise.
2335 * recog.c (general_operand): Likewise.
2336 * regcprop.c (mode_change_ok): Likewise.
2337 * reload1.c (choose_reload_regs): Likewise.
2338 (inherit_piecemeal_p): Likewise.
2339 * rtlanal.c (simplify_subreg_regno): Likewise.
2340 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2341 instead of CANNOT_CHANGE_MODE_CLASS.
2342 (reload_cse_simplify_operands): Likewise.
2343 * reload.c (push_reload): Use targetm.can_change_mode_class
2344 instead of CANNOT_CHANGE_MODE_CLASS.
2345 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2346 REG_CANNOT_CHANGE_MODE_P.
2347 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2348 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2349 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2350 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2351 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2352 (arm_can_change_mode_class): New function.
2353 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2354 than CANNOT_CHANGE_MODE_CLASS in comments.
2355 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2356 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2357 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2358 (ix86_can_change_mode_class): ...this new function, inverting the
2359 sense of the return value.
2360 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2361 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2362 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2363 (ia64_can_change_mode_class): New function.
2364 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2365 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2366 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2367 (m32c_can_change_mode_class): ...this new function, inverting the
2368 sense of the return value.
2369 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2370 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2371 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2372 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2373 (mips_can_change_mode_class): ...this new function, inverting the
2374 sense of the return value.
2375 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2376 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2377 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2378 (msp430_can_change_mode_class): New function.
2379 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2380 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2381 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2382 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2383 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2384 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2385 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2386 (pa_cannot_change_mode_class): Replace with...
2387 (pa_can_change_mode_class): ...this new function, inverting the
2388 sense of the return value.
2389 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2390 than CANNOT_CHANGE_MODE_CLASS in comments.
2391 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2392 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2393 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2394 (pdp11_cannot_change_mode_class): Replace with...
2395 (pdp11_can_change_mode_class): ...this new function, inverting the
2396 sense of the return value.
2397 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2398 * config/powerpcspe/powerpcspe-protos.h
2399 (rs6000_cannot_change_mode_class_ptr): Delete.
2400 * config/powerpcspe/powerpcspe.c
2401 (rs6000_cannot_change_mode_class_ptr): Delete.
2402 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2403 (rs6000_option_override_internal): Assign to
2404 targetm.can_change_mode_class instead of
2405 rs6000_cannot_change_mode_class_ptr.
2406 (rs6000_cannot_change_mode_class): Replace with...
2407 (rs6000_can_change_mode_class): ...this new function, inverting the
2408 sense of the return value.
2409 (rs6000_debug_cannot_change_mode_class): Replace with...
2410 (rs6000_debug_can_change_mode_class): ...this new function.
2411 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2412 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2413 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2414 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2415 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2417 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2418 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2419 (rs6000_option_override_internal): Assign to
2420 targetm.can_change_mode_class instead of
2421 rs6000_cannot_change_mode_class_ptr.
2422 (rs6000_cannot_change_mode_class): Replace with...
2423 (rs6000_can_change_mode_class): ...this new function, inverting the
2424 sense of the return value.
2425 (rs6000_debug_cannot_change_mode_class): Replace with...
2426 (rs6000_debug_can_change_mode_class): ...this new function.
2427 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2428 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2429 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2430 (s390_can_change_mode_class): ...this new function, inverting the
2431 sense of the return value.
2432 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2433 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2434 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2435 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2436 (sh_cannot_change_mode_class): Replace with...
2437 (sh_can_change_mode_class): ...this new function, inverting the
2438 sense of the return value.
2439 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2440 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2441 (sparc_can_change_mode_class): New function.
2442 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2443 * config/spu/spu.c (spu_can_change_mode_class): New function.
2444 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2445 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2446 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2447 (visium_can_change_mode_class): New function.
2448 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2450 2017-09-15 Richard Biener <rguenther@suse.de>
2452 PR tree-optimization/82217
2453 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2454 but not undefined case.
2456 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2459 * postreload.c (reload_cse_simplify_operands): Skip
2460 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2462 2017-09-15 Richard Biener <rguenther@suse.de>
2464 PR tree-optimization/68823
2465 * graphite-scop-detection.c (build_alias_set): If we have a
2466 possible dependence check whether we can handle them by just
2467 looking at the DRs DR_ACCESS_FNs.
2468 (build_scops): If build_alias_set fails, fail the SCOP.
2470 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2472 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2473 to support float128 built-in functions that require the ISA 3.0
2475 (BU_FLOAT128_3_HW): Likewise.
2476 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2478 (FMAF128): Likewise.
2480 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2481 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2482 floating point instructions.
2483 (rs6000_invalid_builtin): Likewise.
2484 (rs6000_builtin_mask_names): Likewise.
2485 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2486 (RS6000_BTM_FLOAT128_HW): Likewise.
2487 (RS6000_BTM_COMMON): Likewise.
2488 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2490 * doc/extend.texi (RS/6000 built-in functions): Document the
2491 IEEE 128-bit floating point square root and fused multiply-add
2494 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2496 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2497 reg (r2) isn't in the set of registers defined in the prologue.
2499 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2500 Alan Hayward <alan.hayward@arm.com>
2501 David Sherwood <david.sherwood@arm.com>
2503 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2504 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2505 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2506 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2507 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2509 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2510 max_vectorization_factor.
2511 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2513 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2514 Alan Hayward <alan.hayward@arm.com>
2515 David Sherwood <david.sherwood@arm.com>
2517 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2518 (vect_worthwhile_without_simd_p): Declare.
2519 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2520 (vectorizable_reduction): Use it.
2521 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2522 (vectorizable_operation): Likewise.
2524 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2525 Alan Hayward <alan.hayward@arm.com>
2526 David Sherwood <david.sherwood@arm.com>
2528 * tree-vectorizer.h (vect_get_num_copies): New function.
2529 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2530 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2531 (vectorizable_induction): Likewise.
2532 (vectorizable_live_operation): Likewise.
2533 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2534 (vectorizable_bswap): Likewise.
2535 (vectorizable_call): Likewise.
2536 (vectorizable_conversion): Likewise.
2537 (vectorizable_assignment): Likewise.
2538 (vectorizable_shift): Likewise.
2539 (vectorizable_operation): Likewise.
2540 (vectorizable_store): Likewise.
2541 (vectorizable_load): Likewise.
2542 (vectorizable_condition): Likewise.
2543 (vectorizable_comparison): Likewise.
2544 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2546 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2547 Alan Hayward <alan.hayward@arm.com>
2548 David Sherwood <david.sherwood@arm.com>
2550 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2551 of vect_init_vector.
2553 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2554 Alan Hayward <alan.hayward@arm.com>
2555 David Sherwood <david.sherwood@arm.com>
2557 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2558 an inline wrapper that provides a location.
2559 (gimple_build_vector): Likewise.
2560 * gimple-fold.c (gimple_build_vector_from_val): New function.
2561 (gimple_build_vector): Likewise.
2562 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2563 functions to build the initial value. Always return a gimple value.
2564 (get_initial_defs_for_reduction): Likewise. Only compute
2566 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2567 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2568 (vectorizable_induction): Use gimple_build_vector rather than
2571 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2572 Alan Hayward <alan.hayward@arm.com>
2573 David Sherwood <david.sherwood@arm.com>
2575 * target.h (vec_perm_indices): New typedef.
2576 (auto_vec_perm_indices): Likewise.
2577 * optabs-query.h: Include target.h
2578 (can_vec_perm_p): Take a vec_perm_indices *.
2579 * optabs-query.c (can_vec_perm_p): Likewise.
2580 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2581 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2582 * tree-vect-generic.c (lower_vec_perm): Likewise.
2583 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2584 (vect_grouped_load_supported): Likewise.
2585 (vect_shift_permute_load_chain): Likewise.
2586 (vect_permute_store_chain): Use auto_vec_perm_indices.
2587 (vect_permute_load_chain): Likewise.
2588 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2589 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2590 Update uses of can_vec_perm_p.
2591 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2592 mode with a number of elements. Take a vec_perm_indices *.
2593 (vect_create_epilog_for_reduction): Update accordingly.
2594 Use auto_vec_perm_indices.
2595 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2596 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2597 (vect_transform_slp_perm_load): Likewise.
2598 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2599 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2600 (vect_gen_perm_mask_checked): Likewise.
2601 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2602 (vect_gen_perm_mask_checked): Likewise.
2603 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2604 (vectorizable_store): Likewise.
2605 (vectorizable_load): Likewise.
2606 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2607 (vectorizable_bswap): Likewise.
2609 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2610 Alan Hayward <alan.hayward@arm.com>
2611 David Sherwood <david.sherwood@arm.com>
2613 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2614 * tree.c (build_vector): Likewise.
2615 (build_vector_from_ctor): Update accordingly.
2616 (build_vector_from_val): Likewise.
2617 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2618 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2619 * tree-vect-generic.c (add_rshift): Likewise.
2620 (expand_vector_divmod): Likewise.
2621 (optimize_vector_constructor): Likewise.
2622 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2623 (vect_transform_slp_perm_load): Likewise.
2624 (vect_schedule_slp_instance): Likewise.
2625 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2626 (vectorizable_call): Likewise.
2627 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2628 * expmed.c (make_tree): Likewise.
2629 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2630 a vector passed to build_vector.
2631 (fold_convert_const): Likewise.
2632 (exact_inverse): Likewise.
2633 (fold_ternary_loc): Likewise.
2634 (fold_relational_const): Likewise.
2635 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2636 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2637 (const_unop): Likewise. Store the reduction accumulator in a
2638 variable rather than an array.
2639 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2640 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2641 the new vector, rather than constructing it after the input arrays.
2642 (native_interpret_vector): Use auto_vec<tree> when building
2643 a vector passed to build_vector. Add elements in order.
2644 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2645 auto_vec<tree> when building a vector passed to build_vector.
2646 (vect_create_epilog_for_reduction): Likewise.
2647 (vectorizable_induction): Likewise.
2648 (get_initial_def_for_reduction): Likewise. Fix indentation of
2650 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2652 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2653 passed to build_vector.
2655 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2656 Alan Hayward <alan.hayward@arm.com>
2657 David Sherwood <david.sherwood@arm.com>
2659 * tree-core.h (tree_base::u): Add an "nelts" field.
2660 (tree_vector): Use VECTOR_CST_NELTS as the length.
2661 * tree.c (tree_size): Likewise.
2662 (make_vector): Initialize VECTOR_CST_NELTS.
2663 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2664 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2665 TYPE_VECTOR_SUBPARTS.
2666 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2667 as the number of units, setting it from VECTOR_CST_NELTS.
2668 (const_vector_from_tree): Likewise.
2669 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2670 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2671 (fold_negate_expr_1): Likewise.
2672 (fold_convert_const): Likewise.
2673 (const_binop): Likewise. Differentiate the number of output and
2675 (const_unop): Likewise.
2676 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2677 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2678 in cases that did the opposite.
2680 2017-09-14 Richard Biener <rguenther@suse.de>
2682 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2685 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2687 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2689 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2692 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2693 if and where to split a bb, except for splitting before debug insn
2694 sequences followed by non-label real insn. Delete debug insns
2695 in between basic blocks.
2697 * combine.c (make_compound_operation_int): Formatting fixes.
2699 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2700 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2701 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2702 * config/sol2.h (LINK_EH_SPEC): Likewise.
2703 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2704 * config/s390/linux.h (LINK_SPEC): Likewise.
2705 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2706 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2707 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2708 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2709 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2710 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2711 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2712 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2714 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2716 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2718 (ENDFILE_LINUX_SPEC): Likewise.
2719 (LINK_EH_SPEC): Likewise.
2720 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2721 (LINK_OS_LINUX_SPEC32): Likewise.
2722 (LINK_OS_LINUX_SPEC64): Likewise.
2723 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2724 (LINK_OS_LINUX_SPEC): Likewise.
2726 2017-09-13 Martin Liska <mliska@suse.cz>
2729 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2730 CASE_HIGH is NULL_TREE.
2732 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2733 Alan Hayward <alan.hayward@arm.com>
2734 David Sherwood <david.sherwood@arm.com>
2736 * target.def (secondary_memory_needed): New hook.
2737 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2738 instead of SECONDARY_MEMORY_NEEDED.
2739 (secondary_memory_needed_mode): Likewise.
2740 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2741 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2742 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2743 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2744 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2745 * doc/tm.texi: Regenerate.
2746 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2747 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2748 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2749 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2750 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2751 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2752 mode argument first and change the reg_class arguments to reg_class_t.
2753 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2754 Make static. Update the call to inline_secondary_memory_needed.
2755 (ix86_register_move_cost): Update the call to
2756 inline_secondary_memory_needed.
2757 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2758 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2760 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2761 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2763 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2764 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2765 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2766 and match hook interface. Add comment from mips.h.
2767 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2768 * config/mmix/mmix.md (truncdfsf2): Refer to
2769 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2771 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2772 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2773 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2774 (pa_secondary_memory_needed): New function.
2775 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2776 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2777 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2778 (pdp11_secondary_memory_needed): Make static and match hook interface.
2779 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2780 * config/powerpcspe/powerpcspe-protos.h
2781 (rs6000_secondary_memory_needed_ptr): Delete.
2782 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2784 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2785 (rs6000_option_override_internal): Assign to
2786 targetm.secondary_memory_needed rather than
2787 rs6000_secondary_memory_needed_ptr.
2788 (rs6000_secondary_memory_needed): Match hook interface.
2789 (rs6000_debug_secondary_memory_needed): Likewise.
2790 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2791 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2792 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2793 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2794 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2795 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2797 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2798 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2799 (rs6000_option_override_internal): Assign to
2800 targetm.secondary_memory_needed rather than
2801 rs6000_secondary_memory_needed_ptr.
2802 (rs6000_secondary_memory_needed): Match hook interface.
2803 (rs6000_debug_secondary_memory_needed): Likewise.
2804 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2805 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2806 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2807 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2808 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2809 (sparc_secondary_memory_needed): New function.
2810 * lra-constraints.c (check_and_process_move): Refer to
2811 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2813 (curr_insn_transform): Likewise.
2814 (process_alt_operands): Use targetm.secondary_memory_needed
2815 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2816 (check_secondary_memory_needed_p): Likewise.
2817 (choose_split_class): Likewise.
2818 * reload.c: Unconditionally include code that was previously
2819 conditional on SECONDARY_MEMORY_NEEDED.
2820 (push_secondary_reload): Use targetm.secondary_memory_needed
2821 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2822 (push_reload): Likewise.
2823 * reload1.c: Unconditionally include code that was previously
2824 conditional on SECONDARY_MEMORY_NEEDED.
2825 (choose_reload_regs): Use targetm.secondary_memory_needed
2826 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2827 (gen_reload): Likewise.
2828 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2830 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2831 Alan Hayward <alan.hayward@arm.com>
2832 David Sherwood <david.sherwood@arm.com>
2834 * target.def (secondary_memory_needed_mode): New hook:
2835 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2836 * targhooks.h (default_secondary_memory_needed_mode): New function.
2837 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2838 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2839 * doc/tm.texi: Regenerate.
2840 * lra-constraints.c (check_and_process_move): Use
2841 targetm.secondary_memory_needed_mode instead of
2842 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2843 (curr_insn_transform): Likewise.
2844 * reload.c (get_secondary_mem): Likewise.
2845 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2846 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2848 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2849 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2850 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2851 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2852 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2854 * config/powerpcspe/powerpcspe-protos.h
2855 (rs6000_secondary_memory_needed_mode): Delete.
2856 * config/powerpcspe/powerpcspe.c
2857 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2858 (rs6000_secondary_memory_needed_mode): Make static.
2859 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2860 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2862 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2864 (rs6000_secondary_memory_needed_mode): Make static.
2865 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2866 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2867 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2868 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2869 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2871 (sparc_secondary_memory_needed_mode): New function.
2872 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2874 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2876 * config/aarch64/constraints.md (Umq): New constraint.
2877 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2879 (mov<mode>): Update condition.
2881 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2883 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2884 when bitposition is the same.
2886 2017-09-13 Richard Biener <rguenther@suse.de>
2888 * dwarf2out.c (output_die_symbol): Remove.
2889 (output_die): Do not output a DIEs symbol.
2891 2017-09-13 Richard Biener <rguenther@suse.de>
2894 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2895 default-def to avoid breaking iterator update with the weird
2896 interaction with cgraph_update_edges_for_call_stmt_node.
2898 2017-09-13 Richard Biener <rguenther@suse.de>
2900 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2901 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2902 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2903 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2905 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2907 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2908 Disable pc relative literal load irrespective of
2909 TARGET_FIX_ERR_A53_84341 for default.
2911 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2913 * config/sparc/sparc.c (output_return): Output the source location of
2914 the insn in the delay slot, if any.
2915 (output_sibcall): Likewise.
2917 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2920 * common.opt (-static-pie): New alias.
2921 (shared): Negate static-pie.
2922 (-no-pie): Update help text.
2924 (static-pie): New option.
2925 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2926 -static-pie support.
2927 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2928 (LINK_EH_SPEC): Likewise.
2929 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2930 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2931 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2932 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2933 (init_gcc_specs): Likewise.
2934 (init_spec): Likewise.
2935 (display_help): Update help message for -pie.
2936 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2939 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2941 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2942 (movdi_aarch64): Likewise.
2943 (movti_aarch64): Likewise.
2945 2017-09-12 Simon Wright <simon@pushface.org>
2948 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2949 calculation of the minor version, always output as 0.
2951 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2955 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2956 on it early, rather than manual conversion late. For
2957 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2958 instead of performing manual conversion.
2960 2017-09-12 Carl Love <cel@us.ibm.com>
2962 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2963 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2965 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2966 VMULOSW): Add definitions.
2967 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2968 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2969 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2970 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2971 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2973 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2975 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2977 (movti_aarch64): Likewise.
2978 (movdf_aarch64): Likewise.
2979 (movtf_aarch64): Likewise.
2980 (load_pairdi): Likewise.
2981 (store_pairdi): Likewise.
2982 (load_pairdf): Likewise.
2983 (store_pairdf): Likewise.
2984 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2985 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2986 (ldr_got_small_<mode>): Likewise.
2987 (ldr_got_small_28k_<mode>): Likewise.
2988 (ldr_got_tiny): Likewise.
2989 * config/aarch64/iterators.md (ldst_sz): New.
2990 (ldpstp_sz): Likewise.
2991 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2993 (thunderx_load): Split load_8 to load_16.
2994 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2996 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2997 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2998 (xgene1_store_pair): Split store_8 to store_16.
2999 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3000 (falkor_st_0_st_sd): Split store_8 to store_16.
3002 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3004 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3005 and store1/2/3/4 to store_4/8/12/16.
3006 * config/aarch64/aarch64.md: Update for rename.
3007 * config/arm/arm.md: Likewise.: Likewise.
3008 * config/arm/arm.c: Likewise.
3009 * config/arm/thumb1.md: Likewise.
3010 * config/arm/thumb2.md: Likewise.
3011 * config/arm/vfp.md: Likewise.
3012 * config/arm/arm-generic.md: Likewise.
3013 * config/arm/arm1020e.md: Likewise.
3014 * config/arm/arm1026ejs.md: Likewise.
3015 * config/arm/arm1136jfs.md: Likewise.
3016 * config/arm/arm926ejs.md: Likewise.
3017 * config/arm/cortex-a15.md: Likewise.
3018 * config/arm/cortex-a17.md: Likewise.
3019 * config/arm/cortex-a5.md: Likewise.
3020 * config/arm/cortex-a53.md: Likewise.
3021 * config/arm/cortex-a57.md: Likewise.
3022 * config/arm/cortex-a7.md: Likewise.
3023 * config/arm/cortex-a8.md: Likewise.
3024 * config/arm/cortex-a9.md: Likewise.
3025 * config/arm/cortex-m4.md: Likewise.
3026 * config/arm/cortex-m7.md: Likewise.
3027 * config/arm/cortex-r4.md: Likewise.
3028 * config/arm/exynos-m1.md: Likewise.
3029 * config/arm/fa526.md: Likewise.
3030 * config/arm/fa606te.md: Likewise.
3031 * config/arm/fa626te.md: Likewise.
3032 * config/arm/fa726te.md: Likewise.
3033 * config/arm/fmp626.md: Likewise.
3034 * config/arm/iwmmxt.md: Likewise.
3035 * config/arm/ldmstm.md: Likewise.
3036 * config/arm/marvell-pj4.md: Likewise.
3037 * config/arm/xgene1.md: Likewise.
3038 * config/aarch64/thunderx.md: Likewise.
3039 * config/aarch64/thunderx2t99.md: Likewise.
3040 * config/aarch64/falkor.md: Likewise.
3042 2017-09-12 Martin Liska <mliska@suse.cz>
3044 * attribs.c (private_lookup_attribute): New function.
3045 * attribs.h (private_lookup_attribute): Declared here.
3046 (lookup_attribute): Called from this place.
3048 2017-09-12 Richard Biener <rguenther@suse.de>
3050 PR tree-optimization/82157
3051 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3052 stmts with side-effects.
3054 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3055 Alan Hayward <alan.hayward@arm.com>
3056 David Sherwood <david.sherwood@arm.com>
3058 * target.def (hard_regno_nregs): New hook.
3059 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3060 * targhooks.h (default_hard_regno_nregs): Declare.
3061 * targhooks.c (default_hard_regno_nregs): New function.
3062 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3063 (TARGET_HARD_REGNO_NREGS): ...this hook.
3064 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3065 (CLASS_MAX_NREGS): Likewise.
3066 * doc/tm.texi: Regenerate.
3067 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3068 instead of HARD_REGNO_NREGS.
3069 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3070 HARD_REGNO_NREGS in the comment.
3071 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3072 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3073 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3074 Return an unsigned int.
3075 (TARGET_HARD_REGNO_NREGS): Redefine.
3076 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3077 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3078 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3079 (arc_hard_regno_nregs): New function.
3080 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3081 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3082 (arm_hard_regno_nregs): New function.
3083 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3084 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3085 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3086 (TARGET_HARD_REGNO_NREGS): Redefine.
3087 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3088 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3089 (HARD_REGNO_NREGS): Delete.
3090 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3091 (cr16_hard_regno_nregs): New function.
3092 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3093 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3094 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3095 (cris_hard_regno_nregs): New function.
3096 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3097 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3098 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3099 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3100 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3101 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3102 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3103 (frv_hard_regno_nregs): Make static. Take and return an
3105 (frv_class_max_nregs): Remove outdated copy of documentation.
3106 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3107 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3108 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3109 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3110 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3111 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3112 (TARGET_HARD_REGNO_NREGS): Redefine.
3113 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3114 (CLASS_MAX_NREGS): Update comment.
3115 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3116 (ia64_hard_regno_nregs): New function.
3117 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3118 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3119 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3120 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3121 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3123 (m32c_hard_regno_nregs): Likewise. Make static.
3124 (TARGET_HARD_REGNO_NREGS): Redefine.
3125 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3126 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3127 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3128 (m68k_hard_regno_nregs): New function.
3129 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3130 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3131 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3132 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3133 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3134 Take and return an unsigned int.
3135 (TARGET_HARD_REGNO_NREGS): Redefine.
3136 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3137 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3138 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3139 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3140 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3141 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3142 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3143 (msp430_hard_regno_nregs): Make static. Take and return an
3145 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3146 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3147 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3148 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3149 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3150 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3151 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3152 (TARGET_HARD_REGNO_NREGS): Redefine.
3153 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3154 (PA_HARD_REGNO_NREGS): ...this.
3155 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3156 (PA_HARD_REGNO_NREGS): ...this.
3157 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3158 (pa_hard_regno_nregs): New function.
3159 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3160 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3161 (pdp11_hard_regno_nregs): New function.
3162 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3163 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3164 (rs6000_hard_regno_nregs_hook): New function.
3165 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3166 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3167 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3168 Take and return an unsigned int. Move earlier in file.
3169 (TARGET_HARD_REGNO_NREGS): Redefine.
3170 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3171 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3172 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3173 (rl78_hard_regno_nregs): Make static. Take and return an
3175 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3176 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3177 (rs6000_hard_regno_nregs_hook): New function.
3178 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3179 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3180 (TARGET_HARD_REGNO_NREGS): Redefine.
3181 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3182 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3183 instead of HARD_REGNO_NREGS.
3184 (s390_hard_regno_nregs): New function.
3185 (s390_hard_regno_mode_ok): Add comment from s390.h.
3186 (TARGET_HARD_REGNO_NREGS): Redefine.
3187 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3188 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3189 (sh_hard_regno_nregs): New function.
3190 (sh_pass_in_reg_p): Use it.
3191 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3192 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3193 (sparc_hard_regno_nregs): New function.
3194 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3195 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3196 (spu_function_arg_advance): Use it, supplying a valid register number.
3197 (TARGET_HARD_REGNO_NREGS): Redefine.
3198 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3199 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3200 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3201 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3202 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3203 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3204 (CLASS_MAX_NREGS): Remove copy of old documentation.
3205 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3206 (visium_hard_regno_nregs): New function.
3207 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3208 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3209 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3210 xtensa_hard_regno_nregs): New function.
3211 * system.h (HARD_REGNO_NREGS): Poison.
3213 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3215 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3216 hard_regno_nregs instead of HARD_REGNO_NREGS.
3217 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3218 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3219 (c6x_expand_epilogue): Likewise.
3220 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3221 (frv_read_iacc_argument): Likewise.
3222 * config/sh/sh.c: Include regs.h.
3223 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3224 (regs_used): Likewise.
3225 (output_stack_adjust): Likewise.
3226 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3227 * expmed.c: Include regs.h.
3228 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3229 * ree.c: Include regs.h.
3230 (combine_reaching_defs): Use hard_regno_nregs instead of
3232 (add_removable_extension): Likewise.
3234 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3236 * regs.h (hard_regno_nregs): Turn into a function.
3237 (end_hard_regno): Update accordingly.
3238 * caller-save.c (setup_save_areas): Likewise.
3239 (save_call_clobbered_regs): Likewise.
3240 (replace_reg_with_saved_mem): Likewise.
3241 (insert_restore): Likewise.
3242 (insert_save): Likewise.
3243 * combine.c (can_change_dest_mode): Likewise.
3244 (move_deaths): Likewise.
3245 (distribute_notes): Likewise.
3246 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3247 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3248 (rs6000_split_multireg_move): Likewise.
3249 (rs6000_register_move_cost): Likewise.
3250 (rs6000_memory_move_cost): Likewise.
3251 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3252 (rs6000_split_multireg_move): Likewise.
3253 (rs6000_register_move_cost): Likewise.
3254 (rs6000_memory_move_cost): Likewise.
3255 * cselib.c (cselib_reset_table): Likewise.
3256 (cselib_lookup_1): Likewise.
3257 * emit-rtl.c (set_mode_and_regno): Likewise.
3258 * function.c (aggregate_value_p): Likewise.
3259 * ira-color.c (setup_profitable_hard_regs): Likewise.
3260 (check_hard_reg_p): Likewise.
3261 (calculate_saved_nregs): Likewise.
3262 (assign_hard_reg): Likewise.
3263 (improve_allocation): Likewise.
3264 (calculate_spill_cost): Likewise.
3265 * ira-emit.c (modify_move_list): Likewise.
3266 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3267 (ira_hard_reg_in_set_p): Likewise.
3268 * ira.c (setup_reg_mode_hard_regset): Likewise.
3269 (clarify_prohibited_class_mode_regs): Likewise.
3270 (check_allocation): Likewise.
3271 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3272 (lra_setup_reg_renumber): Likewise.
3273 (setup_try_hard_regno_pseudos): Likewise.
3274 (spill_for): Likewise.
3275 (assign_hard_regno): Likewise.
3276 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3277 * lra-constraints.c (in_class_p): Likewise.
3278 (lra_constraint_offset): Likewise.
3279 (simplify_operand_subreg): Likewise.
3280 (lra_constraints): Likewise.
3281 (split_reg): Likewise.
3282 (split_if_necessary): Likewise.
3283 (invariant_p): Likewise.
3284 (inherit_in_ebb): Likewise.
3285 * lra-lives.c (process_bb_lives): Likewise.
3286 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3287 (get_hard_regs): Likewise.
3288 (do_remat): Likewise.
3289 * lra-spills.c (assign_spill_hard_regs): Likewise.
3290 * mode-switching.c (create_pre_exit): Likewise.
3291 * postreload.c (reload_combine_recognize_pattern): Likewise.
3292 * recog.c (peep2_find_free_register): Likewise.
3293 * regcprop.c (kill_value_regno): Likewise.
3294 (set_value_regno): Likewise.
3295 (copy_value): Likewise.
3296 (maybe_mode_change): Likewise.
3297 (find_oldest_value_reg): Likewise.
3298 (copyprop_hardreg_forward_1): Likewise.
3299 * regrename.c (check_new_reg_p): Likewise.
3300 (regrename_do_replace): Likewise.
3301 * reload.c (push_reload): Likewise.
3302 (combine_reloads): Likewise.
3303 (find_dummy_reload): Likewise.
3304 (operands_match_p): Likewise.
3305 (find_reloads): Likewise.
3306 (find_equiv_reg): Likewise.
3307 (reload_adjust_reg_for_mode): Likewise.
3308 * reload1.c (count_pseudo): Likewise.
3309 (count_spilled_pseudo): Likewise.
3310 (find_reg): Likewise.
3311 (clear_reload_reg_in_use): Likewise.
3312 (free_for_value_p): Likewise.
3313 (allocate_reload_reg): Likewise.
3314 (choose_reload_regs): Likewise.
3315 (reload_adjust_reg_for_temp): Likewise.
3316 (emit_reload_insns): Likewise.
3317 (delete_output_reload): Likewise.
3318 * rtlanal.c (subreg_get_info): Likewise.
3319 * sched-deps.c (sched_analyze_reg): Likewise.
3320 * sel-sched.c (init_regs_for_mode): Likewise.
3321 (mark_unavailable_hard_regs): Likewise.
3322 (choose_best_reg_1): Likewise.
3323 (verify_target_availability): Likewise.
3324 * valtrack.c (dead_debug_insert_temp): Likewise.
3325 * var-tracking.c (track_loc_p): Likewise.
3326 (emit_note_insn_var_location): Likewise.
3327 * varasm.c (make_decl_rtl): Likewise.
3328 * reginfo.c (choose_hard_reg_mode): Likewise.
3329 (init_reg_modes_target): Refer directly to
3330 this_target_regs->x_hard_regno_nregs.
3332 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3334 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3335 instead of hard_regno_nregs.
3337 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3339 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3340 end_hard_regno instead of hard_regno_nregs.
3341 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3342 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3343 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3344 * ira-color.c (improve_allocation): Likewise.
3345 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3346 * lra-lives.c (mark_regno_live): Likewise.
3347 (mark_regno_dead): Likewise.
3348 * lra-remat.c (operand_to_remat): Likewise.
3349 * lra.c (collect_non_operand_hard_regs): Likewise.
3350 * postreload.c (reload_combine_note_store): Likewise.
3351 (move2add_valid_value_p): Likewise.
3352 * reload.c (regno_clobbered_p): Likewise.
3354 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3356 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3358 * config/v850/v850.c (v850_reorg): Likewise.
3359 * reload.c (refers_to_regno_for_reload_p): Likewise.
3360 (find_equiv_reg): Likewise.
3361 * reload1.c (reload_reg_reaches_end_p): Likewise.
3363 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3365 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3367 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3368 * config/arm/arm.c (output_move_neon): Likewise.
3369 (arm_attr_length_move_neon): Likewise.
3370 (neon_split_vcombine): Likewise.
3371 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3372 (c6x_mark_reg_written): Likewise.
3373 (c6x_dwarf_register_span): Likewise.
3374 * config/i386/i386.c (ix86_save_reg): Likewise.
3375 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3376 (rws_access_reg): Likewise.
3377 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3378 * mode-switching.c (create_pre_exit): Likewise.
3379 * ree.c (combine_reaching_defs): Likewise.
3380 (add_removable_extension): Likewise.
3381 * regcprop.c (find_oldest_value_reg): Likewise.
3382 (copyprop_hardreg_forward_1): Likewise.
3383 * reload.c (reload_inner_reg_of_subreg): Likewise.
3384 (push_reload): Likewise.
3385 (combine_reloads): Likewise.
3386 (find_dummy_reload): Likewise.
3387 (reload_adjust_reg_for_mode): Likewise.
3388 * reload1.c (find_reload_regs): Likewise.
3389 (forget_old_reloads_1): Likewise.
3390 (reload_reg_free_for_value_p): Likewise.
3391 (reload_adjust_reg_for_temp): Likewise.
3392 (emit_reload_insns): Likewise.
3393 (delete_output_reload): Likewise.
3394 * sel-sched.c (choose_best_reg_1): Likewise.
3395 (choose_best_pseudo_reg): Likewise.
3397 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3398 Alan Hayward <alan.hayward@arm.com>
3399 David Sherwood <david.sherwood@arm.com>
3401 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3402 * target.def (slow_unaligned_access): New hook.
3403 * targhooks.h (default_slow_unaligned_access): Declare.
3404 * targhooks.c (default_slow_unaligned_access): New function.
3405 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3406 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3407 * doc/tm.texi: Regenerate.
3408 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3409 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3410 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3412 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3413 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3415 (rs6000_slow_unaligned_access): New function.
3416 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3417 (expand_block_compare): Likewise.
3418 (expand_strn_compare): Likewise.
3419 (rs6000_rtx_costs): Likewise.
3420 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3421 (riscv_slow_unaligned_access): Likewise.
3422 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3423 (riscv_slow_unaligned_access_p): ...this and make static.
3424 (riscv_option_override): Update accordingly.
3425 (riscv_slow_unaligned_access): New function.
3426 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3427 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3428 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3429 (rs6000_slow_unaligned_access): New function.
3430 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3431 (rs6000_rtx_costs): Likewise.
3432 * config/rs6000/rs6000-string.c (expand_block_compare)
3433 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3434 of SLOW_UNALIGNED_ACCESS.
3435 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3436 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3437 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3438 of SLOW_UNALIGNED_ACCESS.
3439 * expmed.c (simple_mem_bitfield_p): Likewise.
3440 * expr.c (alignment_for_piecewise_move): Likewise.
3441 (emit_group_load_1): Likewise.
3442 (emit_group_store): Likewise.
3443 (copy_blkmode_from_reg): Likewise.
3444 (emit_push_insn): Likewise.
3445 (expand_assignment): Likewise.
3446 (store_field): Likewise.
3447 (expand_expr_real_1): Likewise.
3448 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3449 * lra-constraints.c (simplify_operand_subreg): Likewise.
3450 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3451 * gimple-ssa-store-merging.c: Likewise in block comment at start
3453 * tree-ssa-strlen.c: Include target.h.
3454 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3455 of SLOW_UNALIGNED_ACCESS.
3456 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3458 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3460 PR rtl-optimization/82185
3461 * expmed.c (emit_store_flag_int): Only test tem if it has been
3464 2017-09-12 Richard Biener <rguenther@suse.de>
3467 * match.pd ((FTYPE) N CMP CST): Fix typo.
3469 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3471 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3473 (mips_near_type_p): Add 'short_call' attribute as a synonym
3475 * doc/extend.texi (short_call): Document new function attribute.
3477 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3480 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3481 assertion check that in the condition.
3482 (get_atomic_generic_size): Likewise. Before testing if parameter
3483 has pointer type, if it has array type, call for C++
3484 default_conversion to perform array-to-pointer conversion.
3486 2017-09-12 Richard Biener <rguenther@suse.de>
3488 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3489 for operations we cannot scalarize.
3491 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3493 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3494 vectors heap vectors. Clean up comments.
3495 Make visited_bbs a reference.
3496 (profitable_jump_thread_path): Make GC
3497 vectors heap vectors. Clean up comments.
3499 (convert_and_register_jump_thread_path): Make GC vectors heap
3501 (check_subpath_and_update_thread_path): Same. Clean up comments.
3502 Make visited_bbs a reference.
3503 (handle_phi): Abstract common code to to
3504 register_jump_thread_path_if_profitable.
3505 Rename VAR_BB to DEF_BB.
3507 Make GC vectors heap vectors.
3508 Make visited_bbs a reference.
3509 (handle_assignment): Same.
3510 (register_jump_thread_path_if_profitable): New.
3511 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3513 Make GC vectors heap vectors. Clean up comments.
3514 Make visited_bbs a reference.
3515 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3516 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3519 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3522 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3523 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3525 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3527 Revert r251800 and r251799.
3529 2017-09-11 Martin Jambor <mjambor@suse.cz>
3532 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3533 arguments in advance.
3534 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3535 use it to find predecessor edges.
3536 (naive_outof_ssa): Collect vector of predecessors.
3538 2017-09-08 Jason Merrill <jason@redhat.com>
3540 PR c++/70029 - ICE with ref-qualifier and -flto
3541 * langhooks.h (struct lang_hooks_for_types): Add
3542 copy_lang_qualifiers.
3543 * attribs.c (build_type_attribute_qual_variant): Use it.
3544 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3546 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3547 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3549 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3552 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3553 (*mulsi3_sp64): New instruction.
3554 (mulsi3): New expander.
3556 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3558 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3560 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3562 * sancov.c: Include memmodel.h.
3564 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3567 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3570 2017-09-07 Carl Love <cel@us.ibm.com>
3572 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3573 the sldi instruction.
3575 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3577 * sancov.c: Include tm_p.h.
3579 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3582 * output.h (switch_to_other_text_partition): New declaration.
3583 * varasm.c (switch_to_other_text_partition): New function.
3584 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3585 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3586 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3587 to the other text partition before emitting LCL label and switch back
3588 after emitting the word after it.
3590 2017-09-07 Richard Biener <rguenther@suse.de>
3592 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3593 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3594 critical edges here, after loop init.
3595 (pass_data_pre): Remove PROP_no_crit_edges flags.
3596 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3597 for valueization of call args to avoid leaking VN_TOP.
3598 (visit_use): Assert we do not visit default defs.
3599 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3600 Use error_mark_node to more easily detect leaking VN_TOP.
3601 All default-defs are varying, not VN_TOP. Mark them visited.
3602 (run_scc_vn): Make code match comment.
3604 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3606 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3607 OPTION_MASK_FLOAT128_KEYWORD.
3608 (POWERPC_MASKS): Likewise.
3609 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3610 support for the -mfloat128-type option, and make -mfloat128
3611 default on PowerPC Linux systems. Define or undefine
3612 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3613 Define __float128 to be __ieee128 if IEEE 128-bit support is
3614 enabled, or undefine it.
3615 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3616 Delete defining __FLOAT128_TYPE__.
3617 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3618 -mfloat128-type option and make -mfloat128 default on PowerPC
3620 (TARGET_FLOAT128_TYPE): Likewise.
3621 (-mfloat128-type): Likewise.
3622 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3623 Delete the -mfloat128-type option and make -mfloat128 default on
3624 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3625 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3626 128-bit floating point is enabled. Change tests from using
3627 -mfloat128-type to -mfloat128.
3628 (rs6000_mangle_type): Use the correct mangling for the __float128
3629 type even if normal long double is restricted to 64-bits.
3630 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3632 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3633 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3634 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3636 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3637 documentation for -mfloat128.
3639 2017-09-06 Olivier Hainque <hainque@adacore.com>
3641 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3643 2017-09-06 Wish Wu <wishwu007@gmail.com>
3644 Jakub Jelinek <jakub@redhat.com>
3646 * asan.c (initialize_sanitizer_builtins): Add
3647 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3648 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3649 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3650 BT_FN_VOID_UINT64_PTR variables.
3651 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3652 (BT_FN_VOID_UINT16_UINT16): Likewise.
3653 (BT_FN_VOID_UINT32_UINT32): Likewise.
3654 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3655 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3656 (BT_FN_VOID_UINT64_PTR): Likewise.
3657 * common.opt (flag_sanitize_coverage): New variable.
3658 (fsanitize-coverage=trace-pc): Remove.
3659 (fsanitize-coverage=): Add.
3660 * flag-types.h (enum sanitize_coverage_code): New enum.
3661 * fold-const.c (fold_range_test): Disable non-short-circuit
3662 optimization if flag_sanitize_coverage.
3663 (fold_truth_andor): Likewise.
3664 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3665 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3666 (coverage_sanitizer_opts): New array.
3667 (get_closest_sanitizer_option): Add OPTS argument, handle also
3668 OPT_fsanitize_coverage_.
3669 (parse_sanitizer_options): Adjusted to also handle
3670 OPT_fsanitize_coverage_.
3671 (common_handle_option): Add OPT_fsanitize_coverage_.
3672 * sancov.c (instrument_comparison, instrument_switch): New function.
3673 (sancov_pass): Add trace-cmp support.
3674 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3675 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3676 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3677 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3678 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3679 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3680 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3681 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3682 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3683 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3685 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3687 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3688 error. Only quit immediately if parsing is complete.
3689 (BEGIN): Initialize fatal_err and parse_done.
3690 (begin fpu, end fpu): Check number of arguments.
3691 (begin arch, end arch): Likewise.
3692 (begin cpu, end cpu): Likewise.
3693 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3694 (optalias): Likewise.
3696 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3698 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3699 * config/arm/arm-isa.h: Delete. Move definitions to ...
3700 * arm-cpus.in: ... here. Use new feature and fgroup values.
3701 * config/arm/arm.c (arm_option_override): Use lower case for feature
3703 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3704 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3705 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3707 (print_isa_bits_for): New function.
3708 (gen_isa): New function.
3709 (gen_comm_data): Use print_isa_bits_for.
3710 (define feature): New keyword.
3711 (define fgroup): New keyword.
3712 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3713 (arm-isa.h): Add rule to generate file.
3714 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3715 case for feature bit names.
3717 2017-09-06 Richard Biener <rguenther@suse.de>
3719 * tree-ssa-pre.c (NECESSARY): Remove.
3720 (create_expression_by_pieces): Do not touch pass-local flags.
3721 (insert_into_preds_of_block): Likewise.
3722 (do_pre_regular_insertion): Likewise.
3723 (eliminate_insert): Likewise.
3724 (eliminate_dom_walker::before_dom_children): Likewise.
3725 (fini_eliminate): Do not look at inserted_exprs.
3726 (mark_operand_necessary): Remove.
3727 (remove_dead_inserted_code): Replace with simple work-list
3728 algorithm based on inserted_exprs and SSA uses.
3729 (pass_pre::execute): Re-order fini_eliminate and
3730 remove_dead_inserted_code.
3732 2017-09-06 Olivier Hainque <hainque@adacore.com>
3734 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3735 for VxWorks 7. Adjust surrounding comments.
3737 2017-09-06 Richard Biener <rguenther@suse.de>
3739 * gimple-ssa-strength-reduction.c
3740 (find_candidates_dom_walker::before_dom_children): Also allow
3743 2017-09-06 Richard Biener <rguenther@suse.de>
3745 PR tree-optimization/82108
3746 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3747 for gap in the non-permutation SLP case.
3749 2017-09-06 Martin Jambor <mjambor@suse.cz>
3751 PR tree-optimization/82078
3752 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3753 add_access_to_work_queue...
3754 (build_accesses_from_assign): ...here.
3755 (propagate_all_subaccesses): Make sure racc is the group
3756 representative, if there is one.
3758 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3761 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3764 2017-09-06 Richard Biener <rguenther@suse.de>
3766 * gimple-ssa-strength-reduction.c
3767 (find_candidates_dom_walker::before_doom_children): Use a
3768 type and not a mode check.
3770 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3773 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3775 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3777 2017-09-05 Jeff Law <law@redhat.com>
3779 PR tree-optimization/64910
3780 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3781 cases where we have 3 or more operands.
3783 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3786 * omp-low.c (lower_omp_for): Recompute tree invariant if
3787 gimple_omp_for_initial/final is ADDR_EXPR.
3790 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3791 into gimple val before gimplification fo the COND_EXPR.
3793 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3795 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3796 REGION_COPY argument.
3797 (thread_through_all_blocks): Remove unused argument to
3798 duplicate_thread_path.
3800 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3801 Alan Hayward <alan.hayward@arm.com>
3802 David Sherwood <david.sherwood@arm.com>
3804 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3805 Take a scalar_mode rather than a machine_mode.
3806 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3807 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3808 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3809 (aarch64_gen_adjusted_ldpstp): Likewise.
3810 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3812 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3813 Alan Hayward <alan.hayward@arm.com>
3814 David Sherwood <david.sherwood@arm.com>
3816 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3817 Take a scalar_int_mode instead of a machine_mode.
3818 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3819 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3820 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3821 (aarch64_simd_attr_length_rglist): Delete.
3822 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3823 a scalar_int_mode instead of a machine_mode.
3824 (aarch64_add_offset): Likewise.
3825 (aarch64_internal_mov_immediate): Likewise
3826 (aarch64_add_constant_internal): Likewise.
3827 (aarch64_add_constant): Likewise.
3828 (aarch64_movw_imm): Likewise.
3829 (aarch64_rtx_arith_op_extract_p): Likewise.
3830 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3831 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3832 Remove assert that the mode isn't a vector.
3833 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3834 (aarch64_expand_mov_immediate): Update calls after above changes.
3835 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3836 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3837 (aarch64_move_imm): Likewise.
3838 (aarch64_can_const_movi_rtx_p): Likewise.
3839 (aarch64_strip_extend): Likewise.
3840 (aarch64_extr_rtx_p): Likewise.
3841 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3842 a CONST_INT when the mode parameter is VOIDmode.
3843 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3845 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3847 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3848 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3849 (bitwise_type_for_mode): Update accordingly.
3851 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3853 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3854 * stor-layout.c (mode_for_size_tree): Likewise.
3855 (mode_for_array): Update accordingly.
3856 (layout_decl): Likewise.
3857 (compute_record_mode): Likewise. Only set the mode once.
3859 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3861 * target.def (get_mask_mode): Change return type to opt_mode.
3863 * doc/tm.texi: Regenerate.
3864 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3865 * targhooks.c (default_get_mask_mode): Likewise.
3866 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3867 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3868 targetm.get_mask_mode.
3869 * tree.c (build_truth_vector_type): Likewise.
3871 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3873 * machmode.h (mode_for_vector): Return an opt_mode.
3874 * stor-layout.c (mode_for_vector): Likewise.
3875 (mode_for_int_vector): Update accordingly.
3876 (layout_type): Likewise.
3877 * config/i386/i386.c (emit_memmov): Likewise.
3878 (ix86_expand_set_or_movmem): Likewise.
3879 (ix86_expand_vector_init): Likewise.
3880 (ix86_get_mask_mode): Likewise.
3881 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3883 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3884 * expmed.c (extract_bit_field_1): Likewise.
3885 * expr.c (expand_expr_real_2): Likewise.
3886 * optabs-query.c (can_vec_perm_p): Likewise.
3887 (can_vec_mask_load_store_p): Likewise.
3888 * optabs.c (expand_vec_perm): Likewise.
3889 * targhooks.c (default_get_mask_mode): Likewise.
3890 * tree-vect-stmts.c (vectorizable_store): Likewise.
3891 (vectorizable_load): Likewise.
3892 (get_vectype_for_scalar_type_and_size): Likewise.
3894 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3896 * machmode.h (mode_for_int_vector): New function.
3897 * stor-layout.c (mode_for_int_vector): Likewise.
3898 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3899 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3900 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3901 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3902 (s390_expand_vcond): Likewise.
3904 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3906 * machmode.h (opt_machine_mode): New type.
3907 (opt_mode<T>): Allow construction from anything that can be
3909 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3910 (mode_for_size): Return an opt_machine_mode.
3911 * stor-layout.c (mode_for_size): Likewise.
3912 (mode_for_size_tree): Update call accordingly.
3913 (bitwise_mode_for_mode): Likewise.
3914 (make_fract_type): Likewise.
3915 (make_accum_type): Likewise.
3916 * caller-save.c (replace_reg_with_saved_mem): Update call
3918 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3919 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3920 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3921 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3922 * expmed.c (extract_bit_field_1): Likewise.
3923 * reload.c (get_secondary_mem): Likewise.
3924 * varasm.c (assemble_integer): Likewise.
3925 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3928 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3930 * machmode.h (decimal_float_mode_for_size): New function.
3931 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3932 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3933 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3934 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3935 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3936 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3938 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3940 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3941 (get_builtin_sync_mode): Likewise.
3942 (expand_ifn_atomic_compare_exchange): Likewise.
3943 (expand_builtin_atomic_clear): Likewise.
3944 (expand_builtin_atomic_test_and_set): Likewise.
3945 (fold_builtin_atomic_always_lock_free): Likewise.
3946 * calls.c (compute_argument_addresses): Likewise.
3947 (emit_library_call_value_1): Likewise.
3948 (store_one_arg): Likewise.
3949 * combine.c (combine_instructions): Likewise.
3950 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3951 * config/arm/arm.c (arm_function_value): Likewise.
3952 (aapcs_allocate_return_reg): Likewise.
3953 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3954 * config/i386/i386.c (construct_container): Likewise.
3955 (ix86_gimplify_va_arg): Likewise.
3956 (ix86_expand_sse_cmp): Likewise.
3957 (emit_memmov): Likewise.
3958 (emit_memset): Likewise.
3959 (expand_small_movmem_or_setmem): Likewise.
3960 (ix86_expand_pextr): Likewise.
3961 (ix86_expand_pinsr): Likewise.
3962 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3963 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3965 * config/mips/mips.c (mips_function_value_1) Likewise.
3966 (mips_block_move_straight): Likewise.
3967 (mips_expand_ins_as_unaligned_store): Likewise.
3968 * config/powerpcspe/powerpcspe.c
3969 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3970 (rs6000_darwin64_record_arg_flush): Likewise.
3971 * config/rs6000/rs6000.c
3972 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3973 (rs6000_darwin64_record_arg_flush): Likewise.
3974 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3975 (sparc_function_value_1): Likewise.
3976 * config/spu/spu.c (adjust_operand): Likewise.
3977 (spu_emit_branch_or_set): Likewise.
3978 (arith_immediate_p): Likewise.
3979 * emit-rtl.c (gen_lowpart_common): Likewise.
3980 * expr.c (expand_expr_real_1): Likewise.
3981 * function.c (assign_parm_setup_block): Likewise.
3982 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3983 * reload1.c (alter_reg): Likewise.
3984 * stor-layout.c (mode_for_vector): Likewise.
3985 (layout_type): Likewise.
3987 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3989 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3990 (spu_convert_move): Likewise.
3991 * lower-subreg.c (resolve_simple_move): Likewise.
3993 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3996 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3997 define_insn to a define_expand.
3998 (altivec_vsum2sws_direct): New define_insn.
3999 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4001 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4003 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4006 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4011 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4012 realign stack if stack alignment needed is less than incoming
4015 2017-09-05 Marek Polacek <polacek@redhat.com>
4018 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4021 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4023 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4025 2017-09-05 Richard Biener <rguenther@suse.de>
4027 PR tree-optimization/82084
4028 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4030 2017-09-05 Richard Biener <rguenther@suse.de>
4032 PR tree-optimization/82102
4033 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4035 2017-09-05 Martin Liska <mliska@suse.cz>
4037 PR tree-optimization/82032
4038 * tree-cfg.c (generate_range_test): New function.
4039 * tree-cfg.h (generate_range_test): Declared here.
4040 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4041 (cleanup_control_expr_graph): Use it.
4042 * tree-switch-conversion.c (try_switch_expansion): Remove
4044 (emit_case_nodes): Use generate_range_test.
4046 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4049 * config/i386/i386.md (*<btsc><mode>_mask): Add
4050 TARGET_USE_BT to insn constraint.
4051 (*btr<mode>_mask): Ditto.
4053 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4055 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4056 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4058 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4061 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4062 TARGET_NEON and TARGET_IWMMXT.
4063 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4064 TARGET_NEON and TARGET_IWMMXT.
4065 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4067 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4069 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4070 (ix86_rewrite_tls_address): Ditto.
4071 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4072 (ix86_rewrite_tls_address_1): Ditto.
4073 (ix86_rewrite_tls_address): Ditto.
4074 * config/i386/predicates.md (tls_address_pattern): New predicate.
4075 * config/i386/i386.md (TLS address splitter): New splitter.
4077 2017-09-04 Richard Biener <rguenther@suse.de>
4079 PR tree-optimization/82084
4080 * fold-const.h (can_native_encode_string_p): Declare.
4081 * fold-const.c (can_native_encode_string_p): Factor out from ...
4082 (native_encode_string): ... here.
4083 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4084 vectorizing stores from constants we later cannot handle.
4086 2017-09-04 Marek Polacek <polacek@redhat.com>
4089 * doc/invoke.texi: Update -Wtautological-compare documentation.
4091 2017-09-04 Jeff Law <law@redhat.com>
4093 PR tree-optimization/64910
4094 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4095 swap the first and last operand if the last is a constant.
4097 2017-09-04 Marek Polacek <polacek@redhat.com>
4100 * convert.c (do_narrow): When sanitizing signed integer overflows,
4101 bail out for signed types.
4102 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4104 2017-09-04 Richard Biener <rguenther@suse.de>
4106 PR tree-optimization/82060
4107 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4108 Move devirtualization after stmt folding and before EH/AB/noreturn
4109 cleanup to get the stmt refs canonicalized. Use a bool instead
4110 of gimple_modified_p since that doesn't work for NOPs. Schedule
4111 NOPs generated by folding for removal.
4113 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4114 Alan Hayward <alan.hayward@arm.com>
4115 David Sherwood <david.sherwood@arm.com>
4117 * coretypes.h (pad_direction): New enum.
4118 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4119 (FUNCTION_ARG_PADDING): Likewise.
4120 * target.def (function_arg_padding): New hook.
4121 * targhooks.h (default_function_arg_padding): Declare.
4122 * targhooks.c (default_function_arg_padding): New function.
4123 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4124 (TARGET_FUNCTION_ARG_PADDING): ...this.
4125 * doc/tm.texi: Regenerate.
4126 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4127 instead of direction.
4128 (compute_argument_addresses): Likewise.
4129 (load_register_parameters): Likewise.
4130 (emit_library_call_value_1): Likewise.
4131 (store_one_arg): Use targetm.calls.function_arg_padding instead
4132 of FUNCTION_ARG_PADDING.
4133 (must_pass_in_stack_var_size_or_pad): Likewise.
4134 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4135 (emit_group_store): Likewise.
4136 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4137 instead of FUNCTION_ARG_PADDING.
4138 (emit_push_insn): Likewise, and propagate enum change throughout
4140 * function.h (direction): Delete.
4141 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4143 * function.c (assign_parm_find_stack_rtl): Likewise.
4144 (assign_parm_setup_block_p): Likewise.
4145 (assign_parm_setup_block): Likewise.
4146 (gimplify_parameters): Likewise.
4147 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4148 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4150 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4151 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4152 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4153 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4154 (aarch64_function_arg_padding): ...this new function.
4155 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4156 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4157 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4158 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4159 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4160 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4161 (arm_pad_arg_upward): Replace with...
4162 (arm_function_arg_padding): ...this new function.
4163 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4165 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4166 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4167 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4168 (ia64_hpux_function_arg_padding): Replace with...
4169 (ia64_function_arg_padding): ...this new function. Use pad_direction
4170 instead of direction. Check for TARGET_HPUX.
4171 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4172 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4173 (iq2000_function_arg_padding): New function.
4174 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4175 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4176 (mips_function_arg_padding): ...this new function.
4177 (mips_pad_reg_upward): Update accordingly.
4178 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4179 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4180 targetm.calls.function_arg_padding.
4181 (FUNCTION_ARG_PADDING): Delete.
4182 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4183 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4184 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4185 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4186 (nios2_block_reg_padding): Return pad_direction instead of direction.
4187 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4188 instead of direction.
4189 (nios2_function_arg_padding): Likewise. Make static.
4190 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4191 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4192 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4193 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4194 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4195 (pa_function_arg_padding): Make static. Return pad_direction instead
4197 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4198 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4199 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4200 instead of direction. Use targetm.calls.function_arg_padding.
4201 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4202 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4203 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4204 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4205 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4207 (function_arg_padding): Rename to...
4208 (rs6000_function_arg_padding): ...this. Make static. Return
4209 pad_direction instead of direction.
4210 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4211 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4212 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4213 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4214 instead of direction. Use targetm.calls.function_arg_padding.
4215 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4216 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4217 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4218 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4219 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4220 (function_arg_padding): Rename to...
4221 (rs6000_function_arg_padding): ...this. Make static. Return
4222 pad_direction instead of direction.
4223 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4224 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4225 * config/s390/s390.c (s390_function_arg_padding): New function.
4226 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4227 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4228 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4229 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4230 (function_arg_padding): Rename to...
4231 (sparc_function_arg_padding): ...this. Make static. Return
4232 pad_direction instead of direction.
4233 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4234 * config/spu/spu.c (spu_function_arg_padding): New function.
4235 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4236 * system.h (FUNCTION_ARG_PADDING): Poison.
4238 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4239 Alan Hayward <alan.hayward@arm.com>
4240 David Sherwood <david.sherwood@arm.com>
4242 * target.def (modes_tieable_p): New hook.
4243 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4244 (TARGET_MODES_TIEABLE_P): ...this.
4245 * doc/tm.texi.in: Regenerate.
4246 * hooks.h (hook_bool_mode_mode_true): Declare.
4247 * hooks.c (hook_bool_mode_mode_true): New function.
4248 * combine.c (subst): Use targetm.modes_tieable_p instead of
4250 * dse.c (find_shift_sequence): Likewise.
4251 * expmed.c (extract_low_bits): Likewise.
4252 * lower-subreg.c: Include target.h.
4253 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4255 * rtlanal.c (rtx_cost): Likewise.
4256 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4257 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4258 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4259 (TARGET_MODES_TIEABLE_P): Redefine.
4260 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4261 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4262 (TARGET_MODES_TIEABLE_P): Redefine.
4263 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4264 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4265 (arc_modes_tieable_p): New function.
4266 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4267 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4268 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4269 (arm_modes_tieable_p): Make static.
4270 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4271 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4272 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4273 (TARGET_MODES_TIEABLE_P): Redefine.
4274 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4275 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4276 (TARGET_MODES_TIEABLE_P): Redefine.
4277 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4278 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4279 (cr16_modes_tieable_p): New function.
4280 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4281 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4282 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4283 (TRULY_NOOP_TRUNCATION): Update comment.
4284 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4285 (TRULY_NOOP_TRUNCATION): Update comment.
4286 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4287 (frv_modes_tieable_p): New function.
4288 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4289 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4290 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4291 (TARGET_MODES_TIEABLE_P): Redefine.
4292 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4293 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4294 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4295 (TARGET_MODES_TIEABLE_P): Redefine.
4296 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4297 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4298 (ia64_modes_tieable_p): New function.
4299 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4300 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4301 (iq2000_modes_tieable_p): New function.
4302 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4303 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4304 (lm32_modes_tieable_p): New function.
4305 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4306 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4307 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4308 (TARGET_MODES_TIEABLE_P): Redefine.
4309 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4310 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4311 (m32r_modes_tieable_p): New function.
4312 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4313 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4314 (m68k_modes_tieable_p): New function.
4315 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4316 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4317 (mcore_modes_tieable_p): New function.
4318 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4319 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4321 (TARGET_MODES_TIEABLE_P): Redefine.
4322 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4323 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4324 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4325 (TARGET_MODES_TIEABLE_P): Redefine.
4326 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4327 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4328 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4329 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4330 (mn10300_modes_tieable_p): ...this and make static.
4331 (TARGET_MODES_TIEABLE_P): Redefine.
4332 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4333 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4334 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4335 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4336 (msp430_modes_tieable_p): Make static.
4337 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4338 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4339 (TARGET_MODES_TIEABLE_P): Redefine.
4340 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4341 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4342 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4343 (TARGET_MODES_TIEABLE_P): Redefine.
4344 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4345 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4346 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4347 (TARGET_MODES_TIEABLE_P): Redefine.
4348 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4349 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4350 (pdp11_modes_tieable_p): New function.
4351 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4352 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4353 (rs6000_modes_tieable_p): New function.
4354 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4355 * config/powerpcspe/powerpcspe.md: Update comment.
4356 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4357 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4358 (TARGET_MODES_TIEABLE_P): Redefine.
4359 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4360 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4361 (rl78_modes_tieable_p): New function.
4362 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4363 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4364 (rs6000_modes_tieable_p): New function.
4365 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4366 * config/rs6000/rs6000.md: Update comment.
4367 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4368 * config/rx/rx.c (rx_modes_tieable_p): New function.
4369 (TARGET_MODES_TIEABLE_P): Redefine.
4370 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4371 * config/s390/s390.c (s390_modes_tieable_p): New function.
4372 (TARGET_MODES_TIEABLE_P): Redefine.
4373 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4374 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4375 (sh_modes_tieable_p): New function.
4376 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4377 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4378 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4379 (sparc_modes_tieable_p): Make static.
4380 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4381 * config/spu/spu.c (spu_modes_tieable_p): New function.
4382 (TARGET_MODES_TIEABLE_P): Redefine.
4383 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4384 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4385 (TARGET_MODES_TIEABLE_P): Redefine.
4386 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4387 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4388 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4389 * config/v850/v850.c (v850_modes_tieable_p): New function.
4390 (TARGET_MODES_TIEABLE_P): Redefine.
4391 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4392 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4393 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4394 (visium_modes_tieable_p): New function.
4395 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4396 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4397 (xtensa_modes_tieable_p): New function.
4398 * system.h (MODES_TIEABLE_P): Poison.
4400 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4401 Alan Hayward <alan.hayward@arm.com>
4402 David Sherwood <david.sherwood@arm.com>
4404 * target.def (hard_regno_mode_ok): New hook.
4405 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4406 (TARGET_HARD_REGNO_MODE_OK): ...this.
4407 * doc/tm.texi.in: Regenerate.
4408 * hooks.h (hook_bool_uint_mode_true): Declare.
4409 * hooks.c (hook_bool_uint_mode_true): New function.
4410 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4412 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4413 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4414 instead of HARD_REGNO_MODE_OK.
4415 * caller-save.c: Include target.h.
4416 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4418 * combine.c (can_combine_p): Likewise.
4419 (combinable_i3pat): Likewise.
4420 (can_change_dest_mode): Likewise.
4421 * expr.c (init_expr_target): Likewise.
4422 (convert_move): Likewise.
4423 (convert_modes): Likewise.
4424 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4425 (setup_prohibited_mode_move_regs): Likewise.
4426 * ira.h (target_ira): Likewise.
4427 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4428 * lra-constraints.c (process_alt_operands): Likewise.
4429 (split_reg): Likewise.
4430 * recog.c (peep2_find_free_register): Likewise.
4431 * ree.c (combine_reaching_defs): Likewise.
4432 * regcprop.c (maybe_mode_change): Likewise.
4433 * reginfo.c (init_reg_sets_1): Likewise.
4434 (choose_hard_reg_mode): Likewise.
4435 (simplifiable_subregs): Likewise.
4436 * regrename.c (check_new_reg_p): Likewise.
4437 * reload.c (find_valid_class): Likewise.
4438 (find_valid_class_1): Likewise.
4439 (reload_inner_reg_of_subreg): Likewise.
4440 (push_reload): Likewise.
4441 (combine_reloads): Likewise.
4442 (find_dummy_reload): Likewise.
4443 (find_reloads): Likewise.
4444 * reload1.c (find_reg): Likewise.
4445 (set_reload_reg): Likewise.
4446 (allocate_reload_reg): Likewise.
4447 (choose_reload_regs): Likewise.
4448 (reload_adjust_reg_for_temp): Likewise.
4449 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4450 (simplify_subreg_regno): Likewise.
4451 * sel-sched.c (init_regs_for_mode): Likewise.
4452 * varasm.c (make_decl_rtl): Likewise.
4453 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4454 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4456 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4457 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4459 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4460 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4461 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4462 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4463 (arc_mode_class): Delete.
4464 (HARD_REGNO_MODE_OK): Delete.
4465 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4466 (arc_hard_regno_mode_ok): Rename old array to...
4467 (arc_hard_regno_mode_ok_modes): ...this.
4468 (arc_conditional_register_usage): Update accordingly.
4469 (arc_mode_class): Make static.
4470 (arc_hard_regno_mode_ok): New function.
4471 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4472 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4473 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4474 (arm_hard_regno_mode_ok): Make static.
4475 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4477 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4478 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4479 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4481 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4482 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4483 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4484 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4485 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4486 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4487 * config/bfin/predicates.md (valid_reg_operand): Use
4488 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4489 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4490 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4492 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4493 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4494 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4495 (cr16_hard_regno_mode_ok): Make static and return a bool.
4496 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4497 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4498 (cris_hard_regno_mode_ok): New function.
4499 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4500 (epiphany_mode_class): Delete.
4501 (HARD_REGNO_MODE_OK): Delete.
4502 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4503 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4504 (hard_regno_mode_ok): Rename to...
4505 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4506 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4507 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4509 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4510 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4511 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4512 (frv_hard_regno_mode_ok): Make static and return a bool.
4513 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4515 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4516 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4517 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4518 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4520 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4521 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4522 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4523 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4526 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4527 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4528 (ia64_hard_regno_mode_ok): New function.
4529 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4530 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4531 (iq2000_hard_regno_mode_ok): New function.
4532 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4533 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4534 (lm32_hard_regno_mode_ok): New function.
4535 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4536 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4537 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4538 instead of HARD_REGNO_MODE_OK.
4539 (m32c_hard_regno_ok): Rename to...
4540 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4541 (m32c_cannot_change_mode_class): Update accordingly.
4542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4543 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4544 (m32r_mode_class): Delete.
4545 (HARD_REGNO_MODE_OK): Delete.
4546 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4547 (m32r_hard_regno_mode_ok): Rename to...
4548 (m32r_hard_regno_modes): ...this.
4549 (m32r_mode_class): Make static.
4550 (m32r_hard_regno_mode_ok): New function.
4551 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4552 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4553 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4554 (m68k_hard_regno_mode_ok): Make static.
4555 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4556 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4557 (mcore_hard_regno_mode_ok): New function.
4558 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4559 (HARD_REGNO_MODE_OK): Delete.
4560 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4562 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4563 (microblaze_hard_regno_mode_ok): New function.
4564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4565 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4566 (mips_hard_regno_mode_ok): Delete.
4567 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4568 (mips_hard_regno_mode_ok_p): ...this and make static.
4569 (mips_hard_regno_mode_ok_p): Rename to...
4570 (mips_hard_regno_mode_ok_uncached): ...this.
4571 (mips_hard_regno_mode_ok): New function.
4572 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4573 of HARD_REGNO_MODE_OK.
4574 (mips_option_override): Update after above name changes.
4575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4576 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4577 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4578 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4579 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4581 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4582 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4583 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4584 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4585 (msp430_hard_regno_mode_ok): Make static and return a bool.
4586 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4587 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4588 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4590 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4591 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4592 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4593 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4594 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4595 (PA_HARD_REGNO_MODE_OK): ...this
4596 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4597 (PA_HARD_REGNO_MODE_OK): ...this.
4598 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4599 (pa_hard_regno_mode_ok): New function.
4600 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4601 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4602 (pdp11_hard_regno_mode_ok): New function.
4603 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4604 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4606 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4608 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4609 (rs6000_hard_regno_mode_ok): Rename to...
4610 (rs6000_hard_regno_mode_ok_uncached): ...this.
4611 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4612 (rs6000_hard_regno_mode_ok): New function.
4613 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4614 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4615 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4616 (riscv_hard_regno_mode_ok): ...this and make static.
4617 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4618 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4619 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4620 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4621 (rl78_hard_regno_mode_ok): Make static and return bool.
4622 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4623 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4625 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4626 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4627 (rs6000_hard_regno_mode_ok): Rename to...
4628 (rs6000_hard_regno_mode_ok_uncached): ...this.
4629 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4630 (rs6000_hard_regno_mode_ok): New function.
4631 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4632 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4633 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4634 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4635 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4636 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4637 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4638 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4639 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4640 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4641 (sh_hard_regno_mode_ok): Make static.
4642 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4643 instead of HARD_REGNO_MODE_OK.
4644 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4645 (sparc_mode_class): Delete.
4646 (HARD_REGNO_MODE_OK): Delete.
4647 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4648 (hard_regno_mode_classes): Make static.
4649 (sparc_mode_class): Likewise.
4650 (sparc_hard_regno_mode_ok): New function.
4651 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4652 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4653 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4655 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4656 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4657 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4658 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4659 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4661 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4662 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4663 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4664 (visium_hard_regno_mode_ok): New function.
4665 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4666 instead of HARD_REGNO_MODE_OK.
4667 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4668 (HARD_REGNO_MODE_OK): Delete.
4669 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4670 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4671 (xtensa_option_override): Update accordingly.
4672 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4673 (xtensa_hard_regno_mode_ok): New function.
4674 * system.h (HARD_REGNO_MODE_OK): Poison.
4676 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4677 Alan Hayward <alan.hayward@arm.com>
4678 David Sherwood <david.sherwood@arm.com>
4680 * target.def (hard_regno_call_part_clobbered): New hook.
4681 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4682 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4683 * doc/tm.texi: Regenerate.
4684 * hooks.h (hook_bool_uint_mode_false): Declare.
4685 * hooks.c (hook_bool_uint_mode_false): New function.
4686 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4687 * cselib.c (cselib_process_insn): Use
4688 targetm.hard_regno_call_part_clobbered instead of
4689 HARD_REGNO_CALL_PART_CLOBBERED.
4690 * ira-conflicts.c (ira_build_conflicts): Likewise.
4691 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4692 * lra-constraints.c (need_for_call_save_p): Likewise.
4693 * lra-lives.c: Include target.h.
4694 (check_pseudos_live_through_calls): Use
4695 targetm.hard_regno_call_part_clobbered instead of
4696 HARD_REGNO_CALL_PART_CLOBBERED.
4697 * regcprop.c: Include target.h.
4698 (copyprop_hardreg_forward_1): Use
4699 targetm.hard_regno_call_part_clobbered instead of
4700 HARD_REGNO_CALL_PART_CLOBBERED.
4701 * reginfo.c (choose_hard_reg_mode): Likewise.
4702 * regrename.c (check_new_reg_p): Likewise.
4703 * reload.c (find_equiv_reg): Likewise.
4704 * reload1.c (emit_reload_insns): Likewise.
4705 * sched-deps.c (deps_analyze_insn): Likewise.
4706 * sel-sched.c (init_regs_for_mode): Likewise.
4707 (mark_unavailable_hard_regs): Likewise.
4708 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4709 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4710 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4712 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4713 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4714 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4716 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4718 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4719 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4720 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4722 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4723 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4724 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4726 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4727 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4729 * config/powerpcspe/powerpcspe.c
4730 (rs6000_hard_regno_call_part_clobbered): New function.
4731 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4732 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4733 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4735 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4736 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4737 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4739 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4740 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4741 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4743 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4744 Alan Hayward <alan.hayward@arm.com>
4745 David Sherwood <david.sherwood@arm.com>
4747 * rtl.h (subreg_memory_offset): Declare.
4748 * emit-rtl.c (subreg_memory_offset): New function.
4749 * expmed.c (store_bit_field_1): Use it.
4750 * expr.c (undefined_operand_subword_p): Likewise.
4751 * simplify-rtx.c (simplify_subreg): Likewise.
4753 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4755 PR rtl-optimization/57448
4758 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4759 using atomic_load pattern.
4760 (expand_atomic_store): Likewise.
4762 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4765 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4766 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4767 handling. Use replace_call_with_value with NULL instead of
4768 gsi_replace, unlink_stmt_vdef and release_defs.
4770 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4773 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4775 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4778 * rtl.h (emit_library_call_value_1): Declare.
4779 (emit_library_call): Replace declaration with a series of overloads.
4780 Remove the parameter count argument.
4781 (emit_library_call_value): Likewise.
4782 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4783 with an "rtx_mode_t *".
4784 (emit_library_call_value): Delete.
4785 (emit_library_call): Likewise.
4786 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4787 (asan_emit_allocas_unpoison): Likewise.
4788 * builtins.c (expand_builtin_powi): Likewise.
4789 (expand_asan_emit_allocas_unpoison): Likewise.
4790 * cfgexpand.c (expand_main_function): Likewise.
4791 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4792 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4793 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4794 * config/arm/arm.c (arm_trampoline_init): Likewise.
4795 (arm_call_tls_get_addr): Likewise.
4796 (arm_expand_divmod_libfunc): Likewise.
4797 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4798 (smulsi3_highpart): Likewise.
4799 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4800 (c6x_expand_compare): Likewise.
4801 (c6x_expand_movmem): Likewise.
4802 * config/frv/frv.c (frv_trampoline_init): Likewise.
4803 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4804 (ix86_expand_divmod_libfunc): Likewise.
4805 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4806 (ia64_expand_compare): Likewise.
4807 (ia64_profile_hook): Likewise.
4808 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4809 (nonlocal_goto): Likewise.
4810 (restore_stack_nonlocal): Likewise.
4811 * config/m32r/m32r.c (block_move_call): Likewise.
4812 (m32r_trampoline_init): Likewise.
4813 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4814 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4815 (m68k_call_m68k_read_tp): Likewise.
4816 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4817 (microblaze_expand_divide): Likewise.
4818 * config/mips/mips.h (mips_args): Likewise.
4819 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4820 (MIPS_ICACHE_SYNC): Likewise.
4821 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4822 (nios2_trampoline_init): Likewise.
4823 * config/pa/pa.c (hppa_tls_call): Likewise.
4824 (pa_trampoline_init): Likewise.
4825 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4826 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4827 (expand_strn_compare): Likewise.
4828 (rs6000_generate_compare): Likewise.
4829 (rs6000_expand_float128_convert): Likewise.
4830 (output_profile_hook): Likewise.
4831 (rs6000_trampoline_init): Likewise.
4832 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4833 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4834 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4835 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4836 (rs6000_generate_compare): Likewise.
4837 (rs6000_expand_float128_convert): Likewise.
4838 (output_profile_hook): Likewise.
4839 (rs6000_trampoline_init): Likewise.
4840 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4841 * config/sh/sh.c (sh_trampoline_init): Likewise.
4842 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4843 (sparc_emit_float_lib_cmp): Likewise.
4844 (sparc32_initialize_trampoline): Likewise.
4845 (sparc64_initialize_trampoline): Likewise.
4846 (sparc_profile_hook): Likewise.
4847 * config/spu/spu.c (ea_load_store): Likewise.
4848 * config/spu/spu.md (floatunssidf2): Likewise.
4849 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4850 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4851 * config/visium/visium.c (expand_block_move_4): Likewise.
4852 (expand_block_move_2): Likewise.
4853 (expand_block_move_1): Likewise.
4854 (expand_block_set_4): Likewise.
4855 (expand_block_set_2): Likewise.
4856 (expand_block_set_1): Likewise.
4857 (visium_trampoline_init): Likewise.
4858 (visium_profile_hook): Likewise.
4859 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4860 (xtensa_setup_frame_addresses): Likewise.
4861 (xtensa_trampoline_init): Likewise.
4862 * except.c (sjlj_emit_function_enter): Likewise.
4863 (sjlj_emit_function_exit): Likewise.
4864 * explow.c (allocate_dynamic_stack_space): Likewise.
4865 (probe_stack_range): Likewise.
4866 * expr.c (convert_mode_scalar): Likewise.
4867 * optabs.c (expand_binop): Likewise.
4868 (expand_twoval_binop_libfunc): Likewise.
4869 (expand_unop): Likewise.
4870 (prepare_cmp_insn): Likewise.
4871 (prepare_float_lib_cmp): Likewise.
4872 (expand_float): Likewise.
4873 (expand_fix): Likewise.
4874 (expand_fixed_convert): Likewise.
4875 (maybe_emit_sync_lock_test_and_set): Likewise.
4876 (expand_atomic_compare_and_swap): Likewise.
4877 (expand_mem_thread_fence): Likewise.
4878 (expand_atomic_fetch_op): Likewise.
4880 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4882 * doc/generic.texi (OpenACC): Adjust URL.
4883 * doc/invoke.texi (C Dialect Options): Ditto.
4885 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4887 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4888 predicate for operand 1. Add (m,<S>) constraint.
4889 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4890 Prevent memory operand 1 with register operand 2.
4892 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4894 PR rtl-optimization/82024
4895 * combine.c (try_combine): If the combination result is a PARALLEL,
4896 and we only need to retain the SET in there that would be placed
4897 at I2, check that we can place that at I3 instead, before doing so.
4899 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4902 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4904 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4905 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4908 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4909 Jeff Law <law@redhat.com>
4911 * varasm.c (bss_initializer_p): Do not put constants into .bss
4912 (categorize_decl_for_section): Handle bss_initializer_p returning
4913 false when DECL_INITIAL is NULL.
4915 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4918 * config/s390/s390.c (s390_can_inline_p): New function.
4920 2017-09-01 Jeff Law <law@redhat.com>
4922 PR tree-optimization/82052
4923 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4924 Always initialize the returned slot after a hash table miss
4925 when INSERT is true.
4927 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4929 * config/s390/s390.md (mem_signal_fence): Remove.
4930 * doc/md.texi (mem_signal_fence): Remove.
4931 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4933 * target-insns.def (mem_signal_fence): Remove.
4935 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4938 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4941 * asan.c (create_odr_indicator): Strip name encoding from assembler
4942 name before appending it after __odr_asan_.
4944 2017-09-01 Martin Liska <mliska@suse.cz>
4946 PR tree-optimization/82059
4947 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4948 frequency only when an edge is redirected.
4950 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4952 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4953 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4954 (arc_conditional_register_usage): Remove ARC600 lp_count
4956 (arc_file_start): Emit Tag_ARC_CPU_variation.
4957 (arc_can_use_doloop_p): New conditions to use ZOLs.
4958 (hwloop_fail): New function.
4959 (hwloop_optimize): Likewise.
4960 (hwloop_pattern_reg): Likewise.
4961 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4962 (arc_reorg_loops): New function, calls reorg_loops.
4963 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4964 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4966 (arc_loop_hazard): Remove function, functionality moved into
4968 (arc_hazard): Remove arc_loop_hazard call.
4969 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4970 into hwloop_optimize.
4971 (arc_label_align): Remove ZOL handling.
4972 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4973 * config/arc/arc.md (doloop_begin): Remove pattern.
4974 (doloop_begin_i): Likewise.
4975 (doloop_end_i): Likewise.
4976 (doloop_fallback): Likewise.
4977 (doloop_fallback_m): Likewise.
4978 (doloop_end): Reimplement expand.
4979 (arc_lp): New pattern for LP instruction.
4980 (loop_end): New pattern.
4981 (loop_fail): Likewise.
4982 (decrement_and_branch_until_zero): Likewise.
4983 * config/arc/arc.opt (mlpc-width): New option.
4984 * doc/invoke.texi (mlpc-width): Document option.
4986 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4988 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4989 (arc_ccfsm_advance): Fix checking for delay slots.
4990 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4992 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4994 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4996 (movhi_insn): Update store instruction constraint which are saving
4997 6-bit short immediates.
4998 (movsi_insn): Consider also short scaled load operations.
4999 (zero_extendhisi2_i): Use Usd constraint instead of T.
5000 (extendhisi2_i): Add q constraint.
5001 (arc_clzsi2): Add type and length attributes.
5002 (arc_ctzsi2): Likewise.
5003 * config/arc/constraints.md (Usc): Update constraint, the
5004 assembler can parse two relocations for a single instruction.
5006 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5008 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5009 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5011 2017-08-31 Olivier Hainque <hainque@adacore.com>
5013 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5014 match as powerpc-wrs-vxworks*.
5016 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5018 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5019 register constraint for by-element operand.
5020 (aarch64_mls_elt_merge<mode>): Likewise.
5022 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5024 * config/arc/arc.c (arc_can_follow_jump): Check for short
5027 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5029 * config.gcc: Use g.opt for arc.
5030 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5031 functionality moved to ...
5032 (legitimate_scaled_address_p): New function, ...here.
5033 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5034 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5035 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5037 (arc_override_options): Handle G option.
5038 (arc_output_pic_addr_const): Correct function definition.
5039 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5040 (arc_decl_anon_ns_mem_p): Delete.
5041 (arc_in_small_data_p): Overhaul this function to take into
5042 consideration the value given via G option.
5043 (arc_rewrite_small_data_1): Renamed and corrected old
5044 arc_rewrite_small_data function.
5045 (arc_rewrite_small_data): New function.
5046 (small_data_pattern): Don't use pic_offset_table_rtx.
5047 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5048 * config/arc/simdext.md (movmisalignv2hi): Use
5049 prepare_move_operands function.
5051 (movmisalign*): Likewise.
5052 * doc/invoke.texi (ARC options): Document -G option.
5054 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5056 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5058 * config/arc/arc.c (arc_print_operand): Output scalled address for
5059 sdata whenever is possible.
5060 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5061 load/stores are available.
5062 (compact_sda_memory_operand): Check for the alignment required by
5063 code density instructions.
5064 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5066 * config/arc/constraints.md (Usd): Update constraint.
5067 (Us0): New constraint.
5068 (Usc): Update constraint.
5070 2017-08-31 Richard Biener <rguenther@suse.de>
5073 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5076 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5078 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5079 Resize type_signature.
5081 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5082 Alan Hayward <alan.hayward@arm.com>
5083 David Sherwood <david.sherwood@arm.com>
5085 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5086 subregs whose inner modes can be stored in GPRs.
5087 (aarch64_classify_index): Likewise.
5089 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5090 Alan Hayward <alan.hayward@arm.com>
5091 David Sherwood <david.sherwood@arm.com>
5093 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5094 (V_INT_EQUIV): ...this.
5095 (v_cmp_result): Rename to...
5096 (v_int_equiv): ...this.
5097 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5098 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5099 (copysign<mode>3): Likewise.
5100 (aarch64_simd_bsl<mode>_internal): Likewise.
5101 (aarch64_simd_bsl<mode>): Likewise.
5102 (vec_cmp<mode><mode>): Likewise.
5103 (vcond<mode><mode>): Likewise.
5104 (vcond<v_cmp_mixed><mode>): Likewise.
5105 (vcondu<mode><v_cmp_mixed>): Likewise.
5106 (aarch64_cm<optab><mode>): Likewise.
5107 (aarch64_cmtst<mode>): Likewise.
5108 (aarch64_fac<optab><mode>): Likewise.
5109 (vec_perm_const<mode>): Likewise.
5110 (vcond_mask_<mode><v_cmp_result>): Rename to...
5111 (vcond_mask_<mode><v_int_equiv>): ...this.
5112 (vec_cmp<mode><v_cmp_result>): Rename to...
5113 (vec_cmp<mode><v_int_equiv>): ...this.
5115 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5116 Alan Hayward <alan.hayward@arm.com>
5117 David Sherwood <david.sherwood@arm.com>
5119 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5121 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5122 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5123 (UNSPEC_LD4_DREG): New unspecs.
5124 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5125 (aarch64_ld2<mode>_dreg_be): Replace with...
5126 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5128 (aarch64_ld3<mode>_dreg_le)
5129 (aarch64_ld3<mode>_dreg_be): Replace with...
5130 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5132 (aarch64_ld4<mode>_dreg_le)
5133 (aarch64_ld4<mode>_dreg_be): Replace with...
5134 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5137 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5139 PR tree-optimization/81987
5140 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5141 insert an initializer in a location not dominated by the stride
5144 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5146 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5147 on the entire header of the finally block in the fallthru case.
5149 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5151 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5153 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5155 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5156 rs6000_emit_move_from_cr and call renamed function.
5157 (rs6000_emit_prologue): Call renamed functions.
5158 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5159 movesi_from_cr, remove volatile CRs.
5161 2017-08-30 Jon Beniston <jon@beniston.com>
5162 Richard Biener <rguenther@suse.de>
5164 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5165 of VECTOR_MODE_P check.
5166 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5167 element vector types.
5169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5171 * df.h (df_read_modify_subreg_p): Remove in favor of...
5172 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5173 const_rtx instead of an rtx.
5174 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5175 * df-problems.c (df_word_lr_mark_ref): Likewise.
5176 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5177 (mark_pseudo_reg_dead): Likewise.
5178 (mark_ref_dead): Likewise.
5179 * reginfo.c (init_subregs_of_mode): Likewise.
5180 * sched-deps.c (sched_analyze_1): Likewise.
5181 * df-scan.c (df_def_record_1): Likewise.
5182 (df_uses_record): Likewise.
5183 (df_read_modify_subreg_p): Remove in favor of...
5184 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5185 const_rtx instead of an rtx.
5187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5188 Alan Hayward <alan.hayward@arm.com>
5189 David Sherwood <david.sherwood@arm.com>
5191 * rtl.h (partial_subreg_p): New function.
5192 * caller-save.c (save_call_clobbered_regs): Use it.
5193 * calls.c (expand_call): Likewise.
5194 * combine.c (combinable_i3pat): Likewise.
5195 (simplify_set): Likewise.
5196 (make_extraction): Likewise.
5197 (make_compound_operation_int): Likewise.
5198 (gen_lowpart_or_truncate): Likewise.
5199 (force_to_mode): Likewise.
5200 (make_field_assignment): Likewise.
5201 (reg_truncated_to_mode): Likewise.
5202 (record_truncated_value): Likewise.
5203 (move_deaths): Likewise.
5204 * cse.c (record_jump_cond): Likewise.
5205 (cse_insn): Likewise.
5206 * cselib.c (cselib_lookup_1): Likewise.
5207 * expmed.c (extract_bit_field_using_extv): Likewise.
5208 * function.c (assign_parm_setup_reg): Likewise.
5209 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5210 * ira-build.c (create_insn_allocnos): Likewise.
5211 * lra-coalesce.c (merge_pseudos): Likewise.
5212 * lra-constraints.c (match_reload): Likewise.
5213 (simplify_operand_subreg): Likewise.
5214 (curr_insn_transform): Likewise.
5215 * lra-lives.c (process_bb_lives): Likewise.
5216 * lra.c (new_insn_reg): Likewise.
5217 (lra_substitute_pseudo): Likewise.
5218 * regcprop.c (mode_change_ok): Likewise.
5219 (maybe_mode_change): Likewise.
5220 (copyprop_hardreg_forward_1): Likewise.
5221 * reload.c (push_reload): Likewise.
5222 (find_reloads): Likewise.
5223 (find_reloads_subreg_address): Likewise.
5224 * reload1.c (alter_reg): Likewise.
5225 (eliminate_regs_1): Likewise.
5226 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5228 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5230 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5231 back to if statements, including unpack.
5233 2017-08-30 Martin Liska <mliska@suse.cz>
5236 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5238 (func_checker::compare_asm_inputs_outputs): ... this function.
5239 (func_checker::compare_gimple_asm): Use the function to compare
5240 also ASM constrains.
5241 * ipa-icf-gimple.h: Rename the function.
5243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5244 Alan Hayward <alan.hayward@arm.com>
5245 David Sherwood <david.sherwood@arm.com>
5247 * coretypes.h (complex_mode): New type.
5248 * gdbhooks.py (build_pretty_printer): Handle it.
5249 * machmode.h (complex_mode): New class.
5250 (complex_mode::includes_p): New function.
5251 (is_complex_int_mode): Likewise.
5252 (is_complex_float_mode): Likewise.
5253 * genmodes.c (get_mode_class): Handle complex mode classes.
5254 * function.c (expand_function_end): Use is_complex_int_mode.
5256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5257 Alan Hayward <alan.hayward@arm.com>
5258 David Sherwood <david.sherwood@arm.com>
5260 * coretypes.h (scalar_mode_pod): New typedef.
5261 * gdbhooks.py (build_pretty_printer): Handle it.
5262 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5263 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5264 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5265 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5269 Alan Hayward <alan.hayward@arm.com>
5270 David Sherwood <david.sherwood@arm.com>
5272 * machmode.h (mode_for_vector): Take a scalar_mode instead
5274 * stor-layout.c (mode_for_vector): Likewise.
5275 * explow.c (promote_mode): Use as_a <scalar_mode>.
5276 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5279 Alan Hayward <alan.hayward@arm.com>
5280 David Sherwood <david.sherwood@arm.com>
5282 * target.def (preferred_simd_mode): Take a scalar_mode
5283 instead of a machine_mode.
5284 * targhooks.h (default_preferred_simd_mode): Likewise.
5285 * targhooks.c (default_preferred_simd_mode): Likewise.
5286 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5287 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5288 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5289 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5290 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5291 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5292 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5293 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5295 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5296 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5297 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5298 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5299 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5300 * doc/tm.texi: Regenerate.
5301 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5305 Alan Hayward <alan.hayward@arm.com>
5306 David Sherwood <david.sherwood@arm.com>
5308 * target.def (scalar_mode_supported_p): Take a scalar_mode
5309 instead of a machine_mode.
5310 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5311 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5312 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5313 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5314 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5315 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5316 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5317 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5318 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5319 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5320 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5321 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5322 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5323 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5325 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5326 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5327 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5328 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5329 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5331 * doc/tm.texi: Regenerate.
5333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5334 Alan Hayward <alan.hayward@arm.com>
5335 David Sherwood <david.sherwood@arm.com>
5337 * coretypes.h (opt_scalar_mode): New typedef.
5338 * gdbhooks.py (build_pretty_printers): Handle it.
5339 * machmode.h (mode_iterator::get_2xwider): Add overload for
5341 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5343 * expr.c (convert_mode_scalar): Likewise.
5344 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5345 * optabs.c (expand_float): Likewise.
5346 (expand_fix): Likewise.
5347 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5350 Alan Hayward <alan.hayward@arm.com>
5351 David Sherwood <david.sherwood@arm.com>
5353 * optabs.c (expand_float): Explicitly check for scalars before
5354 using a branching expansion.
5355 (expand_fix): Likewise.
5357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5358 Alan Hayward <alan.hayward@arm.com>
5359 David Sherwood <david.sherwood@arm.com>
5361 * expr.c (convert_mode): Split scalar handling out into...
5362 (convert_mode_scalar): ...this new function. Treat the modes
5365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5366 Alan Hayward <alan.hayward@arm.com>
5367 David Sherwood <david.sherwood@arm.com>
5369 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5371 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5374 Alan Hayward <alan.hayward@arm.com>
5375 David Sherwood <david.sherwood@arm.com>
5377 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5378 rather than a machine_mode.
5379 (fixed_from_string): Likewise.
5380 (fixed_convert): Likewise.
5381 (fixed_convert_from_int): Likewise.
5382 (fixed_convert_from_real): Likewise.
5383 (real_convert_from_fixed): Likewise.
5384 * fixed-value.c (fixed_from_double_int): Likewise.
5385 (fixed_from_string): Likewise.
5386 (fixed_convert): Likewise.
5387 (fixed_convert_from_int): Likewise.
5388 (fixed_convert_from_real): Likewise.
5389 (real_convert_from_fixed): Likewise.
5390 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5393 Alan Hayward <alan.hayward@arm.com>
5394 David Sherwood <david.sherwood@arm.com>
5396 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5397 of separate mode class checks. Do not allow vector modes here.
5398 (immed_wide_int_const): Use as_a <scalar_mode>.
5399 * explow.c (trunc_int_for_mode): Likewise.
5400 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5401 (wi::shwi): Likewise.
5402 (wi::min_value): Likewise.
5403 (wi::max_value): Likewise.
5404 * dwarf2out.c (loc_descriptor): Likewise.
5405 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5409 Alan Hayward <alan.hayward@arm.com>
5410 David Sherwood <david.sherwood@arm.com>
5412 * tree.h (SCALAR_TYPE_MODE): New macro.
5413 * expr.c (expand_expr_addr_expr_1): Use it.
5414 (expand_expr_real_2): Likewise.
5415 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5416 (fold_convert_const_fixed_from_int): Likewise.
5417 (fold_convert_const_fixed_from_real): Likewise.
5418 (native_encode_fixed): Likewise
5419 (native_encode_complex): Likewise
5420 (native_encode_vector): Likewise.
5421 (native_interpret_fixed): Likewise.
5422 (native_interpret_real): Likewise.
5423 (native_interpret_complex): Likewise.
5424 (native_interpret_vector): Likewise.
5425 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5426 (simd_clone_adjust_argument_types): Likewise.
5427 (simd_clone_init_simd_arrays): Likewise.
5428 (simd_clone_adjust): Likewise.
5429 * stor-layout.c (layout_type): Likewise.
5430 * tree.c (build_minus_one_cst): Likewise.
5431 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5432 * tree-inline.c (estimate_move_cost): Likewise.
5433 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5434 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5435 (vectorizable_reduction): Likewise.
5436 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5437 (vect_recog_mixed_size_cond_pattern): Likewise.
5438 (check_bool_pattern): Likewise.
5439 (adjust_bool_pattern): Likewise.
5440 (search_type_for_mask_1): Likewise.
5441 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5442 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5443 (vectorizable_load): Likewise.
5444 (vectorizable_store): Likewise.
5445 * ubsan.c (ubsan_encode_value): Likewise.
5446 * varasm.c (output_constant): Likewise.
5448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5449 Alan Hayward <alan.hayward@arm.com>
5450 David Sherwood <david.sherwood@arm.com>
5452 * coretypes.h (scalar_mode): New class.
5453 * machmode.h (scalar_mode): Likewise.
5454 (scalar_mode::includes_p): New function.
5455 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5456 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5457 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5458 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5459 * expmed.c (store_bit_field_1): Likewise.
5460 (extract_bit_field_1): Likewise.
5461 * expr.c (write_complex_part): Likewise.
5462 (read_complex_part): Likewise.
5463 (emit_move_complex_push): Likewise.
5464 (expand_expr_real_2): Likewise.
5465 * function.c (assign_parm_setup_reg): Likewise.
5466 (assign_parms_unsplit_complex): Likewise.
5467 * optabs.c (expand_binop): Likewise.
5468 * rtlanal.c (subreg_get_info): Likewise.
5469 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5470 * varasm.c (output_constant_pool_2): Likewise.
5472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5473 Alan Hayward <alan.hayward@arm.com>
5474 David Sherwood <david.sherwood@arm.com>
5476 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5478 (expmed_mult_highpart_optab): Likewise.
5479 (expmed_mult_highpart): Likewise.
5481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5482 Alan Hayward <alan.hayward@arm.com>
5483 David Sherwood <david.sherwood@arm.com>
5485 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5486 instead of a machine_mode.
5487 (builtin_memset_read_str): Likewise.
5488 * builtins.c (c_readstr): Likewise.
5489 (builtin_memcpy_read_str): Likewise.
5490 (builtin_strncpy_read_str): Likewise.
5491 (builtin_memset_read_str): Likewise.
5492 (builtin_memset_gen_str): Likewise.
5493 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5494 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5495 instead of a machine_mode.
5496 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5498 (make_extraction): Likewise.
5499 (try_widen_shift_mode): Take and return scalar_int_modes instead
5501 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5502 a scalar_int_mode instead of a machine_mode.
5503 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5504 (avr_addr_space_pointer_mode): Likewise.
5505 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5506 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5507 (msp430_unwind_word_mode): Likewise.
5508 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5509 (spu_addr_space_pointer_mode): Likewise.
5510 (spu_addr_space_address_mode): Likewise.
5511 (spu_libgcc_cmp_return_mode): Likewise.
5512 (spu_libgcc_shift_count_mode): Likewise.
5513 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5514 (rl78_addr_space_pointer_mode): Likewise.
5515 (fl78_unwind_word_mode): Likewise.
5516 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5518 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5519 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5520 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5521 (mips_valid_pointer_mode): Likewise.
5522 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5523 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5524 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5526 (ft32_addr_space_address_mode): Likewise.
5527 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5528 scalar_int_mode instead of a machine_mode.
5529 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5531 (m32c_addr_space_address_mode): Likewise.
5532 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5533 (rs6000_eh_return_filter_mode): Likewise.
5534 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5535 (rs6000_eh_return_filter_mode): Likewise.
5536 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5537 (s390_libgcc_shift_count_mode): Likewise.
5538 (s390_unwind_word_mode): Likewise.
5539 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5541 * target.def (mode_rep_extended): Likewise.
5542 (valid_pointer_mode): Likewise.
5543 (addr_space.valid_pointer_mode): Likewise.
5544 (eh_return_filter_mode): Return a scalar_int_mode rather than
5546 (libgcc_cmp_return_mode): Likewise.
5547 (libgcc_shift_count_mode): Likewise.
5548 (unwind_word_mode): Likewise.
5549 (addr_space.pointer_mode): Likewise.
5550 (addr_space.address_mode): Likewise.
5551 * doc/tm.texi: Regenerate.
5552 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5554 (do_jump): Use scalar_int_mode for local variables.
5555 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5556 rather than a machine_mode.
5557 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5558 (scompare_loc_descriptor_wide): Likewise.
5559 (scompare_loc_descriptor_narrow): Likewise.
5560 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5562 * except.c (sjlj_emit_dispatch_table): Likewise.
5563 (expand_builtin_eh_copy_values): Likewise.
5564 * explow.c (convert_memory_address_addr_space_1): Likewise.
5565 Take a scalar_int_mode rather than a machine_mode.
5566 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5567 than a machine_mode.
5568 (memory_address_addr_space): Use scalar_int_mode for local variables.
5569 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5570 rather than a machine_mode.
5571 * expmed.c (mask_rtx): Likewise.
5572 (init_expmed_one_conv): Likewise.
5573 (expand_mult_highpart_adjust): Likewise.
5574 (extract_high_half): Likewise.
5575 (expmed_mult_highpart_optab): Likewise.
5576 (expmed_mult_highpart): Likewise.
5577 (expand_smod_pow2): Likewise.
5578 (expand_sdiv_pow2): Likewise.
5579 (emit_store_flag_int): Likewise.
5580 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5582 (extract_low_bits): Likewise.
5583 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5585 * expr.c (pieces_addr::adjust): Likewise.
5586 (can_store_by_pieces): Likewise.
5587 (store_by_pieces): Likewise.
5588 (clear_by_pieces_1): Likewise.
5589 (expand_expr_addr_expr_1): Likewise.
5590 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5591 (expand_expr_real_1): Likewise.
5592 (try_casesi): Likewise.
5593 * final.c (shorten_branches): Likewise.
5594 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5595 type of "mode" to machine_mode.
5596 * internal-fn.c (expand_arith_overflow_result_store): Take a
5597 scalar_int_mode rather than a machine_mode.
5598 (expand_mul_overflow): Use scalar_int_mode for local variables.
5599 * loop-doloop.c (doloop_modify): Likewise.
5600 (doloop_optimize): Likewise.
5601 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5602 than a machine_mode.
5603 (expand_doubleword_shift_condmove): Likewise.
5604 (expand_doubleword_shift): Likewise.
5605 (expand_doubleword_clz): Likewise.
5606 (expand_doubleword_popcount): Likewise.
5607 (expand_doubleword_parity): Likewise.
5608 (expand_absneg_bit): Use scalar_int_mode for local variables.
5609 (prepare_float_lib_cmp): Likewise.
5610 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5611 rather than a machine_mode.
5612 (convert_memory_address_addr_space): Likewise.
5613 (get_mode_bounds): Likewise.
5614 (get_address_mode): Return a scalar_int_mode rather than a
5616 * rtlanal.c (get_address_mode): Likewise.
5617 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5618 than a machine_mode.
5619 * targhooks.c (default_mode_rep_extended): Likewise.
5620 (default_valid_pointer_mode): Likewise.
5621 (default_addr_space_valid_pointer_mode): Likewise.
5622 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5623 than a machine_mode.
5624 (default_libgcc_cmp_return_mode): Likewise.
5625 (default_libgcc_shift_count_mode): Likewise.
5626 (default_unwind_word_mode): Likewise.
5627 (default_addr_space_pointer_mode): Likewise.
5628 (default_addr_space_address_mode): Likewise.
5629 * targhooks.h (default_eh_return_filter_mode): Likewise.
5630 (default_libgcc_cmp_return_mode): Likewise.
5631 (default_libgcc_shift_count_mode): Likewise.
5632 (default_unwind_word_mode): Likewise.
5633 (default_addr_space_pointer_mode): Likewise.
5634 (default_addr_space_address_mode): Likewise.
5635 (default_mode_rep_extended): Take a scalar_int_mode rather than
5637 (default_valid_pointer_mode): Likewise.
5638 (default_addr_space_valid_pointer_mode): Likewise.
5639 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5641 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5642 rather than a machine_mode.
5643 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5644 for local variables.
5645 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5646 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5647 than a machine_mode.
5649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5650 Alan Hayward <alan.hayward@arm.com>
5651 David Sherwood <david.sherwood@arm.com>
5653 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5654 the mode argument to scalar_int_mode.
5655 (do_jump_by_parts_zero_rtx): Likewise.
5656 (do_jump_by_parts_equality_rtx): Likewise.
5657 (do_jump_by_parts_greater): Take a mode argument.
5658 (do_jump_by_parts_equality): Likewise.
5659 (do_jump_1): Update calls accordingly.
5661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5662 Alan Hayward <alan.hayward@arm.com>
5663 David Sherwood <david.sherwood@arm.com>
5665 * is-a.h (safe_dyn_cast): New function.
5666 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5667 (jump_table_for_label): Likewise.
5668 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5669 instead of an rtx_insn *.
5670 (shorten_branches): Use dyn_cast instead of LABEL_P and
5671 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5672 rtx_jump_table_data::get_data_mode.
5673 (final_scan_insn): Likewise.
5675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5676 Alan Hayward <alan.hayward@arm.com>
5677 David Sherwood <david.sherwood@arm.com>
5679 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5680 trying to combine a full-register integer set with a subreg
5683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5684 Alan Hayward <alan.hayward@arm.com>
5685 David Sherwood <david.sherwood@arm.com>
5687 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5688 that is always either address_mode or pointer_mode.
5690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5691 Alan Hayward <alan.hayward@arm.com>
5692 David Sherwood <david.sherwood@arm.com>
5694 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5695 when the two are known to be equal.
5697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5698 Alan Hayward <alan.hayward@arm.com>
5699 David Sherwood <david.sherwood@arm.com>
5701 * simplify-rtx.c (simplify_const_unary_operation): Use
5702 is_a <scalar_int_mode> instead of checking for a nonzero
5703 precision. Forcibly convert op_mode to a scalar_int_mode
5704 in that case. More clearly differentiate the operand and
5705 result modes and use the former when deciding what the value
5706 of a count-bits operation should be. Use is_int_mode instead
5707 of checking for a MODE_INT. Remove redundant check for whether
5708 this mode has a zero precision.
5710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5711 Alan Hayward <alan.hayward@arm.com>
5712 David Sherwood <david.sherwood@arm.com>
5714 * optabs.c (widen_leading): Change the type of the mode argument
5715 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5716 (widen_bswap): Likewise.
5717 (expand_parity): Likewise.
5718 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5719 (expand_ffs): Likewise.
5720 (epand_unop): Check for scalar integer modes before calling the
5723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5724 Alan Hayward <alan.hayward@arm.com>
5725 David Sherwood <david.sherwood@arm.com>
5727 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5729 (expand_expr_real_1): Update call accordingly.
5731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5732 Alan Hayward <alan.hayward@arm.com>
5733 David Sherwood <david.sherwood@arm.com>
5735 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5736 value_mode arguments. Use scalar_int_mode internally.
5737 (store_bit_field_1): Rename the new integer mode from imode
5738 to op0_mode and use it instead of GET_MODE (op0). Update calls
5739 to store_split_bit_field, store_bit_field_using_insv and
5740 store_fixed_bit_field.
5741 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5742 Use scalar_int_mode internally. Use a bit count rather than a mode
5743 when calculating the largest bit size for get_best_mode.
5744 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5745 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5746 Remove assertion that OP0 has a scalar integer mode.
5747 (store_split_bit_field): Add op0_mode and value_mode arguments.
5748 Update calls to extract_fixed_bit_field.
5749 (extract_bit_field_using_extv): Add an op0_mode argument.
5750 Use scalar_int_mode internally.
5751 (extract_bit_field_1): Rename the new integer mode from imode to
5752 op0_mode and use it instead of GET_MODE (op0). Update calls to
5753 extract_split_bit_field, extract_bit_field_using_extv and
5754 extract_fixed_bit_field.
5755 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5756 to extract_split_bit_field and extract_fixed_bit_field_1.
5757 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5758 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5760 (extract_split_bit_field): Add an op0_mode argument. Update call
5761 to extract_fixed_bit_field.
5763 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5764 Alan Hayward <alan.hayward@arm.com>
5765 David Sherwood <david.sherwood@arm.com>
5767 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5768 * explow.c (hard_function_value): Likewise.
5769 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5770 convert_to_mode call outside the loop.
5771 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5772 for the mode iterator. Require the mode specified by max_pieces
5774 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5776 (copy_blkmode_to_reg): Likewise.
5777 (set_storage_via_setmem): Likewise.
5778 * optabs.c (prepare_cmp_insn): Likewise.
5779 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5780 * stor-layout.c (finish_bitfield_representative): Likewise.
5782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5783 Alan Hayward <alan.hayward@arm.com>
5784 David Sherwood <david.sherwood@arm.com>
5786 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5787 * expr.c (convert_move): Use them.
5788 (convert_modes): Likewise.
5789 (store_expr_with_bounds): Likewise.
5791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5792 Alan Hayward <alan.hayward@arm.com>
5793 David Sherwood <david.sherwood@arm.com>
5795 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5796 parameter for the mode of "x". Remove the "known_x", "known_mode"
5797 and "known_ret" arguments. Change the type of the mode argument
5799 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5800 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5801 (reg_num_sign_bit_copies_for_combine): Likewise.
5802 * rtlanal.c (nonzero_bits1): Likewise.
5803 (num_sign_bit_copies1): Likewise.
5804 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5805 (reg_num_sign_bit_copies_general): Likewise.
5806 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5807 (reg_nonzero_bits_general): Likewise.
5809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5810 Alan Hayward <alan.hayward@arm.com>
5811 David Sherwood <david.sherwood@arm.com>
5813 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5814 than in subroutines. Return 1 for non-integer modes.
5815 (cached_num_sign_bit_copies): Change the type of the mode parameter
5817 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5818 classes. Handle CONST_INT_P first and then check whether X also
5819 has a scalar integer mode. Check the same thing for inner registers
5820 of a SUBREG and for values that are being extended or truncated.
5822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5823 Alan Hayward <alan.hayward@arm.com>
5824 David Sherwood <david.sherwood@arm.com>
5826 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5827 in subroutines. Return the mode mask for non-integer modes.
5828 (cached_nonzero_bits): Change the type of the mode parameter
5830 (nonzero_bits1): Likewise. Remove early exit for other mode
5831 classes. Handle CONST_INT_P first and then check whether X
5832 also has a scalar integer mode.
5834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5835 Alan Hayward <alan.hayward@arm.com>
5836 David Sherwood <david.sherwood@arm.com>
5838 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5839 Return a scalar_int_mode and assert that the size is greater than
5841 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5842 (op_by_pieces_d::op_by_pieces_d): Likewise.
5843 (op_by_pieces_d::run): Likewise.
5844 (can_store_by_pieces): Likewise.
5846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5847 Alan Hayward <alan.hayward@arm.com>
5848 David Sherwood <david.sherwood@arm.com>
5850 * combine.c (extract_left_shift): Add a mode argument and update
5852 (make_compound_operation_int): Change the type of the mode parameter
5853 to scalar_int_mode and update the call to extract_left_shift.
5855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5856 Alan Hayward <alan.hayward@arm.com>
5857 David Sherwood <david.sherwood@arm.com>
5859 * combine.c (simplify_and_const_int): Change the type of the mode
5860 parameter to scalar_int_mode.
5861 (simplify_and_const_int_1): Likewise. Update recursive call.
5863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5864 Alan Hayward <alan.hayward@arm.com>
5865 David Sherwood <david.sherwood@arm.com>
5867 * combine.c (simplify_compare_const): Check that the mode is a
5868 scalar_int_mode (rather than VOIDmode) before testing its
5870 (simplify_comparison): Move COMPARISON_P handling out of the
5871 loop and restrict the latter part of the loop to scalar_int_modes.
5872 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5873 and when considering SUBREG_REGs. Use is_int_mode instead of
5874 checking GET_MODE_CLASS against MODE_INT.
5876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5877 Alan Hayward <alan.hayward@arm.com>
5878 David Sherwood <david.sherwood@arm.com>
5880 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5881 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5882 shift_unit_mode and for modes involved in scalar shifts.
5884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5885 Alan Hayward <alan.hayward@arm.com>
5886 David Sherwood <david.sherwood@arm.com>
5888 * combine.c (force_int_to_mode): New function, split out from...
5889 (force_to_mode): ...here. Keep xmode up-to-date and use it
5890 instead of GET_MODE (x).
5892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5893 Alan Hayward <alan.hayward@arm.com>
5894 David Sherwood <david.sherwood@arm.com>
5896 * optabs-query.h (extraction_insn::struct_mode): Change type to
5897 opt_scalar_int_mode and update comment.
5898 (extraction_insn::field_mode): Change type to scalar_int_mode.
5899 (extraction_insn::pos_mode): Likewise.
5900 * combine.c (make_extraction): Update accordingly.
5901 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5902 (get_optab_extraction_insn): Likewise.
5903 * recog.c (simplify_while_replacing): Likewise.
5904 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5905 parameter to opt_scalar_int_mode.
5907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5908 Alan Hayward <alan.hayward@arm.com>
5909 David Sherwood <david.sherwood@arm.com>
5911 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5912 to a scalar_int_mode instead of a machine_mode.
5913 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5914 (get_best_mode): Return a boolean and use a pointer argument to store
5915 the selected mode. Replace the limit mode parameter with a bit limit.
5916 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5917 for the values returned by bit_field_mode_iterator::next_mode.
5918 (store_bit_field): Update call to get_best_mode.
5919 (store_fixed_bit_field): Likewise.
5920 (extract_fixed_bit_field): Likewise.
5921 * expr.c (optimize_bitfield_assignment_op): Likewise.
5922 * fold-const.c (optimize_bit_field_compare): Likewise.
5923 (fold_truth_andor_1): Likewise.
5924 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5925 Update for new type of m_mode.
5926 (get_best_mode): As above.
5928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5929 Alan Hayward <alan.hayward@arm.com>
5930 David Sherwood <david.sherwood@arm.com>
5932 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5933 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5934 (store_bit_field): Check is_a <scalar_int_mode> before calling
5935 strict_volatile_bitfield_p.
5936 (extract_bit_field): Likewise.
5938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5939 Alan Hayward <alan.hayward@arm.com>
5940 David Sherwood <david.sherwood@arm.com>
5942 * target.def (cstore_mode): Return a scalar_int_mode.
5943 * doc/tm.texi: Regenerate.
5944 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5945 * targhooks.h (default_cstore_mode): Likewise.
5946 * targhooks.c (default_cstore_mode): Likewise, using a forced
5948 * expmed.c (emit_cstore): Expect the target of the cstore to be
5951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5952 Alan Hayward <alan.hayward@arm.com>
5953 David Sherwood <david.sherwood@arm.com>
5955 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5957 (niter_desc): Likewise mode.
5958 (iv_analyze): Add a mode parameter.
5960 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5961 and change its type to scalar_int_mode.
5962 * loop-iv.c: Update commentary at head of file.
5963 (iv_constant): Pass the mode paraeter before the rtx it describes
5964 and change its type to scalar_int_mode. Remove VOIDmode handling.
5965 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5966 (iv_extend): Likewise.
5967 (shorten_into_mode): Likewise.
5968 (iv_add): Use scalar_int_mode.
5969 (iv_mult): Likewise.
5970 (iv_shift): Likewise.
5971 (canonicalize_iv_subregs): Likewise.
5972 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5973 it describes and change its mode to scalar_int_mode. Also change
5974 the type of the returned inner_mode to scalar_int_mode.
5975 (get_biv_step): Likewise, turning outer_mode from a pointer
5976 into a direct parameter. Update call to get_biv_step_1.
5977 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5978 iv_constant and get_biv_step.
5979 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5980 and change its type to scalar_int_mode. Don't initialise iv->mode
5981 to VOIDmode and remove later checks for its still being VOIDmode.
5982 Update calls to iv_analyze_op and iv_analyze_expr. Check
5983 is_a <scalar_int_mode> when changing the mode under consideration.
5984 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5985 Update call to iv_analyze_expr.
5986 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5987 inner register is not also a scalar_int_mode. Update call to
5989 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5990 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5991 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5992 separate mode class checks. Update calls to iv_analyze. Remove
5993 fix-up of VOIDmodes after iv_analyze_biv.
5994 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5995 don't have a scalar_int_mode. Update call to biv_p.
5997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5998 Alan Hayward <alan.hayward@arm.com>
5999 David Sherwood <david.sherwood@arm.com>
6001 * cfgexpand.c (convert_debug_memory_address): Use
6002 as_a <scalar_int_mode>.
6003 * combine.c (expand_compound_operation): Likewise.
6004 (make_extraction): Likewise.
6005 (change_zero_ext): Likewise.
6006 (simplify_comparison): Likewise.
6007 * cse.c (cse_insn): Likewise.
6008 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6009 (mem_loc_descriptor): Likewise.
6010 (loc_descriptor): Likewise.
6011 * expmed.c (init_expmed_one_mode): Likewise.
6012 (synth_mult): Likewise.
6013 (emit_store_flag_1): Likewise.
6014 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6015 of a comparison with size.
6016 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6017 (reduce_to_bit_field_precision): Likewise.
6018 * function.c (expand_function_end): Likewise.
6019 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6020 * loop-doloop.c (doloop_modify): Likewise.
6021 * optabs.c (expand_binop): Likewise.
6022 (expand_unop): Likewise.
6023 (expand_copysign_absneg): Likewise.
6024 (prepare_cmp_insn): Likewise.
6025 (maybe_legitimize_operand): Likewise.
6026 * recog.c (const_scalar_int_operand): Likewise.
6027 * rtlanal.c (get_address_mode): Likewise.
6028 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6029 (simplify_cond_clz_ctz): Likewise.
6030 * tree-nested.c (get_nl_goto_field): Likewise.
6031 * tree.c (build_vector_type_for_mode): Likewise.
6032 * var-tracking.c (use_narrower_mode): Likewise.
6034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6035 Alan Hayward <alan.hayward@arm.com>
6036 David Sherwood <david.sherwood@arm.com>
6038 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6039 * builtins.c (expand_builtin_signbit): Use it.
6040 * cfgexpand.c (expand_debug_expr): Likewise.
6041 * dojump.c (do_jump): Likewise.
6042 (do_compare_and_jump): Likewise.
6043 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6044 * expmed.c (make_tree): Likewise.
6045 * expr.c (expand_expr_real_2): Likewise.
6046 (expand_expr_real_1): Likewise.
6047 (try_casesi): Likewise.
6048 * fold-const-call.c (fold_const_call_ss): Likewise.
6049 * fold-const.c (unextend): Likewise.
6050 (extract_muldiv_1): Likewise.
6051 (fold_single_bit_test): Likewise.
6052 (native_encode_int): Likewise.
6053 (native_encode_string): Likewise.
6054 (native_interpret_int): Likewise.
6055 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6056 * internal-fn.c (expand_addsub_overflow): Likewise.
6057 (expand_neg_overflow): Likewise.
6058 (expand_mul_overflow): Likewise.
6059 (expand_arith_overflow): Likewise.
6060 * match.pd: Likewise.
6061 * stor-layout.c (layout_type): Likewise.
6062 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6063 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6064 * tree-ssanames.c (get_range_info): Likewise.
6065 * tree-switch-conversion.c (array_value_type) Likewise.
6066 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6067 (vect_recog_divmod_pattern): Likewise.
6068 (vect_recog_mixed_size_cond_pattern): Likewise.
6069 * tree-vrp.c (extract_range_basic): Likewise.
6070 (simplify_float_conversion_using_ranges): Likewise.
6071 * tree.c (int_fits_type_p): Likewise.
6072 * ubsan.c (instrument_bool_enum_load): Likewise.
6073 * varasm.c (mergeable_string_section): Likewise.
6074 (narrowing_initializer_constant_valid_p): Likewise.
6075 (output_constant): Likewise.
6077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6078 Alan Hayward <alan.hayward@arm.com>
6079 David Sherwood <david.sherwood@arm.com>
6081 * machmode.h (NARROWEST_INT_MODE): New macro.
6082 * expr.c (alignment_for_piecewise_move): Use it instead of
6083 GET_CLASS_NARROWEST_MODE (MODE_INT).
6084 (push_block): Likewise.
6085 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6087 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6090 Alan Hayward <alan.hayward@arm.com>
6091 David Sherwood <david.sherwood@arm.com>
6093 * postreload.c (move2add_valid_value_p): Change the type of the
6094 mode parameter to scalar_int_mode.
6095 (move2add_use_add2_insn): Add a mode parameter and use it instead
6097 (move2add_use_add3_insn): Likewise.
6098 (reload_cse_move2add): Update accordingly.
6100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6101 Alan Hayward <alan.hayward@arm.com>
6102 David Sherwood <david.sherwood@arm.com>
6104 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6106 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6107 * optabs.c (expand_unop): Likewise.
6109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6110 Alan Hayward <alan.hayward@arm.com>
6111 David Sherwood <david.sherwood@arm.com>
6113 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6114 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6115 (popcount_loc_descriptor): Likewise.
6116 (bswap_loc_descriptor): Likewise.
6117 (rotate_loc_descriptor): Likewise.
6118 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6119 calling the functions above.
6121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6122 Alan Hayward <alan.hayward@arm.com>
6123 David Sherwood <david.sherwood@arm.com>
6125 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6127 (try_combine): Likewise.
6128 (simplify_if_then_else): Likewise.
6129 * cse.c (cse_insn): Likewise.
6130 * dwarf2out.c (mem_loc_descriptor): Likewise.
6131 * emit-rtl.c (gen_lowpart_common): Likewise.
6132 * simplify-rtx.c (simplify_truncation): Likewise.
6133 (simplify_binary_operation_1): Likewise.
6134 (simplify_const_relational_operation): Likewise.
6135 (simplify_ternary_operation): Likewise.
6136 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6139 Alan Hayward <alan.hayward@arm.com>
6140 David Sherwood <david.sherwood@arm.com>
6142 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6143 * reload.c (push_reload): Likewise.
6144 (find_reloads): Likewise.
6146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6147 Alan Hayward <alan.hayward@arm.com>
6148 David Sherwood <david.sherwood@arm.com>
6150 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6151 (make_compound_operation_int): Likewise.
6152 (change_zero_ext): Likewise.
6153 * expr.c (convert_move): Likewise.
6154 (convert_modes): Likewise.
6155 * fwprop.c (forward_propagate_subreg): Likewise.
6156 * loop-iv.c (get_biv_step_1): Likewise.
6157 * optabs.c (widen_operand): Likewise.
6158 * postreload.c (move2add_valid_value_p): Likewise.
6159 * recog.c (simplify_while_replacing): Likewise.
6160 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6161 (simplify_binary_operation_1): Likewise. Remove redundant
6162 mode equality check.
6164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6165 Alan Hayward <alan.hayward@arm.com>
6166 David Sherwood <david.sherwood@arm.com>
6168 * combine.c (combine_simplify_rtx): Add checks for
6169 is_a <scalar_int_mode>.
6170 (simplify_if_then_else): Likewise.
6171 (make_field_assignment): Likewise.
6172 (simplify_comparison): Likewise.
6173 * ifcvt.c (noce_try_bitop): Likewise.
6174 * loop-invariant.c (canonicalize_address_mult): Likewise.
6175 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6178 Alan Hayward <alan.hayward@arm.com>
6179 David Sherwood <david.sherwood@arm.com>
6181 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6182 is_a <scalar_int_mode> instead of != BLKmode.
6184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6185 Alan Hayward <alan.hayward@arm.com>
6186 David Sherwood <david.sherwood@arm.com>
6188 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6189 instead of != VOIDmode.
6190 * combine.c (if_then_else_cond): Likewise.
6191 (change_zero_ext): Likewise.
6192 * dwarf2out.c (mem_loc_descriptor): Likewise.
6193 (loc_descriptor): Likewise.
6194 * rtlanal.c (canonicalize_condition): Likewise.
6195 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6198 Alan Hayward <alan.hayward@arm.com>
6199 David Sherwood <david.sherwood@arm.com>
6201 * simplify-rtx.c (simplify_binary_operation_1): Use
6202 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6205 Alan Hayward <alan.hayward@arm.com>
6206 David Sherwood <david.sherwood@arm.com>
6208 * wide-int.h (int_traits<unsigned char>) New class.
6209 (int_traits<unsigned short>) Likewise.
6210 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6211 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6213 * combine.c (set_nonzero_bits_and_sign_copies): Use
6214 is_a <scalar_int_mode>.
6215 (find_split_point): Likewise.
6216 (combine_simplify_rtx): Likewise.
6217 (simplify_logical): Likewise.
6218 (expand_compound_operation): Likewise.
6219 (expand_field_assignment): Likewise.
6220 (make_compound_operation): Likewise.
6221 (extended_count): Likewise.
6222 (change_zero_ext): Likewise.
6223 (simplify_comparison): Likewise.
6224 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6225 (ucompare_loc_descriptor): Likewise.
6226 (minmax_loc_descriptor): Likewise.
6227 (mem_loc_descriptor): Likewise.
6228 (loc_descriptor): Likewise.
6229 * expmed.c (init_expmed_one_mode): Likewise.
6230 * lra-constraints.c (lra_constraint_offset): Likewise.
6231 * optabs.c (prepare_libcall_arg): Likewise.
6232 * postreload.c (move2add_note_store): Likewise.
6233 * reload.c (operands_match_p): Likewise.
6234 * rtl.h (load_extend_op): Likewise.
6235 * rtlhooks.c (gen_lowpart_general): Likewise.
6236 * simplify-rtx.c (simplify_truncation): Likewise.
6237 (simplify_unary_operation_1): Likewise.
6238 (simplify_binary_operation_1): Likewise.
6239 (simplify_const_binary_operation): Likewise.
6240 (simplify_const_relational_operation): Likewise.
6241 (simplify_subreg): Likewise.
6242 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6243 * var-tracking.c (adjust_mems): Likewise.
6244 (prepare_call_arguments): Likewise.
6246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6247 Alan Hayward <alan.hayward@arm.com>
6248 David Sherwood <david.sherwood@arm.com>
6250 * machmode.h (is_int_mode): New fuction.
6251 * combine.c (find_split_point): Use it.
6252 (combine_simplify_rtx): Likewise.
6253 (simplify_if_then_else): Likewise.
6254 (simplify_set): Likewise.
6255 (simplify_shift_const_1): Likewise.
6256 (simplify_comparison): Likewise.
6257 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6258 * cse.c (notreg_cost): Likewise.
6259 (cse_insn): Likewise.
6260 * cselib.c (cselib_lookup_1): Likewise.
6261 * dojump.c (do_jump_1): Likewise.
6262 (do_compare_rtx_and_jump): Likewise.
6263 * dse.c (get_call_args): Likewise.
6264 * dwarf2out.c (rtl_for_decl_init): Likewise.
6265 (native_encode_initializer): Likewise.
6266 * expmed.c (emit_store_flag_1): Likewise.
6267 (emit_store_flag): Likewise.
6268 * expr.c (convert_modes): Likewise.
6269 (store_field): Likewise.
6270 (expand_expr_real_1): Likewise.
6271 * fold-const.c (fold_read_from_constant_string): Likewise.
6272 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6273 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6274 * optabs.c (expand_binop): Likewise.
6275 (expand_unop): Likewise.
6276 (expand_abs_nojump): Likewise.
6277 (expand_one_cmpl_abs_nojump): Likewise.
6278 * simplify-rtx.c (mode_signbit_p): Likewise.
6279 (val_signbit_p): Likewise.
6280 (val_signbit_known_set_p): Likewise.
6281 (val_signbit_known_clear_p): Likewise.
6282 (simplify_relational_operation_1): Likewise.
6283 * tree.c (vector_type_mode): Likewise.
6285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6286 Alan Hayward <alan.hayward@arm.com>
6287 David Sherwood <david.sherwood@arm.com>
6289 * machmode.h (smallest_mode_for_size): Fix formatting.
6290 (smallest_int_mode_for_size): New function.
6291 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6292 instead of smallest_mode_for_size.
6293 * combine.c (make_extraction): Likewise.
6294 * config/arc/arc.c (arc_expand_movmem): Likewise.
6295 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6296 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6297 * config/s390/s390.c (s390_expand_insv): Likewise.
6298 * config/sparc/sparc.c (assign_int_registers): Likewise.
6299 * config/spu/spu.c (spu_function_value): Likewise.
6300 (spu_function_arg): Likewise.
6301 * coverage.c (get_gcov_type): Likewise.
6302 (get_gcov_unsigned_t): Likewise.
6303 * dse.c (find_shift_sequence): Likewise.
6304 * expmed.c (store_bit_field_1): Likewise.
6305 * expr.c (convert_move): Likewise.
6306 (store_field): Likewise.
6307 * internal-fn.c (expand_arith_overflow): Likewise.
6308 * optabs-query.c (get_best_extraction_insn): Likewise.
6309 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6310 * stor-layout.c (layout_type): Likewise.
6311 (initialize_sizetypes): Likewise.
6312 * targhooks.c (default_get_mask_mode): Likewise.
6313 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6316 Alan Hayward <alan.hayward@arm.com>
6317 David Sherwood <david.sherwood@arm.com>
6319 * machmode.h (opt_mode::else_blk): New function.
6320 (int_mode_for_mode): Declare.
6321 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6322 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6324 * cfgexpand.c (expand_debug_expr): Likewise.
6325 * combine.c (gen_lowpart_or_truncate): Likewise.
6326 (gen_lowpart_for_combine): Likewise.
6327 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6328 * config/avr/avr.c (avr_to_int_mode): Likewise.
6329 (avr_out_plus_1): Likewise.
6330 (avr_out_plus): Likewise.
6331 (avr_out_round): Likewise.
6332 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6333 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6334 (s390_expand_vcond): Likewise.
6335 * config/spu/spu.c (spu_split_immediate): Likewise.
6336 (spu_expand_mov): Likewise.
6337 * dse.c (get_stored_val): Likewise.
6338 * expmed.c (store_bit_field_1): Likewise.
6339 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6341 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6342 (extract_low_bits): Likewise.
6343 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6344 handling rather than repeating the check.
6345 (emit_group_store): Likewise.
6346 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6347 * optabs.c (expand_absneg_bit): Likewise.
6348 (expand_copysign_absneg): Likewise.
6349 (expand_copysign_bit): Likewise.
6350 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6351 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6352 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6353 * var-tracking.c (prepare_call_arguments): Likewise.
6354 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6355 int_mode_for_mode instead of mode_for_size.
6356 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6359 Alan Hayward <alan.hayward@arm.com>
6360 David Sherwood <david.sherwood@arm.com>
6362 * machmode.h (int_mode_for_size): New function.
6363 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6364 instead of mode_for_size.
6365 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6367 * combine.c (expand_field_assignment): Use int_mode_for_size
6368 instead of mode_for_size.
6369 (make_extraction): Likewise.
6370 (simplify_shift_const_1): Likewise.
6371 (simplify_comparison): Likewise.
6372 * dojump.c (do_jump): Likewise.
6373 * dwarf2out.c (mem_loc_descriptor): Likewise.
6374 * emit-rtl.c (init_derived_machine_modes): Likewise.
6375 * expmed.c (flip_storage_order): Likewise.
6376 (convert_extracted_bit_field): Likewise.
6377 * expr.c (copy_blkmode_from_reg): Likewise.
6378 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6379 * internal-fn.c (expand_mul_overflow): Likewise.
6380 * lower-subreg.c (simple_move): Likewise.
6381 * optabs-libfuncs.c (init_optabs): Likewise.
6382 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6383 * tree.c (vector_type_mode): Likewise.
6384 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6385 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6386 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6387 * tree-vect-stmts.c (vectorizable_load): Likewise.
6388 (vectorizable_store): Likewise.
6390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6391 Alan Hayward <alan.hayward@arm.com>
6392 David Sherwood <david.sherwood@arm.com>
6394 * coretypes.h (pod_mode): New type.
6395 (scalar_int_mode_pod): New typedef.
6396 * machmode.h (pod_mode): New class.
6397 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6398 * genmodes.c (emit_mode_int_n): Update accordingly.
6399 * lower-subreg.h (target_lower_subreg): Change type to
6400 scalar_int_mode_pod.
6401 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6402 scalar_int_mode_pod.
6404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6405 Alan Hayward <alan.hayward@arm.com>
6406 David Sherwood <david.sherwood@arm.com>
6408 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6409 machine_mode to scalar_int_mode.
6410 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6411 (rs6000_option_override_internal): Remove cast to int.
6412 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6413 machine_mode to scalar_int_mode.
6414 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6415 (rs6000_option_override_internal): Remove cast to int.
6416 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6417 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6419 * config/s390/s390.c (s390_expand_builtin): Likewise.
6420 * coretypes.h (scalar_int_mode): New type.
6421 (opt_scalar_int_mode): New typedef.
6422 * machmode.h (scalar_int_mode): New class.
6423 (scalar_int_mode::includes_p): New function.
6424 (byte_mode): Change type to scalar_int_mode.
6425 (word_mode): Likewise.
6426 (ptr_mode): Likewise.
6427 * emit-rtl.c (byte_mode): Likewise.
6428 (word_mode): Likewise.
6429 (ptr_mode): Likewise.
6430 (init_derived_machine_modes): Update accordingly.
6431 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6432 and MODE_PARTIAL_INT.
6433 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6434 opt_scalar_int_mode.
6436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6437 Alan Hayward <alan.hayward@arm.com>
6438 David Sherwood <david.sherwood@arm.com>
6440 * target.def (libgcc_floating_mode_supported_p): Take a
6442 * doc/tm.texi: Regenerate.
6443 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6445 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6446 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6450 Alan Hayward <alan.hayward@arm.com>
6451 David Sherwood <david.sherwood@arm.com>
6453 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6454 * doc/tm.texi: Regenerate.
6455 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6456 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6457 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6458 * targhooks.h (default_floatn_mode): Likewise.
6459 * targhooks.c (default_floatn_mode): Likewise.
6460 * tree.c (build_common_tree_nodes): Update accordingly.
6462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6463 Alan Hayward <alan.hayward@arm.com>
6464 David Sherwood <david.sherwood@arm.com>
6466 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6467 (mode_iterator::iterate_p): Likewise.
6468 (mode_iterator::get_wider): Likewise.
6469 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6472 Alan Hayward <alan.hayward@arm.com>
6473 David Sherwood <david.sherwood@arm.com>
6475 * coretypes.h (opt_scalar_float_mode): New typedef.
6476 * machmode.h (float_mode_for_size): New function.
6477 * emit-rtl.c (double_mode): Delete.
6478 (init_emit_once): Use float_mode_for_size.
6479 * stor-layout.c (layout_type): Likewise.
6480 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6483 Alan Hayward <alan.hayward@arm.com>
6484 David Sherwood <david.sherwood@arm.com>
6486 * output.h (assemble_real): Take a scalar_float_mode.
6487 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6488 * config/arm/arm.md (consttable_4): Likewise.
6489 (consttable_8): Likewise.
6490 (consttable_16): Likewise.
6491 * config/mips/mips.md (consttable_float): Likewise.
6492 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6493 * varasm.c (assemble_real): Take a scalar_float_mode.
6494 (output_constant_pool_2): Update accordingly.
6495 (output_constant): Likewise.
6497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6498 Alan Hayward <alan.hayward@arm.com>
6499 David Sherwood <david.sherwood@arm.com>
6501 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6502 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6503 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6504 (native_encode_real): Likewise.
6505 (native_interpret_real): Likewise.
6506 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6507 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6510 Alan Hayward <alan.hayward@arm.com>
6511 David Sherwood <david.sherwood@arm.com>
6513 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6514 <scalar_float_mode>. Simplify.
6515 (gen_extend_conv_libfunc): Likewise.
6517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6518 Alan Hayward <alan.hayward@arm.com>
6519 David Sherwood <david.sherwood@arm.com>
6521 * coretypes.h (scalar_float_mode): New type.
6522 * machmode.h (mode_traits::from_int): Use machine_mode if
6523 USE_ENUM_MODES is defined.
6524 (is_a): New function.
6526 (dyn_cast): Likewise.
6527 (scalar_float_mode): New class.
6528 (scalar_float_mode::includes_p): New function.
6529 (is_float_mode): Likewise.
6530 * gdbhooks.py (MachineModePrinter): New class.
6531 (build_pretty_printer): Use it for scalar_float_mode.
6532 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6533 (format_helper::format_helper): Turn into a template.
6534 * genmodes.c (get_mode_class): New function.
6535 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6536 or machine_mode if none.
6537 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6538 as_a <scalar_float_mode>.
6539 * dwarf2out.c (mem_loc_descriptor): Likewise.
6540 (insert_float): Likewise.
6541 (add_const_value_attribute): Likewise.
6542 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6543 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6544 (expand_unop): Update accordingly.
6545 (expand_abs_nojump): Likewise.
6546 (expand_copysign_absneg): Take a scalar_float_mode.
6547 (expand_copysign_bit): Likewise.
6548 (expand_copysign): Update accordingly.
6550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6551 Alan Hayward <alan.hayward@arm.com>
6552 David Sherwood <david.sherwood@arm.com>
6554 * coretypes.h (opt_mode): New class.
6555 * machmode.h (opt_mode): Likewise.
6556 (opt_mode::else_void): New function.
6557 (opt_mode::require): Likewise.
6558 (opt_mode::exists): Likewise.
6559 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6560 (GET_MODE_2XWIDER_MODE): Likewise.
6561 (mode_iterator::get_wider): Update accordingly.
6562 (mode_iterator::get_2xwider): Likewise.
6563 (mode_iterator::get_known_wider): Likewise, turning into a template.
6564 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6565 forcing a wider mode to exist.
6566 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6567 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6568 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6569 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6570 * lower-subreg.c (init_lower_subreg): Likewise.
6571 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6572 on the final iteration.
6573 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6574 a wider mode exists before asking for a move pattern.
6575 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6576 forcing a wider mode to exist.
6577 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6578 returning false if no such mode exists.
6579 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6580 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6581 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6582 Avoid checking for a MODE_INT if we already know the mode is not a
6584 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6585 forcing a wider mode to exist.
6586 (expmed_mult_highpart_optab): Likewise.
6587 (expmed_mult_highpart): Likewise.
6588 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6590 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6591 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6592 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6593 * internal-fn.c (expand_mul_overflow): Update use of
6594 GET_MODE_2XWIDER_MODE.
6595 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6596 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6597 GET_MODE_WIDER_MODE.
6598 (convert_plusminus_to_widen): Likewise.
6599 * tree-switch-conversion.c (array_value_type): Likewise.
6600 * var-tracking.c (emit_note_insn_var_location): Likewise.
6601 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6602 Return false inside rather than outside the loop if no wider mode
6604 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6605 and GET_MODE_2XWIDER_MODE
6606 (can_compare_p): Use else_void.
6607 * gdbhooks.py (OptMachineModePrinter): New class.
6608 (build_pretty_printer): Use it for opt_mode.
6610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6611 Alan Hayward <alan.hayward@arm.com>
6612 David Sherwood <david.sherwood@arm.com>
6614 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6615 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6618 Alan Hayward <alan.hayward@arm.com>
6619 David Sherwood <david.sherwood@arm.com>
6621 * machmode.h (mode_traits): New structure.
6622 (get_narrowest_mode): New function.
6623 (mode_iterator::start): Likewise.
6624 (mode_iterator::iterate_p): Likewise.
6625 (mode_iterator::get_wider): Likewise.
6626 (mode_iterator::get_known_wider): Likewise.
6627 (mode_iterator::get_2xwider): Likewise.
6628 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6629 (FOR_EACH_MODE): Likewise.
6630 (FOR_EACH_MODE_FROM): Likewise.
6631 (FOR_EACH_MODE_UNTIL): Likewise.
6632 (FOR_EACH_WIDER_MODE): Likewise.
6633 (FOR_EACH_2XWIDER_MODE): Likewise.
6634 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6635 * combine.c (simplify_comparison): Likewise
6636 * config/i386/i386.c (type_natural_mode): Likewise.
6637 * cse.c (cse_insn): Likewise.
6638 * dse.c (find_shift_sequence): Likewise.
6639 * emit-rtl.c (init_derived_machine_modes): Likewise.
6640 (init_emit_once): Likewise.
6641 * explow.c (hard_function_value): Likewise.
6642 * expmed.c (extract_fixed_bit_field_1): Likewise.
6643 (extract_bit_field_1): Likewise.
6644 (expand_divmod): Likewise.
6645 (emit_store_flag_1): Likewise.
6646 * expr.c (init_expr_target): Likewise.
6647 (convert_move): Likewise.
6648 (alignment_for_piecewise_move): Likewise.
6649 (widest_int_mode_for_size): Likewise.
6650 (emit_block_move_via_movmem): Likewise.
6651 (copy_blkmode_to_reg): Likewise.
6652 (set_storage_via_setmem): Likewise.
6653 (compress_float_constant): Likewise.
6654 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6655 * optabs-query.c (get_best_extraction_insn): Likewise.
6656 * optabs.c (expand_binop): Likewise.
6657 (expand_twoval_unop): Likewise.
6658 (expand_twoval_binop): Likewise.
6659 (widen_leading): Likewise.
6660 (widen_bswap): Likewise.
6661 (expand_parity): Likewise.
6662 (expand_unop): Likewise.
6663 (prepare_cmp_insn): Likewise.
6664 (prepare_float_lib_cmp): Likewise.
6665 (expand_float): Likewise.
6666 (expand_fix): Likewise.
6667 (expand_sfix_optab): Likewise.
6668 * postreload.c (move2add_use_add2_insn): Likewise.
6669 * reg-stack.c (reg_to_stack): Likewise.
6670 * reginfo.c (choose_hard_reg_mode): Likewise.
6671 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6672 * stor-layout.c (mode_for_size): Likewise.
6673 (smallest_mode_for_size): Likewise.
6674 (mode_for_vector): Likewise.
6675 (finish_bitfield_representative): Likewise.
6676 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6677 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6678 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6679 * var-tracking.c (prepare_call_arguments): Likewise.
6681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6682 Alan Hayward <alan.hayward@arm.com>
6683 David Sherwood <david.sherwood@arm.com>
6685 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6686 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6687 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6688 * machmode.h (mode_size): Move earlier in file.
6689 (mode_precision): Likewise.
6690 (mode_inner): Likewise.
6691 (mode_nunits): Likewise.
6692 (mode_unit_size): Likewise.
6693 (unit_unit_precision): Likewise.
6694 (mode_wider): Likewise.
6695 (mode_2xwider): Likewise.
6696 (machine_mode): New class.
6697 (mode_to_bytes): New function.
6698 (mode_to_bits): Likewise.
6699 (mode_to_precision): Likewise.
6700 (mode_to_inner): Likewise.
6701 (mode_to_unit_size): Likewise.
6702 (mode_to_unit_precision): Likewise.
6703 (mode_to_nunits): Likewise.
6704 (GET_MODE_SIZE): Use mode_to_bytes.
6705 (GET_MODE_BITSIZE): Use mode_to_bits.
6706 (GET_MODE_PRECISION): Use mode_to_precision.
6707 (GET_MODE_INNER): Use mode_to_inner.
6708 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6709 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6710 (GET_MODE_NUNITS): Use mode_to_nunits.
6711 * system.h (ALWAYS_INLINE): New macro.
6712 * config/powerpcspe/powerpcspe-c.c
6713 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6714 int for arg1_mode and arg2_mode.
6716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6717 Alan Hayward <alan.hayward@arm.com>
6718 David Sherwood <david.sherwood@arm.com>
6720 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6721 Prefix mode names with E_ in case statements.
6722 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6723 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6724 (aarch64_split_simd_move): Likewise.
6725 (aarch64_gen_storewb_pair): Likewise.
6726 (aarch64_gen_loadwb_pair): Likewise.
6727 (aarch64_gen_store_pair): Likewise.
6728 (aarch64_gen_load_pair): Likewise.
6729 (aarch64_get_condition_code_1): Likewise.
6730 (aarch64_constant_pool_reload_icode): Likewise.
6731 (get_rsqrte_type): Likewise.
6732 (get_rsqrts_type): Likewise.
6733 (get_recpe_type): Likewise.
6734 (get_recps_type): Likewise.
6735 (aarch64_gimplify_va_arg_expr): Likewise.
6736 (aarch64_simd_container_mode): Likewise.
6737 (aarch64_emit_load_exclusive): Likewise.
6738 (aarch64_emit_store_exclusive): Likewise.
6739 (aarch64_expand_compare_and_swap): Likewise.
6740 (aarch64_gen_atomic_cas): Likewise.
6741 (aarch64_emit_bic): Likewise.
6742 (aarch64_emit_atomic_swap): Likewise.
6743 (aarch64_emit_atomic_load_op): Likewise.
6744 (aarch64_evpc_trn): Likewise.
6745 (aarch64_evpc_uzp): Likewise.
6746 (aarch64_evpc_zip): Likewise.
6747 (aarch64_evpc_ext): Likewise.
6748 (aarch64_evpc_rev): Likewise.
6749 (aarch64_evpc_dup): Likewise.
6750 (aarch64_gen_ccmp_first): Likewise.
6751 (aarch64_gen_ccmp_next): Likewise.
6752 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6753 (alpha_emit_xfloating_libcall): Likewise.
6754 (emit_insxl): Likewise.
6755 (alpha_arg_type): Likewise.
6756 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6757 (arc_preferred_simd_mode): Likewise.
6758 (arc_secondary_reload): Likewise.
6759 (get_arc_condition_code): Likewise.
6760 (arc_print_operand): Likewise.
6761 (arc_legitimate_constant_p): Likewise.
6762 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6763 * config/arc/arc.md (casesi_load): Likewise.
6764 (casesi_compact_jump): Likewise.
6765 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6766 (cc_use_register): Likewise.
6767 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6768 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6769 (arm_init_iwmmxt_builtins): Likewise.
6770 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6771 (neon_expand_vector_init): Likewise.
6772 (arm_attr_length_move_neon): Likewise.
6773 (maybe_get_arm_condition_code): Likewise.
6774 (arm_emit_vector_const): Likewise.
6775 (arm_preferred_simd_mode): Likewise.
6776 (arm_output_iwmmxt_tinsr): Likewise.
6777 (thumb1_output_casesi): Likewise.
6778 (thumb2_output_casesi): Likewise.
6779 (arm_emit_load_exclusive): Likewise.
6780 (arm_emit_store_exclusive): Likewise.
6781 (arm_expand_compare_and_swap): Likewise.
6782 (arm_evpc_neon_vuzp): Likewise.
6783 (arm_evpc_neon_vzip): Likewise.
6784 (arm_evpc_neon_vrev): Likewise.
6785 (arm_evpc_neon_vtrn): Likewise.
6786 (arm_evpc_neon_vext): Likewise.
6787 (arm_validize_comparison): Likewise.
6788 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6789 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6790 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6791 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6792 (c6x_preferred_simd_mode): Likewise.
6793 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6794 (epiphany_rtx_costs): Likewise.
6795 * config/epiphany/predicates.md (proper_comparison_operator):
6797 * config/frv/frv.c (condexec_memory_operand): Likewise.
6798 (frv_emit_move): Likewise.
6799 (output_move_single): Likewise.
6800 (output_condmove_single): Likewise.
6801 (frv_hard_regno_mode_ok): Likewise.
6802 (frv_matching_accg_mode): Likewise.
6803 * config/h8300/h8300.c (split_adds_subs): Likewise.
6804 (h8300_rtx_costs): Likewise.
6805 (h8300_print_operand): Likewise.
6806 (compute_mov_length): Likewise.
6807 (output_logical_op): Likewise.
6808 (compute_logical_op_length): Likewise.
6809 (compute_logical_op_cc): Likewise.
6810 (h8300_shift_needs_scratch_p): Likewise.
6811 (output_a_shift): Likewise.
6812 (compute_a_shift_length): Likewise.
6813 (compute_a_shift_cc): Likewise.
6814 (expand_a_rotate): Likewise.
6815 (output_a_rotate): Likewise.
6816 * config/i386/i386.c (classify_argument): Likewise.
6817 (function_arg_advance_32): Likewise.
6818 (function_arg_32): Likewise.
6819 (function_arg_64): Likewise.
6820 (function_value_64): Likewise.
6821 (ix86_gimplify_va_arg): Likewise.
6822 (ix86_legitimate_constant_p): Likewise.
6823 (put_condition_code): Likewise.
6824 (split_double_mode): Likewise.
6825 (ix86_avx256_split_vector_move_misalign): Likewise.
6826 (ix86_expand_vector_logical_operator): Likewise.
6827 (ix86_split_idivmod): Likewise.
6828 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6829 (ix86_build_const_vector): Likewise.
6830 (ix86_build_signbit_mask): Likewise.
6831 (ix86_match_ccmode): Likewise.
6832 (ix86_cc_modes_compatible): Likewise.
6833 (ix86_expand_branch): Likewise.
6834 (ix86_expand_sse_cmp): Likewise.
6835 (ix86_expand_sse_movcc): Likewise.
6836 (ix86_expand_int_sse_cmp): Likewise.
6837 (ix86_expand_vec_perm_vpermi2): Likewise.
6838 (ix86_expand_vec_perm): Likewise.
6839 (ix86_expand_sse_unpack): Likewise.
6840 (ix86_expand_int_addcc): Likewise.
6841 (ix86_split_to_parts): Likewise.
6842 (ix86_vectorize_builtin_gather): Likewise.
6843 (ix86_vectorize_builtin_scatter): Likewise.
6844 (avx_vpermilp_parallel): Likewise.
6845 (inline_memory_move_cost): Likewise.
6846 (ix86_tieable_integer_mode_p): Likewise.
6847 (x86_maybe_negate_const_int): Likewise.
6848 (ix86_expand_vector_init_duplicate): Likewise.
6849 (ix86_expand_vector_init_one_nonzero): Likewise.
6850 (ix86_expand_vector_init_one_var): Likewise.
6851 (ix86_expand_vector_init_concat): Likewise.
6852 (ix86_expand_vector_init_interleave): Likewise.
6853 (ix86_expand_vector_init_general): Likewise.
6854 (ix86_expand_vector_set): Likewise.
6855 (ix86_expand_vector_extract): Likewise.
6856 (emit_reduc_half): Likewise.
6857 (ix86_emit_i387_round): Likewise.
6858 (ix86_mangle_type): Likewise.
6859 (ix86_expand_round_sse4): Likewise.
6860 (expand_vec_perm_blend): Likewise.
6861 (canonicalize_vector_int_perm): Likewise.
6862 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6863 (expand_vec_perm_1): Likewise.
6864 (expand_vec_perm_interleave3): Likewise.
6865 (expand_vec_perm_even_odd_pack): Likewise.
6866 (expand_vec_perm_even_odd_1): Likewise.
6867 (expand_vec_perm_broadcast_1): Likewise.
6868 (ix86_vectorize_vec_perm_const_ok): Likewise.
6869 (ix86_expand_vecop_qihi): Likewise.
6870 (ix86_expand_mul_widen_hilo): Likewise.
6871 (ix86_expand_sse2_abs): Likewise.
6872 (ix86_expand_pextr): Likewise.
6873 (ix86_expand_pinsr): Likewise.
6874 (ix86_preferred_simd_mode): Likewise.
6875 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6876 * config/i386/sse.md (*andnot<mode>3): Likewise.
6877 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6878 (*<code><mode>3): Likewise.
6879 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6880 (ia64_expand_atomic_op): Likewise.
6881 (ia64_arg_type): Likewise.
6882 (ia64_mode_to_int): Likewise.
6883 (ia64_scalar_mode_supported_p): Likewise.
6884 (ia64_vector_mode_supported_p): Likewise.
6885 (expand_vec_perm_broadcast): Likewise.
6886 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6887 (iq2000_function_arg_advance): Likewise.
6888 (iq2000_function_arg): Likewise.
6889 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6890 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6891 (m68k_libcall_value): Likewise.
6892 (m68k_function_value): Likewise.
6893 (sched_attr_op_type): Likewise.
6894 * config/mcore/mcore.c (mcore_output_move): Likewise.
6895 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6897 (microblaze_function_arg): Likewise.
6898 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6899 (mips_print_operand): Likewise.
6900 (mips_mode_ok_for_mov_fmt_p): Likewise.
6901 (mips_vector_mode_supported_p): Likewise.
6902 (mips_preferred_simd_mode): Likewise.
6903 (mips_expand_vpc_loongson_even_odd): Likewise.
6904 (mips_expand_vec_unpack): Likewise.
6905 (mips_expand_vi_broadcast): Likewise.
6906 (mips_expand_vector_init): Likewise.
6907 (mips_expand_vec_reduc): Likewise.
6908 (mips_expand_msa_cmp): Likewise.
6909 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6910 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6911 (cc_flags_for_mode): Likewise.
6912 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6913 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6914 (nds32_output_casesi_pc_relative): Likewise.
6915 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6916 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6917 (nvptx_gen_unpack): Likewise.
6918 (nvptx_gen_pack): Likewise.
6919 (nvptx_gen_shuffle): Likewise.
6920 (nvptx_gen_wcast): Likewise.
6921 (nvptx_preferred_simd_mode): Likewise.
6922 * config/pa/pa.c (pa_secondary_reload): Likewise.
6923 * config/pa/predicates.md (base14_operand): Likewise.
6924 * config/powerpcspe/powerpcspe-c.c
6925 (altivec_resolve_overloaded_builtin): Likewise.
6926 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6928 (rs6000_preferred_simd_mode): Likewise.
6929 (output_vec_const_move): Likewise.
6930 (rs6000_expand_vector_extract): Likewise.
6931 (rs6000_split_vec_extract_var): Likewise.
6932 (reg_offset_addressing_ok_p): Likewise.
6933 (rs6000_legitimate_offset_address_p): Likewise.
6934 (rs6000_legitimize_address): Likewise.
6935 (rs6000_emit_set_const): Likewise.
6936 (rs6000_const_vec): Likewise.
6937 (rs6000_emit_move): Likewise.
6938 (spe_build_register_parallel): Likewise.
6939 (rs6000_darwin64_record_arg_recurse): Likewise.
6940 (swap_selector_for_mode): Likewise.
6941 (spe_init_builtins): Likewise.
6942 (paired_init_builtins): Likewise.
6943 (altivec_init_builtins): Likewise.
6944 (do_load_for_compare): Likewise.
6945 (rs6000_generate_compare): Likewise.
6946 (rs6000_expand_float128_convert): Likewise.
6947 (emit_load_locked): Likewise.
6948 (emit_store_conditional): Likewise.
6949 (rs6000_output_function_epilogue): Likewise.
6950 (rs6000_handle_altivec_attribute): Likewise.
6951 (rs6000_function_value): Likewise.
6952 (emit_fusion_gpr_load): Likewise.
6953 (emit_fusion_p9_load): Likewise.
6954 (emit_fusion_p9_store): Likewise.
6955 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6956 (fusion_gpr_mem_load): Likewise.
6957 (fusion_addis_mem_combo_load): Likewise.
6958 (fusion_addis_mem_combo_store): Likewise.
6959 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6960 (fusion_gpr_mem_load): Likewise.
6961 (fusion_addis_mem_combo_load): Likewise.
6962 (fusion_addis_mem_combo_store): Likewise.
6963 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6965 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6966 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6967 (rs6000_preferred_simd_mode): Likewise.
6968 (output_vec_const_move): Likewise.
6969 (rs6000_expand_vector_extract): Likewise.
6970 (rs6000_split_vec_extract_var): Likewise.
6971 (reg_offset_addressing_ok_p): Likewise.
6972 (rs6000_legitimate_offset_address_p): Likewise.
6973 (rs6000_legitimize_address): Likewise.
6974 (rs6000_emit_set_const): Likewise.
6975 (rs6000_const_vec): Likewise.
6976 (rs6000_emit_move): Likewise.
6977 (rs6000_darwin64_record_arg_recurse): Likewise.
6978 (swap_selector_for_mode): Likewise.
6979 (paired_init_builtins): Likewise.
6980 (altivec_init_builtins): Likewise.
6981 (rs6000_expand_float128_convert): Likewise.
6982 (emit_load_locked): Likewise.
6983 (emit_store_conditional): Likewise.
6984 (rs6000_output_function_epilogue): Likewise.
6985 (rs6000_handle_altivec_attribute): Likewise.
6986 (rs6000_function_value): Likewise.
6987 (emit_fusion_gpr_load): Likewise.
6988 (emit_fusion_p9_load): Likewise.
6989 (emit_fusion_p9_store): Likewise.
6990 * config/rx/rx.c (rx_gen_move_template): Likewise.
6991 (flags_from_mode): Likewise.
6992 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6993 (s390_slb_comparison): Likewise.
6994 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6995 (s390_vector_mode_supported_p): Likewise.
6996 (s390_cc_modes_compatible): Likewise.
6997 (s390_match_ccmode_set): Likewise.
6998 (s390_canonicalize_comparison): Likewise.
6999 (s390_emit_compare_and_swap): Likewise.
7000 (s390_branch_condition_mask): Likewise.
7001 (s390_rtx_costs): Likewise.
7002 (s390_secondary_reload): Likewise.
7003 (__SECONDARY_RELOAD_CASE): Likewise.
7004 (s390_expand_cs): Likewise.
7005 (s390_preferred_simd_mode): Likewise.
7006 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7007 * config/sh/sh.c (sh_print_operand): Likewise.
7008 (dump_table): Likewise.
7009 (sh_secondary_reload): Likewise.
7010 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7011 * config/sh/sh.md (casesi_worker_1): Likewise.
7012 (casesi_worker_2): Likewise.
7013 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7014 (fcc_comparison_operator): Likewise.
7015 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7016 (emit_soft_tfmode_cvt): Likewise.
7017 (sparc_preferred_simd_mode): Likewise.
7018 (output_cbranch): Likewise.
7019 (sparc_print_operand): Likewise.
7020 (sparc_expand_vec_perm_bmask): Likewise.
7021 (vector_init_bshuffle): Likewise.
7022 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7023 (spu_vector_mode_supported_p): Likewise.
7024 (spu_expand_insv): Likewise.
7025 (spu_emit_branch_or_set): Likewise.
7026 (spu_handle_vector_attribute): Likewise.
7027 (spu_builtin_splats): Likewise.
7028 (spu_builtin_extract): Likewise.
7029 (spu_builtin_promote): Likewise.
7030 (spu_expand_sign_extend): Likewise.
7031 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7032 (tilegx_simd_int): Likewise.
7033 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7034 (tilepro_simd_int): Likewise.
7035 * config/v850/v850.c (const_double_split): Likewise.
7036 (v850_print_operand): Likewise.
7037 (ep_memory_offset): Likewise.
7038 * config/vax/vax.c (vax_rtx_costs): Likewise.
7039 (vax_output_int_move): Likewise.
7040 (vax_output_int_add): Likewise.
7041 (vax_output_int_subtract): Likewise.
7042 * config/visium/predicates.md (visium_branch_operator): Likewise.
7043 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7044 (visium_print_operand_address): Likewise.
7045 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7046 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7047 (xtensa_expand_conditional_branch): Likewise.
7048 (xtensa_copy_incoming_a7): Likewise.
7049 (xtensa_output_literal): Likewise.
7050 * dfp.c (decimal_real_maxval): Likewise.
7051 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7054 Alan Hayward <alan.hayward@arm.com>
7055 David Sherwood <david.sherwood@arm.com>
7057 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7058 (mode_nunits_inline): Likewise.
7059 (mode_inner_inline): Likewise.
7060 (mode_unit_size_inline): Likewise.
7061 (mode_unit_precision_inline): Likewise.
7062 (emit_insn_modes_h): Likewise. Also emit a #define of the
7064 (emit_mode_wider): Add an E_ prefix to mode names.
7065 (emit_mode_complex): Likewise.
7066 (emit_mode_inner): Likewise.
7067 (emit_mode_adjustments): Likewise.
7068 (emit_mode_int_n): Likewise.
7069 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7070 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7071 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7072 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7073 (CRC32_BUILTIN, ENTRY): Likewise.
7074 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7075 (aarch64_pop_regs): Likewise.
7076 (aarch64_process_components): Likewise.
7077 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7078 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7079 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7080 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7081 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7082 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7084 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7085 (pcmpistr): Likewise.
7086 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7087 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7088 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7090 * config/rl78/rl78.c (mduc_regs): Likewise.
7091 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7092 (htm_expand_builtin): Likewise.
7093 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7094 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7095 * config/xtensa/xtensa.c (print_operand): Likewise.
7096 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7097 (NUM_MODE_VECTOR_INT): Likewise.
7098 * genoutput.c (null_operand): Likewise.
7099 (output_operand_data): Likewise.
7100 * genrecog.c (print_parameter_value): Likewise.
7101 * lra.c (debug_operand_data): Likewise.
7103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7105 * dwarf2out.c (scompare_loc_descriptor_wide)
7106 (scompare_loc_descriptor_narrow): New functions, split out from...
7107 (scompare_loc_descriptor): ...here.
7108 * expmed.c (emit_store_flag_int): New function, split out from...
7109 (emit_store_flag): ...here.
7111 2017-08-30 Richard Biener <rguenther@suse.de>
7113 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7114 (dwarf2out_early_finish): Move setting of AT_pubnames from
7115 early debug output to early finish.
7117 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7119 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7120 and -mdata-region to the assembler.
7122 2017-08-30 Richard Biener <rguenther@suse.de>
7124 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7126 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7127 (dwarf2out_early_global_decl): Only generate a DIE for the
7128 abstract origin if it doesn't already exist or is a declaration DIE.
7129 (resolve_addr): Do not add the linkage name twice when
7130 generating a stub DIE for the DW_TAG_GNU_call_site target.
7132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7134 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7135 Use machine_mode rather than int for arg1_mode.
7137 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7140 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7141 that the second argument of the built-in functions to unpack
7142 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7143 switch statement instead a lot of if statements.
7144 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7145 Allow 64-bit values to be in Altivec registers as well as
7146 traditional floating point registers.
7147 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7149 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7151 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7152 MAX_REGS_PER_ADDRESS == 1.
7154 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7156 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7157 * config/i386/i386.c (ix86_option_override_internal): Simplify
7158 setting of opts->x_flag_entry.
7160 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7161 Jakub Jelinek <jakub@redhat.com>
7162 Richard Biener <rguenther@suse.de>
7164 PR tree-optimization/81503
7165 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7166 folded constant fits in the target type; reorder tests for clarity.
7168 2017-08-29 Martin Liska <mliska@suse.cz>
7170 * passes.def: Include pass_lower_switch.
7171 * stmt.c (dump_case_nodes): Remove and move to
7172 tree-switch-conversion.
7173 (case_values_threshold): Likewise.
7174 (expand_switch_as_decision_tree_p): Likewise.
7175 (emit_case_decision_tree): Likewise.
7176 (expand_case): Likewise.
7177 (balance_case_nodes): Likewise.
7178 (node_has_low_bound): Likewise.
7179 (node_has_high_bound): Likewise.
7180 (node_is_bounded): Likewise.
7181 (emit_case_nodes): Likewise.
7182 (struct simple_case_node): New struct.
7183 (add_case_node): Remove.
7184 (emit_case_dispatch_table): Use vector instead of case_list.
7185 (reset_out_edges_aux): Remove.
7186 (compute_cases_per_edge): Likewise.
7187 (expand_case): Build list of simple_case_node.
7188 (expand_sjlj_dispatch_table): Use it.
7189 * tree-switch-conversion.c (struct case_node): Moved from
7190 stmt.c and adjusted.
7191 (emit_case_nodes): Likewise.
7192 (node_has_low_bound): Likewise.
7193 (node_has_high_bound): Likewise.
7194 (node_is_bounded): Likewise.
7195 (case_values_threshold): Likewise.
7196 (reset_out_edges_aux): Likewise.
7197 (compute_cases_per_edge): Likewise.
7198 (add_case_node): Likewise.
7199 (dump_case_nodes): Likewise.
7200 (balance_case_nodes): Likewise.
7201 (expand_switch_as_decision_tree_p): Likewise.
7202 (emit_jump): Likewise.
7203 (emit_case_decision_tree): Likewise.
7204 (try_switch_expansion): Likewise.
7205 (do_jump_if_equal): Likewise.
7206 (emit_cmp_and_jump_insns): Likewise.
7207 (fix_phi_operands_for_edge): New function.
7208 (record_phi_operand_mapping): Likewise.
7209 (class pass_lower_switch): New pass.
7210 (pass_lower_switch::execute): New function.
7211 (make_pass_lower_switch): Likewise.
7212 (conditional_probability):
7213 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7214 * tree-pass.h: Add make_pass_lower_switch.
7216 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7219 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7222 2017-08-29 Richard Biener <rguenther@suse.de>
7224 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7225 we do not add a DW_AT_inline attribute twice.
7226 (gen_subprogram_die): Remove code setting DW_AT_inline on
7227 DECL_ABSTRACT_P nodes.
7229 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7231 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7232 calls to internal functions.
7233 (gimplify_modify_expr): Likewise.
7234 * tree-call-cdce.c (use_internal_fn): Likewise.
7235 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7236 (convert_to_divmod): Set the nothrow flag.
7237 * tree-if-conv.c (predicate_mem_writes): Likewise.
7238 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7239 (vectorizable_call): Likewise.
7240 (vectorizable_store): Likewise.
7241 (vectorizable_load): Likewise.
7242 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7243 (vect_recog_mask_conversion_pattern): Likewise.
7245 2017-08-29 Martin Liska <mliska@suse.cz>
7248 * gcc.c (driver_handle_option): Add new argument.
7249 * opts-common.c (handle_option): Pass
7250 target_option_override_hook.
7251 * opts-global.c (lang_handle_option): Add new option.
7252 (set_default_handlers): Add new argument.
7253 (decode_options): Likewise.
7254 * opts.c (target_handle_option): Likewise.
7255 (common_handle_option): Call target_option_override_hook.
7256 * opts.h (struct cl_option_handler_func): Add hook for
7257 target option override.
7258 (struct cl_option_handlers): Likewise.
7259 (set_default_handlers): Add new argument.
7260 (decode_options): Likewise.
7261 (common_handle_option): Likewise.
7262 (target_handle_option): Likewise.
7263 * toplev.c (toplev::main): Pass targetm.target_option.override
7266 2017-08-29 Richard Biener <rguenther@suse.de>
7267 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7269 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7270 life to the active subtree.
7272 2017-08-28 Jeff Law <law@redhat.com>
7274 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7275 derive_equivalences.
7276 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7278 (edge_info::derive_equivalences): New private member function
7280 * tree-ssa-dom.c (class edge_info): Changed from a struct
7281 to a class. Add ctor/dtor, methods and data members.
7282 (edge_info::edge_info): Renamed from allocate_edge_info.
7283 Initialize additional members.
7284 (edge_info::~edge_info): New.
7285 (free_dom_edge_info): Delete the edge info.
7286 (record_edge_info): Use new class & associated member functions.
7287 Tighten forms for testing for edge equivalences.
7288 (record_temporary_equivalences): Iterate over the simple
7289 equivalences rather than assuming there's only one per edge.
7290 (cprop_into_successor_phis): Iterate over the simple
7291 equivalences rather than assuming there's only one per edge.
7292 (optimize_stmt): Use operand_equal_p rather than pointer
7293 equality for mini-DSE code.
7295 2017-08-28 Nathan Sidwell <nathan@acm.org>
7297 * gcc.c (execute): Fold SIGPIPE handling into switch
7298 statement. Adjust internal error message.
7300 2017-08-28 Richard Biener <rguenther@suse.de>
7303 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7304 Do nothing for removed DIEs.
7306 2017-08-28 Richard Biener <rguenther@suse.de>
7308 PR tree-optimization/81977
7309 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7312 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7315 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7316 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7317 using targetm.gen_mem_thread_fence.
7319 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7322 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7323 predicate to register_operand. Reorder operands.
7324 (*btr<mode>): Ditto.
7325 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7326 (*btr<mode>_mask): Ditto.
7328 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7330 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7331 * config/rs6000/xmmintrin.h: New file.
7332 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7334 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7337 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7338 parameter and_insn and return it.
7339 (recombine_lvx_pattern): Insert a copy to ensure availability of
7340 the base register of the copied masking operation at the point of
7341 the instruction replacement.
7342 (recombine_stvx_pattern): Likewise.
7344 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7346 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7347 undocumented switches.
7348 (-mpower9-dform-vector): Likewise.
7349 (-mpower9-dform): Likewise.
7350 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7351 comments to delete references to -mpower9-dform* switches.
7352 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7353 Delete reference to -mpower9-dform* switches, test for
7354 -mpower9-vector instead.
7355 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7356 (OTHER_P9_VECTOR_MASKS): Likewise.
7357 (POWERPC_MASKS): Likewise.
7358 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7359 tests against -mpower9-dform* to -mpower9-vector. Delete code
7360 that checked for -mpower9-dform* consistancy with other options.
7361 Add test for -mpower9-misc to enable other power9 switches.
7362 (rs6000_init_hard_regno_mode_ok): Likewise.
7363 (rs6000_option_override_internal): Likewise.
7364 (rs6000_emit_prologue): Likewise.
7365 (rs6000_emit_epilogue): Likewise.
7366 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7367 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7368 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7370 (emit_fusion_p9_store): Likewise.
7371 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7372 resetting these macros if the assembler does not support ISA 3.0
7374 (TARGET_P9_DFORM_VECTOR): Likewise.
7375 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7376 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7378 2017-08-25 Alan Modra <amodra@gmail.com>
7380 PR rtl-optimization/81747
7381 * cse.c (cse_extended_basic_block): Don't attempt to record
7382 equivalences for degenerate conditional branches that branch
7383 to their fall-through.
7385 2017-08-24 Martin Sebor <msebor@redhat.com>
7388 * gimple-fold.c (size_must_be_zero_p): New function.
7389 (gimple_fold_builtin_memory_op): Call it.
7391 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7393 * config/rs6000/mm_malloc.h: New file.
7395 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7397 PR tree-optimization/81913
7398 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7399 analysis when either IVs in condition can wrap.
7401 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7403 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7404 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7406 2017-08-24 Richard Biener <rguenther@suse.de>
7409 * targhooks.c (default_target_can_inline_p): Properly
7410 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7411 is present and always compare.
7412 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7413 infer -mfpmath=sse from TARGET_SSE_P.
7414 (ix86_can_inline_p): Properly use target_option_default_node when
7415 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7417 2017-08-24 Richard Biener <rguenther@suse.de>
7420 * dwarf2out.c (output_die): Handle flag_generate_offload like
7422 (output_comp_unit): Likewise.
7423 (gen_array_type_die): Likewise.
7424 (dwarf2out_early_finish): Likewise.
7425 (note_variable_value_in_expr): Likewise.
7426 (dwarf2out_finish): Likewise. Adjust assert.
7427 * cgraphunit.c (symbol_table::compile): Move setting of
7428 flag_generate_offload earlier ...
7429 (symbol_table::finalize_compilation_unit): ... here, before
7430 early debug finalization.
7432 2017-08-24 Richard Biener <rguenther@suse.de>
7434 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7435 and ipa-fnsummary.h.
7436 (ix86_can_inline_p): When ix86_fpmath flags do not match
7437 check whether the callee uses FP math at all.
7439 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7442 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7443 instead of TYPE_PRECISION.
7445 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7446 Alan Hayward <alan.hayward@arm.com>
7447 David Sherwood <david.sherwood@arm.com>
7449 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7451 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7453 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7455 (ix86_frame::stack_realign_allocate): New field.
7456 (struct machine_frame_state): Modify comments.
7457 (machine_frame_state::sp_realigned_fp_end): New field.
7458 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7460 (sp_valid_at): Add assertion to assure no attempt to access invalid
7461 offset of a realigned stack.
7462 (fp_valid_at): Likewise.
7463 (choose_baseaddr): Modify comments.
7464 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7465 ix86_expand_prologue.
7466 (ix86_expand_prologue): Modify stack realignment and allocation.
7467 (ix86_expand_epilogue): Modify comments.
7468 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7469 avx2_runtime, avx512f, and avx512f_runtime.
7471 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7473 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7474 (mstackrealign): Do not init to -1.
7475 * config/i386/i386.c (ix86_option_override_internal):
7476 Check opts_set, not opts when setting default value of
7477 opts->x_ix86_force_align_arg_pointer.
7479 2017-08-23 Richard Biener <rguenther@suse.de>
7481 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7482 lang_hooks.decl_printable_name.
7483 * print-rtl-function.c (print_rtx_function): Likewise.
7484 * tree-pretty-print.c (dump_function_header): Likewise.
7486 2017-08-23 Richard Biener <rguenther@suse.de>
7489 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7490 -g0 at compile-time.
7492 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7495 * doc/sourcebuild.texi (Other hardware attributes):
7498 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7501 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7502 Fix single-use check.
7504 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7506 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7508 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7510 * doc/install.texi: Modify to add more details on running selected
7513 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7515 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7516 is combined with -mabi=ms.
7517 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7520 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7522 PR tree-optimization/81488
7523 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7524 and cached_basis fields.
7525 (MAX_SPREAD): New constant.
7526 (alloc_cand_and_find_basis): Initialize new fields.
7527 (clear_visited): New function.
7528 (create_phi_basis_1): Rename from create_phi_basis, set visited
7529 and cached_basis fields.
7530 (create_phi_basis): New wrapper function.
7531 (phi_add_costs_1): Rename from phi_add_costs, add spread
7532 parameter, set visited field, short-circuit when limits reached.
7533 (phi_add_costs): New wrapper function.
7534 (record_phi_increments_1): Rename from record_phi_increments, set
7536 (record_phi_increments): New wrapper function.
7537 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7538 (phi_incr_cost): New wrapper function.
7539 (all_phi_incrs_profitable_1): Rename from
7540 all_phi_incrs_profitable, set visited field.
7541 (all_phi_incrs_profitable): New wrapper function.
7543 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7544 Alan Hayward <alan.hayward@arm.com>
7545 David Sherwood <david.sherwood@arm.com>
7547 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7548 that takes the outer and inner modes.
7549 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7550 comparison as the canonical test for a paradoxical subreg.
7551 * combine.c (simplify_set): Use paradoxical_subreg_p.
7552 (make_extraction): Likewise.
7553 (force_to_mode): Likewise.
7554 (rtx_equal_for_field_assignment_p): Likewise.
7555 (gen_lowpart_for_combine): Likewise.
7556 (simplify_comparison): Likewise.
7557 * cse.c (equiv_constant): Likewise.
7558 * expmed.c (store_bit_field_1): Likewise.
7559 * final.c (alter_subreg): Likewise.
7560 * fwprop.c (propagate_rtx): Likewise.
7561 (forward_propagate_subreg): Likewise.
7562 * ira-conflicts.c (ira_build_conflicts): Likewise.
7563 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7564 * lra-constraints.c (curr_insn_transform): Likewise.
7565 (split_reg): Likewise.
7566 * lra-eliminations.c (move_plus_up): Likewise.
7567 (lra_eliminate_regs_1): Likewise.
7568 * recog.c (general_operand): Likewise.
7569 * ree.c (combine_reaching_defs): Likewise.
7570 * reload.c (push_reload): Likewise.
7571 (find_reloads): Likewise.
7572 * reload1.c (elimination_effects): Likewise.
7573 (compute_reload_subreg_offset): Likewise.
7574 (choose_reload_regs): Likewise.
7575 * rtlanal.c (subreg_lsb_1): Likewise.
7576 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7577 (simplify_subreg): Likewise.
7578 * var-tracking.c (track_loc_p): Likewise.
7579 * emit-rtl.c (byte_lowpart_offset): Likewise.
7580 (paradoxical_subreg_p): Delete out-of-line definition.
7582 2017-08-22 Jeff Law <law@redhat.com>
7584 PR tree-optimization/81741
7585 PR tree-optimization/71947
7586 * tree-ssa-dom.c: Include tree-inline.h.
7587 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7588 equivalences if one is more expensive to compute than the other.
7589 * tree-ssa-scopedtables.h (class const_or_copies): Make
7590 record_const_or_copy_raw method private.
7591 (class avail_exprs_stack): New method simplify_binary_operation.
7592 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7593 avail_exprs_stack::simplify_binary_operation as needed.
7594 (avail_exprs_stack::simplify_binary_operation): New function.
7596 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7598 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7599 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7600 (DOT_SYMBOLS): Likewise.
7601 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7602 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7603 (RS6000_ABI_NAME): Likewise.
7604 (TARGET_CMODEL): Likewise.
7605 (TOC_SECTION_ASM_OP): Likewise.
7606 (SET_CMODEL): New macro.
7607 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7609 2017-08-22 Richard Biener <rguenther@suse.de>
7611 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7612 to allow for free-lang-data replacements similar to verify_type_variant.
7614 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7615 Alan Hayward <alan.hayward@arm.com>
7616 David Sherwood <david.sherwood@arm.com>
7618 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7619 VOIDmode for the LABEL_REF.
7621 2017-08-22 Richard Biener <rguenther@suse.de>
7623 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7625 2017-08-22 Richard Biener <rguenther@suse.de>
7627 * common.opt (feliminate-dwarf2-dups): Ignore.
7628 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7629 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7630 same_die_p_wrap, compute_section_prefix,
7631 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7632 (comdat_symbol_id, comdat_symbol_number): Likewise.
7633 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7635 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7636 (output_die): Mark unreachable path unreachable.
7637 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7638 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7639 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7640 (dwarf2out_early_finish): Likewise.
7642 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7644 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7646 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7649 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7650 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7651 (avr_attribute_table) <io, io_low, address>: Initialize
7652 .decl_required with true.
7654 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7656 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7657 undocumented debugging options.
7658 (-mvsx-scalar-double): Likewise.
7659 (-mallow-df-permute): Likewise.
7660 (-mvectorize-builtins): Likewise.
7661 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7662 (rs6000_builtin_vectorized_function): Likewise.
7663 (rs6000_builtin_md_vectorized_function): Likewise.
7664 (rs6000_opt_vars): Likewise.
7666 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7669 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7670 (*btrq_imm): Rename from *btrq.
7671 (*btcq_imm): Rename from *btcq.
7672 (btsc): New code attribute.
7673 (*<btsc><mode>): New insn pattern.
7674 (*btr<mode>): Ditto.
7675 (*<btsc><mode>_mask): New insn_and_split pattern.
7676 (*btr<mode>_mask): Ditto.
7678 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7679 Alan Hayward <alan.hayward@arm.com>
7680 David Sherwood <david.sherwood@arm.com>
7682 * function.c (pad_below): Simplify padding calculation.
7684 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7685 Alan Hayward <alan.hayward@arm.com>
7686 David Sherwood <david.sherwood@arm.com>
7688 * target.def (function_prologue): Remove frame size argument.
7689 (function_epilogue): Likewise.
7690 * doc/tm.texi: Regenerate.
7691 * final.c (final_start_function): Update call to function_prologue.
7692 (final_end_function): Update call to function_epilogue.
7693 (default_function_pro_epilogue): Remove frame size argument.
7694 * output.h (default_function_pro_epilogue): Likewise.
7695 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7696 (arm_output_function_prologue): Likewise.
7697 * config/frv/frv.c (frv_function_prologue): Likewise.
7698 (frv_function_epilogue): Likewise.
7699 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7700 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7701 (ia64_output_function_epilogue): Likewise.
7702 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7703 (m32r_output_function_epilogue): Likewise.
7704 * config/microblaze/microblaze.c (microblaze_function_prologue)
7705 (microblaze_function_epilogue): Likewise.
7706 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7707 (mips_output_function_epilogue): Likewise.
7708 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7709 (mmix_target_asm_function_epilogue): Likewise.
7710 * config/msp430/msp430.c (msp430_start_function): Likewise.
7711 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7712 (nds32_asm_function_epilogue): Likewise.
7713 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7714 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7715 (pa_output_function_epilogue): Likewise.
7716 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7717 (rs6000_output_function_epilogue): Likewise.
7718 * config/rl78/rl78.c (rl78_start_function): Likewise.
7719 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7720 (rs6000_output_function_epilogue): Likewise.
7721 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7722 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7723 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7724 (sparc_asm_function_epilogue): Likewise.
7726 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7728 * tree.h (type_has_mode_precision_p): New function.
7729 * convert.c (convert_to_integer_1): Use it.
7730 * expr.c (expand_expr_real_2): Likewise.
7731 (expand_expr_real_1): Likewise.
7732 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7733 * match.pd: Likewise.
7734 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7735 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7736 * tree-tailcall.c (process_assignment): Likewise.
7737 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7738 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7739 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7740 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7741 (vectorizable_assignment): Likewise.
7742 (vectorizable_shift): Likewise.
7743 (vectorizable_operation): Likewise.
7744 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7746 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7748 * match.pd: Add pow (C, x) simplification.
7750 2017-08-21 Richard Biener <rguenther@suse.de>
7752 PR tree-optimization/81900
7753 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7754 for blocks with abnormal predecessors.
7755 (compute_antic): Do not set visited flag prematurely.
7757 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7760 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7762 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7764 * stor-layout.h (vector_type_mode): Move to...
7765 * tree.h (vector_type_mode): ...here.
7766 * stor-layout.c (vector_type_mode): Move to...
7767 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7769 2017-08-21 Richard Biener <rguenther@suse.de>
7771 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7772 register_external_die hooks.
7773 (debug_false_tree_charstarstar_uhwistar): Declare.
7774 (debug_nothing_tree_charstar_uhwi): Likewise.
7775 * debug.c (do_nothing_debug_hooks): Adjust.
7776 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7777 (debug_nothing_tree_charstar_uhwi): Likewise.
7778 * dbxout.c (dbx_debug_hooks): Adjust.
7779 (xcoff_debug_hooks): Likewise.
7780 * sdbout.c (sdb_debug_hooks): Likewise.
7781 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7782 * dwarf2out.c (macinfo_label_base): New global.
7783 (dwarf2out_register_external_die): New function for the
7784 register_external_die hook.
7785 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7786 (dwarf2_debug_hooks): Use them.
7787 (dwarf2_lineno_debug_hooks): Adjust.
7788 (struct die_struct): Add with_offset flag.
7789 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7790 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7791 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7792 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7793 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7794 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7795 defining section names for the early LTO debug variants.
7796 (reset_indirect_string): New helper.
7797 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7798 (print_dw_val): Add support for offsetted symbol references.
7799 (get_ultimate_context): Split out from is_cxx.
7800 (is_cxx): Use get_ultimate_context.
7801 (is_fortran): Add decl overload.
7802 (compute_comp_unit_symbol): Split out worker from
7803 compute_section_prefix.
7804 (compute_section_prefix): Call compute_comp_unit_symbol and
7805 set comdat_type_p here.
7806 (output_die): Skip DIE symbol output for the LTO added one.
7807 Handle DIE symbol references with offset.
7808 (output_comp_unit): Guard section name mangling properly.
7809 For LTO debug sections emit a symbol at the section beginning
7810 which we use to refer to its DIEs.
7811 (add_abstract_origin_attribute): For DIEs registered via
7812 dwarf2out_register_external_die directly refer to the early
7813 DIE rather than indirectly through the shadow one we created.
7814 Remove obsolete call to dwarf2out_abstract_function for
7815 non-function/block origins.
7816 (gen_array_type_die): When generating early LTO debug do
7817 not emit DW_AT_string_length.
7818 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7819 late when in LTO. As suggested place a gcc_unreachable for
7820 the DECL_ABSTRACT_P case.
7821 (gen_subprogram_die): Avoid another specification DIE
7822 for early built declarations/definitions for the late LTO case.
7823 (gen_variable_die): Add type references for late duplicated VLA dies
7825 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7826 we have the abstract instance already.
7827 (process_scope_var): Adjust decl DIE contexts in LTO which
7828 first puts them in limbo.
7829 (gen_decl_die): Do not generate type DIEs late apart from
7830 types for VLAs or for decls we do not yet have a DIE. Do not
7831 call dwarf2out_abstract_function late.
7832 (dwarf2out_early_global_decl): Make sure to create DIEs
7833 for abstract instances of a decl first.
7834 (dwarf2out_late_global_decl): Adjust comment.
7835 (output_macinfo_op): With multiple macro sections use
7836 macinfo_label_base to distinguish labels.
7837 (output_macinfo): Likewise. Update macinfo_label_base.
7838 Pass in the line info label.
7839 (note_variable_value_in_expr): When generating LTO resolve
7840 all variable values here by generating DIEs as needed.
7841 (init_sections_and_labels): Add early LTO debug flag parameter
7842 and generate different sections and names if set. Add generation
7843 counter for the labels so we can have multiple of them.
7844 (reset_dies): Helper to allow DIEs to be output multiple times.
7845 (dwarf2out_finish): When outputting DIEs to the fat part of an
7846 LTO object first reset DIEs.
7847 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7848 (modified_type_die): Check for decl_ultimate_origin being self
7850 (gen_type_die_with_usage): Likewise.
7851 (gen_typedef_die): Allow decl_ultimate_origin being self.
7852 (set_decl_abstract_flags): Remove.
7853 (set_block_abstract_flags): Likewise.
7854 (dwarf2out_abstract_function): Treat the early generated DIEs
7855 as the abstract copy and only add DW_AT_inline and
7856 DW_AT_artificial here and call set_decl_origin_self.
7857 If the DIE has an abstract origin don't do anything.
7858 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7859 if we have none yet (Go fails to build one, PR78628).
7860 (variably_modified_type_p): Prevent endless recursion for Ada
7861 cyclic pointer types.
7862 * lto-streamer-in.c: Include debug.h.
7863 (dref_queue): New global.
7864 (lto_read_tree_1): Stream in DIE references.
7865 (lto_input_tree): Register DIE references.
7866 (input_function): Stream DECL_DEBUG_ARGS.
7867 * lto-streamer-out.c: Include debug.h.
7868 (lto_write_tree_1): Output DIE references.
7869 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7870 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7871 (output_function): Stream DECL_DEBUG_ARGS.
7872 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7873 Stream DECL_ABSTRACT_ORIGIN.
7874 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7875 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7876 DECL_CONTEXT for file-scope decls.
7877 * lto-streamer.h (struct dref_entry): Declare.
7878 (dref_queue): Likewise.
7879 * cfgexpand.c (pass_expand::execute): Do not call the
7880 outlining_inline_function hook here.
7881 * lto-wrapper.c (debug_obj): New global.
7882 (tool_cleanup): Unlink it if required.
7883 (debug_objcopy): New function.
7884 (run_gcc): Handle early debug sections in the IL files by
7885 extracting them to separate files, partially linkin them and
7886 feeding the result back as result to the linker.
7887 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7888 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7889 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7890 sections into a separate segment.
7891 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7893 (darwin_asm_dwarf_section): Likewise.
7894 (darwin_asm_output_dwarf_offset): Likewise.
7895 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7897 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7898 Alan Hayward <alan.hayward@arm.com>
7899 David Sherwood <david.sherwood@arm.com>
7901 * read-md.h (md_reader::record_potential_iterator_use): Replace
7902 pointer argument with an rtx and an index.
7903 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7904 (apply_mode_iterator): Likewise.
7905 (apply_code_iterator): Likewise.
7906 (apply_int_iterator): Likewise.
7907 (apply_subst_iterator): Likewise.
7908 (record_iterator_use): Likewise.
7909 (record_attribute_use): Likewise.
7910 (md_reader::record_potential_iterator_use): Likewise. Update calls
7911 to record_iterator_use and apply_iterator.
7912 (iterator_use): Replace ptr with x and index.
7913 (attribute_use): Likewise.
7914 (apply_attribute_uses): Update calls to apply_iterator.
7915 (apply_iterators): Likewise. Update initialization of iterator_use.
7916 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7917 and record_potential_iterator_use.
7918 (rtx_reader::read_rtx_operand): Likewise.
7920 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7921 Alan Hayward <alan.hayward@arm.com>
7922 David Sherwood <david.sherwood@arm.com>
7924 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7927 2017-08-21 Richard Biener <rguenther@suse.de>
7930 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7931 at struct end conservatively when comparing common bases.
7933 2017-08-21 Richard Biener <rguenther@suse.de>
7935 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7936 (mem_ref_in_stmt): Remove.
7937 (determine_max_movement): Use ref index to get at the reference.
7938 (invariantness_dom_walker::before_dom_children): Deal with
7939 lim data already initialized.
7940 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7942 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7944 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7945 (TARGET_ISA_ROUND): Ditto.
7946 (TARGET_ROUND): Ditto.
7947 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7948 * config/i386/i386.md: Ditto.
7949 * config/i386/sse.md: Ditto.
7950 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7951 with OPTION_MASK_ISA_SSE4_1.
7953 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7956 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7957 __builtin_ia32_lzcnt_u16.
7959 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7962 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7963 (rs6000_set_current_function): Rewrite function to use it.
7965 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7968 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7969 and TYPE_WARN_IF_NOT_ALIGN.
7970 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7971 (handle_warn_if_not_align): New.
7972 (place_union_field): Call handle_warn_if_not_align.
7973 (place_field): Call handle_warn_if_not_align.
7974 Copy TYPE_WARN_IF_NOT_ALIGN.
7975 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7976 (layout_type): Likewise.
7977 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7979 (tree_decl_common): Add warn_if_not_align.
7980 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7981 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7982 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7983 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7984 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7985 * doc/extend.texi: Document warn_if_not_aligned attribute.
7986 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7988 2017-08-17 Martin Liska <mliska@suse.cz>
7991 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7992 (get_data_dependence): Use it as pointer type.
7993 (distribute_loop): Likewise.
7995 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7997 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7998 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7999 (p8_vmrgow_<mode>_direct): New define_insn.
8000 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8001 handle endianness for vmrgew and vmrgow permute patterns.
8003 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8005 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8006 * config/rs6000/rs6000-cpus.def: Remove comment.
8007 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8008 (POWERPC_MASKS): Likewise.
8009 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8010 use of TARGET_VSX_TIMODE.
8011 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8012 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8013 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8014 (rs6000_option_override_internal): Remove dead code.
8015 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8016 (rs6000_legitimize_reload_address): Likewise.
8017 (rs6000_legitimate_address_p): Likewise.
8018 (rs6000_opt_masks): Delete "vsx-timode".
8019 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8020 from function comment.
8021 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8022 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8023 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8025 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8026 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8027 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8030 (define_peephole2): Likewise.
8032 2017-08-17 Martin Sebor <msebor@redhat.com>
8035 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8036 past the end of an array.
8037 (test_pp_format): Add test cases.
8039 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8041 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8042 missing ECF_NOTHROW flags.
8044 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8047 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8048 operands residing in integer registers.
8049 (*vsx_le_perm_load_<mode>): Likewise.
8050 (*vsx_le_perm_store_<mode>): Likewise.
8051 (define_peephole2): Add peepholes to optimize the above.
8053 2017-08-17 Marek Polacek <polacek@redhat.com>
8056 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8057 to mimic what shorten_compare did. Change the return type to bool.
8058 (fold_cond_expr_with_comparison): Update call to
8059 operand_equal_for_comparison_p.
8060 (fold_ternary_loc): Likewise.
8062 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8064 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8066 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8068 2017-08-17 Richard Biener <rguenther@suse.de>
8070 * tree-ssa-structalias.c (solve_graph): When propagating
8071 to successors update the graphs succ edges and avoid duplicate work.
8073 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8076 * config/i386/i386.c (ix86_option_override_internal): Save target
8077 specific options after ix86_stack_protector_guard_reg was changed.
8079 2017-08-17 Richard Biener <rguenther@suse.de>
8081 PR tree-optimization/81827
8082 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8083 (new_var_info): Initialize it conservatively.
8084 (get_call_vi): Mark register vars.
8085 (new_scalar_tmp_constraint_exp): Likewise.
8086 (handle_rhs_call): Likewise.
8087 (handle_const_call): Likewise.
8088 (create_function_info_for): Likewise.
8089 (solve_constraints): Sort varinfos to separate register from
8090 non-register vars to pack points-to solution bitmaps during
8093 2017-08-17 Marek Polacek <polacek@redhat.com>
8095 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8097 2017-08-17 Richard Biener <rguenther@suse.de>
8099 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8100 to true when overflow is undefined and we saturated the result.
8102 2017-08-17 Alan Modra <amodra@gmail.com>
8105 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8106 Don't use store multiple if only one reg needs saving.
8107 (interesting_frame_related_regno): New function.
8108 (rs6000_frame_related): Don't emit frame info for regs that
8110 (rs6000_emit_epilogue): Likewise.
8112 2017-08-16 Nathan Sidwell <nathan@acm.org>
8114 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8115 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8116 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8117 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8118 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8119 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8120 (verify_type): Adjust for TYPE_BINFO move.
8121 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8122 process TYPE_BINFO directly.
8123 (hash_tree): Likewise.
8124 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8126 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8129 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8131 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8133 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8136 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8137 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8138 Add ix86_binary_operator_ok to insn constraint.
8139 (*iordi_1_bts): Ditto.
8140 (*xordi_1_btc): Ditto.
8141 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8142 Update corresponding peephole2 pattern.
8146 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8148 PR tree-optimization/81832
8149 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8150 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8152 2017-08-16 Marek Polacek <polacek@redhat.com>
8155 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8156 regarding size_zero_node.
8158 2017-08-16 Martin Liska <mliska@suse.cz>
8161 * config.gcc: Respect previously set extra_objs in case
8164 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8166 PR tree-optimization/81835
8167 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8168 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8169 not depend on the phi.
8171 2017-08-16 Alan Modra <amodra@gmail.com>
8173 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8176 2017-08-16 Alan Modra <amodra@gmail.com>
8178 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8179 (save_reg_p): ..into this. Update all callers.
8180 (first_reg_to_save): Simplify.
8182 2017-08-16 Alan Modra <amodra@gmail.com>
8184 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8187 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8191 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8192 constant count if that count is less than 32.
8194 2017-08-15 Nathan Sidwell <nathan@acm.org>
8196 * gcc.c (execute): Emit friendlier message if inferior is killed
8197 by an external cause.
8199 2017-08-15 Richard Biener <rguenther@suse.de>
8201 PR tree-optimization/81790
8202 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8203 CONSTRUCTORs from simplifying and VN.
8205 2017-08-14 Martin Sebor <msebor@redhat.com>
8207 * builtin-attrs.def: Add comments.
8209 2017-08-14 Martin Sebor <msebor@redhat.com>
8212 * doc/extend.texi (attribute nonstring): Document new attribute.
8214 2017-08-14 Martin Sebor <msebor@redhat.com>
8217 * tree-diagnostic.c (default_tree_printer): Handle %G.
8218 * gimple-pretty-print.h (percent_G_format): Declare new function.
8219 * gimple-pretty-print.c (percent_G_format): Define.
8220 * tree-pretty-print.c (percent_K_format): Add argument.
8222 2017-08-14 Martin Sebor <msebor@redhat.com>
8224 PR translation/79998
8225 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8226 Remove a stray space.
8228 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8231 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8232 (*iordi_1_bts): Ditto.
8233 (*xordi_1_btc): Ditto.
8235 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8238 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8239 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8241 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8242 quoted strings, and make more translator-friendly.
8243 (darwin_rs6000_override_options): Likewise.
8244 (rs6000_option_override_internal): Likewise.
8245 (rs6000_return_in_memory): Fix overlong line.
8246 (init_cmulative_args): Use quoted strings, and make more
8247 translator-friendly.
8248 (rs6000_pass_by_reference): Fix overlong line.
8249 (def_builtin): Use quoted strings.
8250 (altivec_expand_predicate_builtin): Use quoted strings, and make
8251 more translator-friendly.
8252 (htm_expand_builtin): Use quoted strings.
8253 (cpu_expand_builtin): Use quoted strings, and make more
8254 translator-friendly.
8255 (altivec_expand_builtin): Likewise.
8256 (paired_expand_predicate_builtin): Likewise.
8257 (rs6000_invalid_builtin): Likewise.
8258 (builtin_function_type): Use quoted strings.
8259 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8260 more translator-friendly.
8261 (rs6000_trampoline_init): Likewise.
8262 (rs6000_handle_altivec_attribute): Likewise.
8263 (rs6000_inner_target_options): Use quoted strings.
8264 (rs6000_disable_incompatible_switches): Likewise.
8265 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8266 strings, and make more translator-friendly.
8267 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8269 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8271 PR tree-optimization/81799
8272 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8273 cond_expr to simple gimple operand.
8275 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8278 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8280 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8284 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8285 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8286 TARGET_GASISR_PROLOGUES.
8287 * config/avr/avr.c (avr_option_override): Same.
8288 (avr_pass_pre_proep::execute): Same.
8290 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8293 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8294 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8296 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8298 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8299 define_insn_and_split. Split to a memory load from 0 in
8300 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8301 using PTR mode iterator.
8302 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8303 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8304 (*add_tp_<mode>): Redefine as define_insn_and_split.
8305 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8306 address space. Merge with *add_tp_x32 using PTR mode iterator.
8307 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8308 Split to an add with a memory load from 0 in
8309 DEFAULT_TLS_SEG_REG address space.
8311 2017-08-12 Andrew Pinski <apinski@cavium.com>
8313 * config/aarch64/aarch64-option-extensions.def (rdma):
8314 Fix feature string to what Linux prints out in /proc/cpuinfo.
8316 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8319 * dwarf2out.c (modified_type_die): For C typedef types that have
8320 an ultimate origin, process the ultimate origin instead of the
8322 (gen_typedef_die): Assert that input DECLs have no ultimate
8324 (gen_type_die_with_usage): For typedef variants that have an
8325 ultimate origin, just call gen_decl_die on the original DECL.
8326 (process_scope_var): Avoid creating DIEs for local typedefs and
8327 concrete static variables.
8329 2017-08-12 Alan Modra <amodra@gmail.com>
8333 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8334 match gnu-user.h startfile.
8335 (ENDFILE_LINUX_SPEC): Similarly.
8337 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8340 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8342 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8344 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8345 * config/aarch64/aarch64.md (mov<mode>): Change.
8346 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8347 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8348 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8350 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8352 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8353 for storage order barriers.
8355 2017-08-11 Martin Liska <mliska@suse.cz>
8358 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8359 variables of void type.
8361 2017-08-11 Martin Liska <mliska@suse.cz>
8363 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8364 TARGET_SUPPORTS_ALIASES.
8365 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8366 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8367 (optimize_weakref): Likewise.
8368 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8369 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8370 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8372 2017-08-11 Martin Liska <mliska@suse.cz>
8375 * config/i386/i386.c (make_resolver_func): Do complete
8376 refactoring of the function.
8378 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8381 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8382 * config/i386/i386.c (ix86_stack_protect_guard): Use
8383 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8384 * doc/invoke.texi (x86 Options): Document
8385 -mstack-protector-guard-symbol= option.
8387 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8389 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8390 * config/i386/i386.c (ix86_split_stack_guard): New function.
8391 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8392 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8393 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8394 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8395 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8396 (split_stack_space_check): Call ix86_split_stack_guard.
8398 2017-08-10 Martin Sebor <msebor@redhat.com>
8400 * print-tree.c (print_node): Print location using the established
8402 Replace spaces with colons.
8403 (debug_raw, debug): Ditto.
8405 2017-08-10 Martin Sebor <msebor@redhat.com>
8408 * pretty-print.c (pp_format): Correct the handling of %s precision.
8410 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8413 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8415 (ix86_finalize_stack_frame_flags): This. Also clear
8416 frame_pointer_needed if -fno-omit-frame-pointer is used without
8418 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8419 with ix86_finalize_stack_frame_flags.
8420 (ix86_expand_epilogue): Likewise.
8421 (ix86_expand_split_stack_prologue): Likewise.
8422 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8424 2017-08-10 Martin Liska <mliska@suse.cz>
8427 * c-attribs.c (handle_target_attribute):
8428 Report warning for an empty string argument of target attribute.
8430 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8433 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8435 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8436 or DECL_NONLOCAL labels.
8437 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8438 or DECL_NONLOCAL labels here.
8440 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8443 to indicate when early gimple folding has been disabled.
8444 (rs6000_gimple_fold_builtin): Add debug content.
8445 (rs6000_invalid_builtin): Fix whitespace.
8446 (rs6000_expand_builtin): Fix whitespace.
8447 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8449 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8452 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8453 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8455 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8457 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8458 (qdf24xx): Likewise.
8459 * config/aarch64/aarch64.md: Include falkor.md.
8460 * config/aarch64/falkor.md: New.
8462 2017-08-09 Marek Polacek <polacek@redhat.com>
8465 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8466 * diagnostic.c (emit_diagnostic): Add a comment.
8467 (emit_diagnostic_valist): New function.
8469 2017-08-09 Marek Polacek <polacek@redhat.com>
8472 * input.c (make_location): New overload.
8473 * input.h (make_location): Declare.
8475 2017-08-08 Alan Modra <amodra@gmail.com>
8476 H.J. Lu <hongjiu.lu@intel.com>
8479 * gcc.c (NO_PIE_SPEC): Delete.
8480 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8482 (LINK_PIE_SPEC): ..to here.
8483 (LINK_COMMAND_SPEC): Support -no-pie.
8484 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8485 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8487 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8488 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8489 (ENDFILE_CRTEND_SPEC): Similarly.
8491 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8494 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8495 (mstack-protector-guard-offset=): Ditto.
8496 * config/i386/i386.c (ix86_option_override): Handle
8497 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8499 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8500 ix86_stack_protect_guard_offset variables.
8501 (TARGET_STACK_PROTECT_GUARD): Always define.
8502 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8503 and -mstack-protector-guard-offset= options.
8505 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8507 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8508 boundary case for the last candidate.
8510 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8512 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8513 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8515 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8518 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8519 * config/aarch64/aarch64-builtins.c
8520 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8521 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8522 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8524 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8525 Andrew Pinski <pinskia@gmail.com>
8528 * internal-fn.def (XORSIGN): New.
8529 * optabs.def (xorsign_optab): New.
8530 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8531 (convert_expand_mult_copysign): New.
8532 (pass_optimize_widening_mul::execute): Call
8533 convert_expand_mult_copysign.
8535 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8537 PR tree-optimization/81354
8538 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8539 Insert on edges rather than explicitly creating landing pads.
8540 (analyze_candidates_and_replace): Commit edge inserts.
8542 2017-08-08 Richard Biener <rguenther@suse.de>
8545 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8546 (expand_simple_operations): Also look through ADDR_EXPRs with
8547 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8549 2017-08-08 Richard Biener <rguenther@suse.de>
8551 PR tree-optimization/81723
8552 * tree-vect-slp.c (struct bst_traits): New hash traits.
8553 (bst_fail): New global.
8554 (vect_build_slp_tree_2): New worker, split out from ...
8555 (vect_build_slp_tree): ... this now wrapping it with using
8556 bst_fail set to cache SLP tree build fails. Properly handle
8558 (vect_analyze_slp_instance): Allocate and free bst_fail.
8560 2017-08-08 Martin Liska <mliska@suse.cz>
8563 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8564 LABEL_DECLs that can be from a different function.
8566 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8568 PR tree-optimization/81744
8569 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8570 loop's number of iterations.
8572 2017-08-08 Martin Liska <mliska@suse.cz>
8574 * asan.c: Include header files.
8575 * attribs.c (build_decl_attribute_variant): New function moved
8577 (build_type_attribute_qual_variant): Likewise.
8578 (cmp_attrib_identifiers): Likewise.
8579 (simple_cst_list_equal): Likewise.
8580 (omp_declare_simd_clauses_equal): Likewise.
8581 (attribute_value_equal): Likewise.
8582 (comp_type_attributes): Likewise.
8583 (build_type_attribute_variant): Likewise.
8584 (lookup_ident_attribute): Likewise.
8585 (remove_attribute): Likewise.
8586 (merge_attributes): Likewise.
8587 (merge_type_attributes): Likewise.
8588 (merge_decl_attributes): Likewise.
8589 (merge_dllimport_decl_attributes): Likewise.
8590 (handle_dll_attribute): Likewise.
8591 (attribute_list_equal): Likewise.
8592 (attribute_list_contained): Likewise.
8593 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8594 (lookup_attribute_by_prefix): Likewise.
8595 * bb-reorder.c: Include header files.
8596 * builtins.c: Likewise.
8597 * calls.c: Likewise.
8598 * cfgexpand.c: Likewise.
8599 * cgraph.c: Likewise.
8600 * cgraphunit.c: Likewise.
8601 * convert.c: Likewise.
8602 * dwarf2out.c: Likewise.
8603 * final.c: Likewise.
8604 * fold-const.c: Likewise.
8605 * function.c: Likewise.
8606 * gimple-expr.c: Likewise.
8607 * gimple-fold.c: Likewise.
8608 * gimple-pretty-print.c: Likewise.
8609 * gimple.c: Likewise.
8610 * gimplify.c: Likewise.
8611 * hsa-common.c: Likewise.
8612 * hsa-gen.c: Likewise.
8613 * internal-fn.c: Likewise.
8614 * ipa-chkp.c: Likewise.
8615 * ipa-cp.c: Likewise.
8616 * ipa-devirt.c: Likewise.
8617 * ipa-fnsummary.c: Likewise.
8618 * ipa-inline.c: Likewise.
8619 * ipa-visibility.c: Likewise.
8621 * lto-cgraph.c: Likewise.
8622 * omp-expand.c: Likewise.
8623 * omp-general.c: Likewise.
8624 * omp-low.c: Likewise.
8625 * omp-offload.c: Likewise.
8626 * omp-simd-clone.c: Likewise.
8627 * opts-global.c: Likewise.
8628 * passes.c: Likewise.
8629 * predict.c: Likewise.
8630 * sancov.c: Likewise.
8631 * sanopt.c: Likewise.
8632 * symtab.c: Likewise.
8633 * toplev.c: Likewise.
8634 * trans-mem.c: Likewise.
8635 * tree-chkp.c: Likewise.
8636 * tree-eh.c: Likewise.
8637 * tree-into-ssa.c: Likewise.
8638 * tree-object-size.c: Likewise.
8639 * tree-parloops.c: Likewise.
8640 * tree-profile.c: Likewise.
8641 * tree-ssa-ccp.c: Likewise.
8642 * tree-ssa-live.c: Likewise.
8643 * tree-ssa-loop.c: Likewise.
8644 * tree-ssa-sccvn.c: Likewise.
8645 * tree-ssa-structalias.c: Likewise.
8646 * tree-ssa.c: Likewise.
8647 * tree-streamer-in.c: Likewise.
8648 * tree-vectorizer.c: Likewise.
8649 * tree-vrp.c: Likewise.
8651 * ubsan.c: Likewise.
8652 * varasm.c: Likewise.
8653 * varpool.c: Likewise.
8654 * tree.c: Remove functions moved to attribs.[ch].
8656 * config/aarch64/aarch64.c: Add attrs.h header file.
8657 * config/alpha/alpha.c: Likewise.
8658 * config/arc/arc.c: Likewise.
8659 * config/arm/arm.c: Likewise.
8660 * config/avr/avr.c: Likewise.
8661 * config/bfin/bfin.c: Likewise.
8662 * config/c6x/c6x.c: Likewise.
8663 * config/cr16/cr16.c: Likewise.
8664 * config/cris/cris.c: Likewise.
8665 * config/darwin.c: Likewise.
8666 * config/epiphany/epiphany.c: Likewise.
8667 * config/fr30/fr30.c: Likewise.
8668 * config/frv/frv.c: Likewise.
8669 * config/ft32/ft32.c: Likewise.
8670 * config/h8300/h8300.c: Likewise.
8671 * config/i386/winnt.c: Likewise.
8672 * config/ia64/ia64.c: Likewise.
8673 * config/iq2000/iq2000.c: Likewise.
8674 * config/lm32/lm32.c: Likewise.
8675 * config/m32c/m32c.c: Likewise.
8676 * config/m32r/m32r.c: Likewise.
8677 * config/m68k/m68k.c: Likewise.
8678 * config/mcore/mcore.c: Likewise.
8679 * config/microblaze/microblaze.c: Likewise.
8680 * config/mips/mips.c: Likewise.
8681 * config/mmix/mmix.c: Likewise.
8682 * config/mn10300/mn10300.c: Likewise.
8683 * config/moxie/moxie.c: Likewise.
8684 * config/msp430/msp430.c: Likewise.
8685 * config/nds32/nds32-isr.c: Likewise.
8686 * config/nds32/nds32.c: Likewise.
8687 * config/nios2/nios2.c: Likewise.
8688 * config/nvptx/nvptx.c: Likewise.
8689 * config/pa/pa.c: Likewise.
8690 * config/pdp11/pdp11.c: Likewise.
8691 * config/powerpcspe/powerpcspe.c: Likewise.
8692 * config/riscv/riscv.c: Likewise.
8693 * config/rl78/rl78.c: Likewise.
8694 * config/rx/rx.c: Likewise.
8695 * config/s390/s390.c: Likewise.
8696 * config/sh/sh.c: Likewise.
8697 * config/sol2.c: Likewise.
8698 * config/sparc/sparc.c: Likewise.
8699 * config/spu/spu.c: Likewise.
8700 * config/stormy16/stormy16.c: Likewise.
8701 * config/tilegx/tilegx.c: Likewise.
8702 * config/tilepro/tilepro.c: Likewise.
8703 * config/v850/v850.c: Likewise.
8704 * config/vax/vax.c: Likewise.
8705 * config/visium/visium.c: Likewise.
8706 * config/xtensa/xtensa.c: Likewise.
8708 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8711 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8712 constraints since the -mupper-regs-* switches have been
8714 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8715 into a vector from a double word element that was extracted from
8716 another vector, and eliminate extra XXPERMDI instructions.
8717 (vsx_concat_<mode>_2): Likewise.
8718 (vsx_concat_<mode>_3): Likewise.
8719 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8720 concat to allow optimizing inserts from previous extracts.
8722 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8724 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8725 memory reference to a SSP offset in TLS address space.
8726 (ix86_print_operand) <case '@'>: Remove.
8727 (ix86_print_operand_punct_valid_p): Remove '@' code.
8728 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8730 (stack_tls_protect_set_<mode>): Remove.
8731 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8732 (stack_tls_protect_test_<mode>): Remove.
8733 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8735 2017-08-07 Olivier Hainque <hainque@adacore.com>
8738 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8740 2017-08-07 Douglas Rupp <rupp@adacore.com>
8742 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8743 variable was referenced as multidir in command.
8745 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8748 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8751 2017-08-07 Martin Liska <mliska@suse.cz>
8753 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8754 * config/rl78/rl78.c: Add include of attribs.h.
8755 * config/sh/sh.c: Likewise.
8756 * config/v850/v850.c: Likewise.
8758 2017-08-07 Tom de Vries <tom@codesourcery.com>
8761 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8763 2017-08-07 Martin Liska <mliska@suse.cz>
8765 * config/mips/mips.c: Include attribs.h.
8767 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8770 * doc/invoke.texi: Document change in behvaior for -Ofast for
8773 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8775 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8777 (aarch64_pop_regs): Likewise.
8778 (aarch64_gen_load_pair): Likewise.
8779 (aarch64_save_callee_saves): Likewise.
8780 (aarch64_restore_callee_saves): Likewise.
8782 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8784 * config/i386/i386.c: Revert the last change.
8786 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8789 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8791 (ix86_finalize_stack_frame_flags): This. Also clear
8792 frame_pointer_needed if -fno-omit-frame-pointer is used without
8794 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8795 with ix86_finalize_stack_frame_flags.
8796 (ix86_expand_epilogue): Likewise.
8797 (ix86_expand_split_stack_prologue): Likewise.
8799 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8802 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8803 to P_AES for Westmere.
8805 2017-08-07 Jonathan Yong <10walls@gmail.com>
8807 * config/i386/mingw.opt (fset-stack-executable): Removed.
8808 * config/i386/cygming.opt (fset-stack-executable): Moved
8810 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8812 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8814 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8816 2017-08-07 Marek Polacek <polacek@redhat.com>
8819 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8821 2017-08-07 Martin Liska <mliska@suse.cz>
8823 * attribs.h (canonicalize_attr_name): New function.
8824 (cmp_attribs): Move from c-format.c and adjusted.
8825 (is_attribute_p): Moved from tree.h.
8826 * tree-inline.c: Add new includes.
8827 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8828 (private_is_attribute_p): Remove.
8829 (private_lookup_attribute): Likewise.
8830 (private_lookup_attribute_by_prefix): Simplify.
8831 (remove_attribute): Use is_attribute_p.
8832 * tree.h: Remove removed declarations.
8834 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8837 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8838 instead of computing it in the function. Formatting fix.
8839 (expand_case): Don't rely on default_edge being the first edge,
8840 clear it if removing it, pass default_edge to
8841 emit_case_dispatch_table.
8842 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8845 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8847 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8848 insn in the function, emit NOP after the insn.
8850 2017-08-06 Tom de Vries <tom@codesourcery.com>
8852 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8855 2017-08-06 Tom de Vries <tom@codesourcery.com>
8857 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8860 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8862 PR tree-optimization/57371
8863 * match.pd: New pattern.
8865 2017-08-04 Marek Polacek <polacek@redhat.com>
8868 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8869 perform the computation in offset_int.
8871 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8873 PR tree-optimization/81136
8874 * tree-vectorizer.h: Include tree-hash-traits.h.
8875 (vec_base_alignments): New typedef.
8876 (vec_info): Add a base_alignments field.
8877 (vect_record_base_alignments): Declare.
8878 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8880 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8881 (create_data_ref): Add an is_conditional_in_stmt argument.
8882 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8883 the is_conditional_in_stmt field.
8884 (data_ref_loc): Add an is_conditional_in_stmt field.
8885 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8886 (find_data_references_in_stmt): Update call to create_data_ref.
8887 (graphite_find_data_references_in_stmt): Likewise.
8888 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8889 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8890 (vect_record_base_alignment): New function.
8891 (vect_record_base_alignments): Likewise.
8892 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8893 for nested statements even if we fail to compute a misalignment.
8894 Use pooled base alignments for unconditional references.
8895 (vect_find_same_alignment_drs): Compare base addresses instead
8897 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8898 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8900 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8902 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8903 Add an explicit name for the enum. Use auto_vec for slp_instances
8905 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8907 (_bb_vec_info): Add a constructor and destructor.
8908 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8909 (destroy_loop_vec_info): Delete.
8910 (vect_destroy_datarefs): Likewise.
8911 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8912 (vec_info::vec_info): New function.
8913 (vec_info::~vec_info): Likewise.
8914 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8915 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8916 destroy_loop_vec_info.
8917 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8918 (_loop_vec_info::_loop_vec_info): ...this.
8919 (destroy_loop_vec_info): Replace with...
8920 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8921 the stmt_vec_infos. Leave handling of vec_info information to its
8922 destructor. Remove explicit vector releases.
8923 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8924 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8925 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8926 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8927 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8928 (destroy_bb_vec_info): Replace with...
8929 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8930 information to its destructor.
8931 (vect_slp_analyze_bb_1): Use new and delete instead of
8932 new_bb_vec_info and destroy_bb_vec_info.
8933 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8936 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8938 * tree-data-ref.h (subscript): Add access_fn field.
8939 (data_dependence_relation): Add could_be_independent_p.
8940 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8941 (same_access_functions): Move to tree-data-ref.c.
8942 * tree-data-ref.c (ref_contains_union_access_p): New function.
8943 (access_fn_component_p): Likewise.
8944 (access_fn_components_comparable_p): Likewise.
8945 (dr_analyze_indices): Add a reference to access_fn_component_p.
8946 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8948 (constant_access_functions): Likewise.
8949 (add_other_self_distances): Likewise.
8950 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8951 (initialize_data_dependence_relation): Use XCNEW and remove
8952 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8953 of access functions that have the same type. Allow the
8954 subsequence to end with different bases in some circumstances.
8955 Record the chosen access functions in SUB_ACCESS_FN.
8956 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8957 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8958 (subscript_dependence_tester_1): Likewise dra and drb.
8959 (build_classic_dist_vector): Update calls accordingly.
8960 (subscript_dependence_tester): Likewise.
8961 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8962 DDR_COULD_BE_INDEPENDENT_P.
8963 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8964 comp_alias_ddrs instead of may_alias_ddrs.
8965 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8967 (vect_analyze_data_ref_dependence): Use it if
8968 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8969 distance vectors if that fails.
8970 (dependence_distance_ge_vf): New function.
8971 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8972 LOOP_VINFO_MAY_ALIAS_DDRS.
8974 2017-08-04 Richard Biener <rguenther@suse.de>
8977 * fold-const.c (fold_binary_loc): Properly restrict
8978 minus_var0 && minus_var1 case when associating undefined overflow
8981 2017-08-04 Tom de Vries <tom@codesourcery.com>
8983 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8985 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8987 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8988 Don't start diagnostic messages with a capital letter.
8989 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8991 (rs6000_invalid_builtin): Likewise.
8992 (rs6000_trampoline_init): Likewise.
8994 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8997 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8998 after setting changeable df flags.
9000 2017-08-03 Richard Biener <rguenther@suse.de>
9002 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9003 up if the use is in USE - X.
9005 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9007 * toplev.c (dumpfile.h): New include.
9008 (internal_error_reentered): New static function. Use it...
9009 (internal_error_function): ...here to handle reentered internal_error.
9011 2017-08-03 Richard Biener <rguenther@suse.de>
9014 * fold-const.c (split_tree): Add minus_var and minus_con
9015 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9016 here but always use minus_*.
9017 (associate_trees): Assert we never associate with MINUS_EXPR
9018 and NULL first operand. Do not recurse for PLUS_EXPR operands
9019 when associating as MINUS_EXPR either.
9020 (fold_binary_loc): Track minus_var and minus_con.
9022 2017-08-03 Tom de Vries <tom@codesourcery.com>
9025 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9026 ACCEL_COMPILER, apply finish_options on
9027 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9029 2017-08-03 Tom de Vries <tom@codesourcery.com>
9032 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9033 function_entry_patch_area_size > 0.
9035 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9038 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9039 instead of 10??LU, perform unit multiplication in wide_int,
9040 don't change alloc_object_size_limit if the limit is larger
9043 PR tree-optimization/81655
9044 PR tree-optimization/81588
9045 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9046 the case when ranges[i].low and high are 1 for unsigned type with
9050 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9051 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9053 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9055 * tree-vrp.h: Add include guard.
9057 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9060 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9061 (ud2): New insn pattern.
9062 * config/i386/i386.c (ix86_expand_epilogue):
9063 For naked functions, generate ud2 instead of trap insn.
9065 2017-08-02 Marek Polacek <polacek@redhat.com>
9068 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9070 2017-08-02 Tom de Vries <tom@codesourcery.com>
9071 Cesar Philippidis <cesar@codesourcery.com>
9073 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9074 Add missing edge probabilities.
9076 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9078 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9081 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9084 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9085 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9086 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9087 prologue_seq sequences - if any.
9089 2017-08-02 Richard Biener <rguenther@suse.de>
9091 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9092 via vectors if supported, integer extracts via punning if supported
9093 or otherwise vector extracts.
9095 2017-08-02 Richard Biener <rguenther@suse.de>
9097 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9099 (bitmap_insert_into_set): ... this.
9101 2017-08-02 Richard Biener <rguenther@suse.de>
9103 PR tree-optimization/81633
9105 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9107 PR tree-optimization/71752
9108 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9110 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9112 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9113 (machine_function::call_ms2sysv_pad_out): Remove field.
9114 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9115 (ix86_compute_frame_layout): Likewise.
9117 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9120 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9121 attribute with interrupt attribute.
9123 2017-08-01 Andrew Pinski <apinski@cavium.com>
9125 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9126 BIT_INSERT_EXPR's operand 1
9127 to see if the types precision matches.
9129 2017-08-01 Martin Liska <mliska@suse.cz>
9132 * builtins.c (expand_builtin_memcpy_args): Remove.
9133 (expand_builtin_memcpy): Call newly added function
9134 expand_builtin_memory_copy_args.
9135 (expand_builtin_memcpy_with_bounds): Likewise.
9136 (expand_builtin_mempcpy): Remove last argument.
9137 (expand_builtin_mempcpy_with_bounds): Likewise.
9138 (expand_builtin_memory_copy_args): New function created from
9139 expand_builtin_mempcpy_args with small modifications.
9140 (expand_builtin_mempcpy_args): Remove.
9141 (expand_builtin_stpcpy): Remove unused argument.
9142 (expand_builtin): Likewise.
9143 (expand_builtin_with_bounds): Likewise.
9145 2017-08-01 Martin Liska <mliska@suse.cz>
9148 Make mempcpy more optimal (PR middle-end/70140).
9150 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9153 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9154 __builtin_vec_cmpne verify both arguments are compatible vectors
9155 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9156 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9157 move computation of aligned to after checking the argument types.
9161 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9162 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9165 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9166 Doug Rupp <rupp@adacore.com>
9167 Olivier Hainque <hainque@adacore.com>
9169 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9170 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9172 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9173 for TARGET_OS_CPP_BUILTIN.
9174 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9175 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9177 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9178 passing required abi options to the assembler for EABI configurations.
9179 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9180 of .text.hot and .text.unlikely sections for kernel modules when
9181 using ARM style exceptions.
9182 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9183 options. Add EXTRA_CC1_SPEC.
9184 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9186 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9188 (ARM_TARGET2_DWARF_FORMAT): Define.
9189 * config/arm/t-vxworks: Adjust multilib control to removal of the
9190 Diab command line options.
9192 2017-08-01 Martin Liska <mliska@suse.cz>
9194 PR gcov-profile/81561
9195 * gcov.c (unblock): Make unblocking safe as we need to preserve
9196 index correspondence of blocks and block_lists.
9198 2017-08-01 Richard Biener <rguenther@suse.de>
9200 PR tree-optimization/81181
9201 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9202 (compute_antic): ... end of iteration here.
9204 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9206 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9207 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9208 (ftree-slp-vectorize): Likewise.
9209 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9210 can no longer be set independent of flag_tree_loop_vectorize.
9211 * omp-general.c (emp_max_vf): Likewise.
9212 * opts.c (enable_fdo_optimizations): Remove references to
9213 flag_tree_vectorize, these are now implicit.
9214 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9215 and leave it for the options machinery.
9217 2017-08-01 Martin Liska <mliska@suse.cz>
9220 * builtins.c (expand_builtin_memcpy_args): Remove.
9221 (expand_builtin_memcpy): Call newly added function
9222 expand_builtin_memory_copy_args.
9223 (expand_builtin_memcpy_with_bounds): Likewise.
9224 (expand_builtin_mempcpy): Remove last argument.
9225 (expand_builtin_mempcpy_with_bounds): Likewise.
9226 (expand_builtin_memory_copy_args): New function created from
9227 expand_builtin_mempcpy_args with small modifications.
9228 (expand_builtin_mempcpy_args): Remove.
9229 (expand_builtin_stpcpy): Remove unused argument.
9230 (expand_builtin): Likewise.
9231 (expand_builtin_with_bounds): Likewise.
9233 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9236 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9237 print "ds:" only for immediates in generic address space.
9239 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9242 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9243 (ix86_function_ok_for_sibcall): Return false for naked functions.
9245 2017-08-01 Richard Biener <rguenther@suse.de>
9247 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9248 (compute_antic): Seed worklist with exit block predecessors.
9249 * cfganal.c (dfs_find_deadend): For a cycle return the source
9250 of the edge closing it.
9252 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9254 * config/aarch64/aarch64.c
9255 (aarch64_can_const_movi_rtx_p): Move 0 check.
9257 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9259 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9260 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9262 * match.pd: Ditto in address comparison pattern.
9264 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9266 PR tree-optimization/81627
9267 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9268 closed ssa form for store-store chain.
9270 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9272 PR tree-optimization/81620
9273 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9274 for store-store chain.
9276 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9278 PR tree-optimization/81588
9279 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9280 ranges[i].in_p, invert comparison code ccode. For >/>=,
9281 swap rhs1 and rhs2 and comparison code unconditionally,
9282 for </<= don't do that. Don't swap rhs1/rhs2 again if
9283 ranges[i].in_p, instead invert comparison code ccode if
9284 opcode or oe->rank is BIT_IOR_EXPR.
9287 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9288 a direct optab to conversion optab.
9289 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9290 with GET_MODE_INNER as last argument instead of optab_handler.
9291 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9292 vector extraction if possible and optab is available.
9293 * expr.c (store_constructor): Use convert_optab_handler instead
9294 of optab_handler. Use vector initialization from smaller
9295 vectors if possible and optab is available.
9296 * tree-vect-stmts.c (vectorizable_load): Likewise.
9297 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9299 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9300 of vec_init from half-sized vectors with the same element mode.
9301 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9302 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9303 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9304 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9305 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9306 after mode in gen_vec_extract* calls.
9307 (vec_extract<mode>): Renamed to ...
9308 (vec_extract<mode><ssescalarmodelower>): ... this.
9309 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9310 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9311 element mode after mode in gen_vec_init* calls.
9312 (VEC_INIT_HALF_MODE): New mode iterator.
9313 (vec_init<mode>): Renamed to ...
9314 (vec_init<mode><ssescalarmodelower>): ... this.
9315 (vec_init<mode><ssehalfvecmodelower>): New expander.
9316 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9317 (vec_extractv2sfsf): ... this.
9318 (vec_initv2sf): Renamed to ...
9319 (vec_initv2sfsf): ... this.
9320 (vec_extractv2si): Renamed to ...
9321 (vec_extractv2sisi): ... this.
9322 (vec_initv2si): Renamed to ...
9323 (vec_initv2sisi): ... this.
9324 (vec_extractv4hi): Renamed to ...
9325 (vec_extractv4hihi): ... this.
9326 (vec_initv4hi): Renamed to ...
9327 (vec_initv4hihi): ... this.
9328 (vec_extractv8qi): Renamed to ...
9329 (vec_extractv8qiqi): ... this.
9330 (vec_initv8qi): Renamed to ...
9331 (vec_initv8qiqi): ... this.
9332 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9333 (vec_init<mode>): Renamed to ...
9334 (vec_init<mode><VEC_base_l>): ... this.
9335 (vec_extract<mode>): Renamed to ...
9336 (vec_extract<mode><VEC_base_l>): ... this.
9337 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9338 (vec_initv2sfsf): ... this.
9339 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9340 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9341 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9342 element mode after mode in gen_vec_init* calls.
9343 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9344 (vec_init<mode><Vel>): ... this.
9345 (vec_extract<mode>): Renamed to ...
9346 (vec_extract<mode><Vel>): ... this.
9347 * config/aarch64/iterators.md (Vel): New mode attribute.
9348 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9349 Add element mode after mode in gen_vec_extract* calls.
9350 * config/s390/vector.md (non_vec_l): New mode attribute.
9351 (vec_extract<mode>): Renamed to ...
9352 (vec_extract<mode><non_vec_l>): ... this.
9353 (vec_init<mode>): Renamed to ...
9354 (vec_init<mode><non_vec_l>): ... this.
9355 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9356 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9358 * config/arm/iterators.md (V_elem_l): New mode attribute.
9359 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9360 (vec_extract<mode><V_elem_l>): ... this.
9361 (vec_extractv2di): Renamed to ...
9362 (vec_extractv2didi): ... this.
9363 (vec_init<mode>): Renamed to ...
9364 (vec_init<mode><V_elem_l>): ... this.
9365 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9366 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9367 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9368 Add element mode after gen_vec_extract* calls.
9369 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9370 (vec_init<mode><unitmode>): ... this.
9371 (vec_extract<mode>): Renamed to ...
9372 (vec_extract<mode><unitmode>): ... this.
9373 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9374 (vec_init<mode><unitmode>): ... this.
9375 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9376 (vec_initv2sfsf): ... this.
9377 (vec_extractv2sf): Renamed to ...
9378 (vec_extractv2sfsf): ... this.
9379 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9380 Add element mode after gen_vec_extract* calls.
9381 * config/mips/mips.md (unitmode): New mode iterator.
9382 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9383 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9384 * config/spu/spu.md (inner_l): New mode attribute.
9385 (vec_init<mode>): Renamed to ...
9386 (vec_init<mode><inner_l>): ... this.
9387 (vec_extract<mode>): Renamed to ...
9388 (vec_extract<mode><inner_l>): ... this.
9389 * config/sparc/sparc.md (veltmode): New mode iterator.
9390 (vec_init<VMALL:mode>): Renamed to ...
9391 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9392 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9393 (vec_initv2sisi): ... this.
9394 (vec_initv2sf): Renamed to ...
9395 (vec_initv2sfsf): ... this.
9396 (vec_extractv2sf): Renamed to ...
9397 (vec_extractv2sfsf): ... this.
9398 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9399 (vec_init<mode>): Renamed to ...
9400 (vec_init<mode><VEC_base_l>): ... this.
9401 (vec_extract<mode>): Renamed to ...
9402 (vec_extract<mode><VEC_base_l>): ... this.
9403 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9404 (vec_initv2sfsf): ... this.
9405 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9406 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9407 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9408 gen_vec_init* calls.
9410 2017-08-01 Richard Biener <rguenther@suse.de>
9412 PR tree-optimization/81297
9413 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9414 TREE_OVERFLOW from INTEGER_CSTs.
9416 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9418 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9420 2017-07-31 Carl Love <cel@us.ibm.com>
9422 * config/rs6000/rs6000-c: Add support for built-in functions
9423 vector signed char vec_xl_be (signed long long, signed char *);
9424 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9425 vector signed int vec_xl_be (signed long long, signed int *);
9426 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9427 vector signed long long vec_xl_be (signed long long, signed long long *);
9428 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9429 vector signed short vec_xl_be (signed long long, signed short *);
9430 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9431 vector double vec_xl_be (signed long long, double *);
9432 vector float vec_xl_be (signed long long, float *);
9433 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9434 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9435 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9437 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9438 (altivec_expand_builtin): Add switch statement to call
9439 altivec_expand_xl_be for each builtin.
9440 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9441 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9442 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9443 __builtin_vsx_le_be_v16qi.
9444 * doc/extend.texi: Update the built-in documentation file for the
9445 new built-in functions.
9447 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9450 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9452 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9454 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9456 * config.gcc: Add z14.
9457 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9458 CPU model numbers for z13s and z14.
9459 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9461 * config/s390/s390-opts.h (enum processor_type): Rename
9462 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9463 * config/s390/s390.c (processor_table): Add field for CPU name to
9464 be passed to Binutils.
9465 (s390_asm_output_machine_for_arch): Use the new field in
9466 processor_table for Binutils.
9467 (s390_expand_builtin): Replace arch12 with z14.
9468 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9469 (s390_get_sched_attrmask): Likewise.
9470 (s390_get_unit_mask): Likewise.
9471 * config/s390/s390.opt: Add z14 to processor_type enum.
9473 2017-07-31 Martin Jambor <mjambor@suse.cz>
9476 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9477 regardless of optimization level.
9479 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9480 Martin Liska <mliska@suse.cz>
9482 * predict.def: Remove old comment and adjust probability.
9483 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9486 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9489 * config/i386/i386.c (ix86_function_naked): New function.
9490 (ix86_can_use_return_insn_p): Return false for naked functions.
9491 (ix86_expand_prologue): Skip prologue for naked functions.
9492 (ix86_expand_epilogue): Skip epilogue for naked functions
9493 and emit trap instruction.
9494 (ix86_warn_func_return): New function.
9495 (ix86_attribute_table): Add "naked" attribute specification.
9496 (TARGET_WARN_FUNC_RETURN): Define.
9497 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9499 2017-07-31 Martin Liska <mliska@suse.cz>
9501 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9502 (dump_gimple_bb_header): Always dump BB info.
9503 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9505 2017-07-31 Martin Liska <mliska@suse.cz>
9508 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9509 also with current_function_decl non-null equality.
9511 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9514 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9515 change type to the element type, instead add eltype variable and
9516 use it where we are interested in the element type.
9518 PR tree-optimization/81603
9519 * ipa-polymorphic-call.c
9520 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9521 offset arithmetic in offset_int, bail out if the resulting bit offset
9522 doesn't fit into shwi.
9524 2017-07-31 Martin Liska <mliska@suse.cz>
9526 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9527 (gimplify_save_expr): Fix comment.
9529 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9532 * config/i386/i386.c (ix86_function_arg): Update arguments for
9534 (ix86_compute_frame_layout): Set the initial stack offset to
9535 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9536 INCOMING_FRAME_SP_OFFSET.
9537 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9538 stack before exception handler returns.
9539 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9540 the 'ERROR_CODE' for exception handler.
9542 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9544 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9545 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9546 (ASM_OUTPUT_REG_POP): Ditto.
9547 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9548 instead of asm_fprintf to output pure string.
9550 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9552 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9553 to imported_module_or_decl hook.
9554 (debug_nothing_tree_tree_tree_bool): Remove.
9555 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9556 * debug.c (do_nothing_debug_hooks): Use
9557 debug_nothing_tree_tree_tree_bool_bool instead of
9558 debug_nothing_tree_tree_tree_bool.
9559 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9560 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9561 * sdbout.c (sdb_debug_hooks): Likewise.
9562 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9563 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9565 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9566 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9567 attribute, don't add anything.
9569 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9571 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9572 (fold_build2_stat_loc): Likewise.
9573 (fold_build3_stat_loc): Likewise.
9574 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9575 (fold_build1_loc): Remove macro.
9576 (fold_build2_loc): Likewise.
9577 (fold_build3_loc): Likewise.
9579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9581 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9582 (gimple_build_debug_bind_source_stat): Likewise.
9583 * gimple.h (gimple_build_debug_bind): Remove macro.
9584 (gimple_build_debug_bind_source): Likewise.
9586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9588 * bitmap.c (bitmap_alloc): Adjust.
9589 (bitmap_gc_alloc): Likewise.
9590 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9592 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9594 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9595 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9596 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9597 (bitmap_gc_alloc_stat): Likewise.
9598 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9600 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9602 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9603 * rtl.h (shallow_copy_rtx): Remove macro.
9605 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9607 * emit-rtl.c (gen_raw_REG): Adjust.
9608 * gengenrtl.c (gendef): Likewise.
9609 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9610 * rtl.h (rtx_alloc): Remove macro.
9612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9614 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9615 (build_tree_list_stat): Likewise.
9616 * tree.h (build_tree_list): Remove macro.
9617 (build_tree_list_vec): Likewise.
9619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9621 * tree.c (make_vector_stat): Remove _stat from name.
9622 (build_vector_stat): Likewise.
9623 * tree.h (make_vector_stat): Remove macro.
9624 (build_vector_stat): Likewise.
9626 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9628 * tree.h (build_var_debug_value): Remove prototype.
9630 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9632 * tree.c (tree_cons_stat): Remove _stat from name.
9633 * tree.h (tree_cons): Remove macro.
9635 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9637 * tree.c (build_vl_exp_stat): Remove _stat from name.
9638 * tree.h (build_vl_exp): Remove macro.
9640 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9642 * tree.c (build_decl_stat): Remove _stat from name.
9643 * tree.h (build_decl): Remove macro.
9645 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9647 * gimple.c (gimple_build_with_ops_stat): Adjust.
9648 (gimple_alloc_stat): Remove _stat from name.
9649 * gimple.h (gimple_alloc): Remove macro.
9651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9653 * tree.c (make_tree_vec_stat): Remove _stat from name.
9654 (grow_tree_vec_stat): Likewise.
9655 * tree.h (make_tree_vec_stat): Adjust prototype.
9656 (grow_tree_vec_stat): Likewise.
9657 (make_tree_vec): Remove macro.
9658 (grow_tree_vec): Likewise.
9660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9662 * fold-const.c (fold_build1_stat_loc): Adjust.
9663 (fold_build2_stat_loc): Likewise.
9664 (fold_build3_stat_loc): Likewise.
9665 * tree.c (build0_stat): Remove _stat from name.
9666 (build1_stat): Likewise.
9667 (build2_stat): Likewise.
9668 (build3_stat): Likewise.
9669 (build4_stat): Likewise.
9670 (build5_stat): Likewise.
9671 * tree.h (build1_loc): Remove macro, and rename _stat function
9673 (build2_loc): Likewise.
9674 (build3_loc): Likewise.
9675 (build4_loc): Likewise.
9676 (build5_loc): Likewise.
9678 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9680 * tree.c (make_int_cst_stat): Remove _stat from name.
9681 * tree.h (make_int_cst_stat): Adjust prototype.
9682 (make_int_cst): Remove macro.
9684 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9686 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9687 * tree.h (make_tree_binfo_stat): Adjust prototype.
9688 (make_tree_binfo): Remove.
9690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9692 * tree.c (copy_node_stat): Rename to copy_node.
9693 (build_distinct_type_copy): Adjust.
9694 * tree.h (copy_node_stat): Adjust prototype.
9695 (copy_node): Remove macro.
9697 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9699 * tree.c (make_node_stat): rename to make_node.
9700 (build_tree_list_stat): Adjust.
9701 (build0_stat): Likewise.
9702 (build2_stat): Likewise.
9703 (build3_stat): Likewise.
9704 (build4_stat): Likewise.
9705 (build5_stat): Likewise.
9706 (build_decl_stat): Likewise.
9707 * tree.h (make_node_stat): Adjust prototype.
9708 (make_node): remove macro.
9710 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9712 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9713 (PPC_FEATURE2_SCV): Likewise.
9714 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9716 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9718 * config/aarch64/aarch64.c
9719 (aarch64_internal_mov_immediate): Add new special pattern.
9720 * config/aarch64/aarch64.md (*movdi_aarch64):
9721 Add reg/32bit const mov case.
9723 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9724 Richard Sandiford <richard.sandiford@linaro.org>
9726 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9727 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9728 Add integer and movi cases.
9729 (movi-split-hf-df-sf split, fp16): New.
9730 (enabled): Added TARGET_FP_F16INST.
9731 * config/aarch64/iterators.md (GPF_HF): New.
9732 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9734 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9736 * config/aarch64/aarch64.c
9737 (aarch64_simd_container_mode): Add prototype.
9738 (aarch64_expand_mov_immediate): Add HI support.
9739 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9740 (aarch64_can_const_movi_rtx_p): New.
9741 (aarch64_preferred_reload_class):
9742 Remove restrictions of using FP registers for certain SIMD operations.
9743 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9744 (aarch64_valid_floating_const): Add integer move validation.
9745 (aarch64_simd_imm_scalar_p): Remove.
9746 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9747 (aarch64_legitimate_constant_p): Expand list of supported cases.
9748 * config/aarch64/aarch64-protos.h
9749 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9750 (aarch64_reinterpret_float_as_int): New.
9751 (aarch64_simd_imm_scalar_p): Remove.
9752 * config/aarch64/constraints.md (Uvi): New.
9753 (Dd): Split into Ds and new Dd.
9754 * config/aarch64/aarch64.md (*movsi_aarch64):
9756 (*movdi_aarch64): Add SIMD mov case.
9758 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9760 * tree-predcom.c: (struct chain): Handle store-store chain in which
9761 stores for elimination only store loop invariant values.
9762 (execute_pred_commoning_chain): Ditto.
9763 (prepare_initializers_chain_store_elim): Ditto.
9764 (prepare_finalizers): Ditto.
9765 (is_inv_store_elimination_chain): New function.
9766 (initialize_root_vars_store_elim_1): New function.
9768 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9770 * tree-predcom.c: Revise general description of the pass.
9771 (enum chain_type): New enum type for store elimination.
9772 (struct chain): New field supporting store elimination.
9773 (struct component): Ditto.
9774 (dump_chain): Dump store-stores chain.
9775 (release_chain): Release resources.
9776 (split_data_refs_to_components): Compute and create component
9777 contains only stores for elimination.
9778 (get_chain_last_ref_at): New function.
9779 (make_invariant_chain): Initialization.
9780 (make_rooted_chain): Specify chain type in parameter and record it.
9781 (add_looparound_copies): Skip for store-stores chain.
9782 (determine_roots_comp): Compute type of chain and pass it to
9784 (initialize_root_vars_store_elim_2): New function.
9785 (finalize_eliminated_stores): New function.
9786 (remove_stmt): Handle store for elimination.
9787 (execute_pred_commoning_chain): Execute predictive commoning on
9789 (determine_unroll_factor): Skip unroll for store-stores chain.
9790 (prepare_initializers_chain_store_elim): New function.
9791 (prepare_initializers_chain): Hanlde store-store chain.
9792 (prepare_finalizers_chain, prepare_finalizers): New function.
9793 (tree_predictive_commoning_loop): Return integer value indicating
9794 if loop is unrolled or lcssa form is corrupted.
9795 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9797 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9799 * tree-predcom.c (initialize_root): Delete.
9800 (execute_pred_commoning_chain): Initialize root vars and replace
9801 reference of non-combined chain directly, rather than call above
9804 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9806 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9807 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9809 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9811 * tree-predcom.c (struct chain): New field init_seq.
9812 (release_chain): Release init_seq.
9813 (prepare_initializers_chain): Record intialization stmts in above
9815 (insert_init_seqs): New function.
9816 (tree_predictive_commoning_loop): Call insert_init_seqs.
9818 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9820 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9822 2017-07-28 Richard Biener <rguenther@suse.de>
9824 * match.pd: Remove superfluous :c.
9825 * genmatch.c (simplify::id): Add member.
9826 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9828 (current_id): New global.
9829 (dt_node::parent): Move from ...
9830 (dt_operand::parent): ... here. Add for_id member.
9831 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9832 (decision_tree::find_node): Relax order requirement when
9833 merging DT_TRUE nodes to ones inbetween the current simplify
9834 and the one we try to merge with. Add diagnostic whenever
9835 we need to enforce pattern order by not merging.
9836 (decision_tree::insert): Set current_id.
9837 (decision_tree::print_node): Dump parent node and for_id.
9838 (parser::last_id): Add member.
9839 (parser::push_simplify): Assign unique id.
9840 (parser::parser): Initialize last_id.
9842 2017-07-28 Martin Liska <mliska@suse.cz>
9845 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9846 gimple_build_debug_bind.
9848 2017-07-28 Richard Biener <rguenther@suse.de>
9850 PR tree-optimization/81502
9851 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9853 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9855 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9856 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9857 for BIT_FIELD_REF args.
9858 * fold-const.c (make_bit_field_ref): Likewise.
9859 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9861 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9864 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9865 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9866 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9867 Or it into SANITIZER_UNDEFINED.
9868 * ubsan.c: Include gimple-fold.h and varasm.h.
9869 (ubsan_expand_ptr_ifn): New function.
9870 (instrument_pointer_overflow): New function.
9871 (maybe_instrument_pointer_overflow): New function.
9872 (instrument_object_size): Formatting fix.
9873 (pass_ubsan::execute): Call instrument_pointer_overflow
9874 and maybe_instrument_pointer_overflow.
9875 * internal-fn.c (expand_UBSAN_PTR): New function.
9876 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9877 * sanitizer.def (__ubsan_handle_pointer_overflow,
9878 __ubsan_handle_pointer_overflow_abort): New builtins.
9879 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9880 * internal-fn.def (UBSAN_PTR): New internal function.
9881 * opts.c (sanitizer_opts): Add pointer-overflow.
9882 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9883 * fold-const.c (build_range_check): Compute pointer range check in
9884 integral type if pointer arithmetics would be needed. Formatting
9887 2017-07-28 Martin Liska <mliska@suse.cz>
9890 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9891 parameters that are of a variable-length.
9893 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9895 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9897 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9898 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9899 (CRT_CALL_STATIC_FUNCTION): Likewise.
9900 (ASM_DEFAULT_SPEC): New define.
9901 (ASM_SPEC32): Likewise.
9902 (ASM_SPEC64): Likewise.
9903 (ASM_SPEC_COMMON): Likewise.
9904 (ASM_SPEC): Likewise.
9905 (INVALID_64BIT): Likewise.
9906 (LINK_OS_DEFAULT_SPEC): Likewise.
9907 (LINK_OS_SPEC32): Likewise.
9908 (LINK_OS_SPEC64): Likewise.
9909 (POWERPC_LINUX): Likewise.
9910 (PTRDIFF_TYPE): Likewise.
9911 (RESTORE_FP_PREFIX): Likewise.
9912 (RESTORE_FP_SUFFIX): Likewise.
9913 (SAVE_FP_PREFIX): Likewise.
9914 (SAVE_FP_SUFFIX): Likewise.
9915 (SIZE_TYPE): Likewise.
9916 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9917 (TARGET_64BIT): Likewise.
9918 (TARGET_64BIT): Likewise.
9919 (TARGET_AIX): Likewise.
9920 (WCHAR_TYPE_SIZE): Likewise.
9921 (WCHAR_TYPE): Undefine.
9922 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9923 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9924 (CPP_OS_RTEMS_SPEC): Delete.
9925 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9926 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9928 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9930 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9932 PR tree-optimization/81578
9933 * tree-parloops.c (build_new_reduction): Bail out if
9934 reduction_code isn't one of the standard OpenMP reductions.
9935 Move the details printing after that decision.
9937 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9939 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9940 related to reload_in_progress.
9941 (splat_input_operand): Likewise.
9942 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9944 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9946 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9947 (TARGET_INSTANTIATE_DECLS): Likewise.
9948 (legitimate_indexed_address_p): Delete reload_in_progress code.
9949 (rs6000_debug_legitimate_address_p): Likewise.
9950 (rs6000_eliminate_indexed_memrefs): Likewise.
9951 (rs6000_emit_le_vsx_store): Likewise.
9952 (rs6000_emit_move_si_sf_subreg): Likewise.
9953 (rs6000_emit_move): Likewise.
9954 (register_to_reg_type): Likewise.
9955 (rs6000_pre_atomic_barrier): Likewise.
9956 (rs6000_machopic_legitimize_pic_address): Likewise.
9957 (rs6000_allocate_stack_temp): Likewise.
9958 (rs6000_address_for_fpconvert): Likewise.
9959 (rs6000_address_for_altivec): Likewise.
9960 (rs6000_secondary_memory_needed_rtx): Delete function.
9961 (rs6000_check_sdmode): Likewise.
9962 (rs6000_alloc_sdmode_stack_slot): Likewise.
9963 (rs6000_instantiate_decls): Likewise.
9964 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9965 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9966 Delete reload_in_progress.
9967 (*vec_reload_and_plus_<mptrsize>): Likewise.
9968 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9969 (vsx_div_v2di): Likewise.
9970 (vsx_udiv_v2di): Likewise.
9972 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9974 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9975 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9976 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9977 (rs6000_debug_reg_global): Delete print of LRA status.
9978 (rs6000_option_override_internal): Delete dead LRA related code.
9979 (rs6000_lra_p): Delete function.
9980 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9982 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9984 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9985 * config/riscv/rtems.h: New file.
9987 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9988 Sudakshina Das <sudi.das@arm.com>
9990 * config/aarch64/aarch64.md
9991 (define_split for and<mode>3nr_compare): Move
9992 non aarch64_logical_operand to a register.
9993 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9994 register immediate operand to a register.
9995 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9997 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10000 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10002 2017-07-27 Richard Biener <rguenther@suse.de>
10004 PR tree-optimization/81573
10005 PR tree-optimization/81494
10006 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10007 multi defuse cycle case.
10009 2017-07-27 Richard Biener <rguenther@suse.de>
10011 PR tree-optimization/81571
10012 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10015 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10017 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10018 earlier and only if MASK_FPU is set. Adjust formatting.
10020 2017-07-27 Martin Liska <mliska@suse.cz>
10022 * opt-functions.awk: Add validation of value of Init.
10023 * optc-gen.awk: Pass new argument.
10025 2017-07-27 Martin Liska <mliska@suse.cz>
10027 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10028 Fix wrong condition.
10030 2017-07-27 Martin Liska <mliska@suse.cz>
10032 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10033 BBs and edges seen by autoFDO.
10035 2017-07-27 Richard Biener <rguenther@suse.de>
10037 PR tree-optimization/81502
10038 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10039 with incompatible but same sized type.
10040 (execute_update_addresses_taken): Likewise.
10042 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10044 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10045 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10047 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10050 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10051 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10052 Change s_operand to memory_operand.
10054 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10056 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10057 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10058 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10059 Emit instructions rather than returning an expression. Handle TFmode
10060 and KFmode by casting to TImode.
10061 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10062 (rs6000_emit_le_vsx_store): Likewise.
10063 * config/rs6000/vsx.md (VSX_TI): New iterator.
10064 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10065 (*vsx_le_undo_permute_<mode>): Likewise.
10066 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10067 emit the split sequence.
10068 (*vsx_le_perm_store_<mode>): Likewise.
10070 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10072 PR tree-optimization/81555
10073 PR tree-optimization/81556
10074 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10075 if true, force CHANGED for the recursive invocation.
10076 (reassociate_bb): Remember original length of ops array, pass
10077 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10079 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10080 attributes for noipa attribute. For naked attribute use
10081 lookup_attribute first before lookup_attribute_spec.
10082 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10084 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10085 for functions with noipa attribute.
10086 (cgraph_externally_visible_p): Return true for functions with noipa
10088 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10089 for functions with noipa attribute.
10090 * doc/extend.texi: Document noipa function attribute.
10091 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10092 also for functions with noipa attribute.
10093 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10095 2017-07-26 Andrew Pinski <apinski@cavium.com>
10097 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10098 vec_unalign_load_cost and vec_unalign_store_cost.
10100 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10102 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10103 -mvsx-small-integer option.
10104 (ISA_3_0_MASKS_IEEE): Likewise.
10105 (OTHER_VSX_VECTOR_MASKS): Likewise.
10106 (POWERPC_MASKS): Likewise.
10107 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10108 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10109 code, only testing for DImode being allowed in non-VSX floating
10111 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10112 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10114 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10115 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10116 TARGET_P8_VECTOR test.
10117 (rs6000_secondary_reload_simple_move): Likewise.
10118 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10119 since TARGET_P9_VECTOR was already tested.
10120 (rs6000_opt_masks): Remove -mvsx-small-integer.
10121 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10122 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10124 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10125 test for TARGET_VEXTRACTUB was used, and that uses
10127 (p9 extract splitter): Likewise.
10128 (vsx_extract_<mode>_di_p9): Likewise.
10129 (vsx_extract_<mode>_store_p9): Likewise.
10130 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10131 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10132 the elimination of TARGET_VSX_SMALL_INTEGER.
10133 (vsx_extract_<mode>_p8): Likewise.
10134 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10135 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10136 (vsx_set_<mode>_p9): Likewise.
10137 (vsx_set_v4sf_p9): Likewise.
10138 (vsx_set_v4sf_p9_zero): Likewise.
10139 (vsx_insert_extract_v4sf_p9): Likewise.
10140 (vsx_insert_extract_v4sf_p9_2): Likewise.
10141 * config/rs6000/rs6000.md (sign extend splitter): Change
10142 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10143 (floatsi<mode>2_lfiwax_mem): Likewise.
10144 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10145 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10146 since a test for TARGET_P9_VECTOR was used.
10147 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10148 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10149 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10150 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10151 TARGET_P8_VECTOR test.
10152 (fix_trunc<mode>si2_stfiwx): Likewise.
10153 (fix_trunc<mode>si2_internal): Likewise.
10154 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10155 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10157 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10158 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10159 TARGET_P8_VECTOR test.
10160 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10161 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10162 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10164 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10165 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10166 since a test for TARGET_P9_VECTOR was used.
10167 (splitter for loading small constants): Likewise.
10169 2017-07-26 Andrew Pinski <apinski@cavium.com>
10171 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10174 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10177 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10178 (fp_valid_at): Likewise.
10180 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10182 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10183 (qdf24xx_addrcost_table): Likewise.
10184 (cortexa57_tunings): Update to use generic_branch_cost.
10185 (cortexa72_tunings): Likewise.
10186 (cortexa73_tunings): Likewise.
10187 (qdf24xx_tunings): Likewise.
10189 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10191 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10192 (thunderx2t99_branch_cost): Likewise.
10193 (cortexa35_tunings): Update to use generic_branch_cost.
10194 (cortexa53_tunings): Likewise.
10195 (cortexa57_tunings): Likewise.
10196 (cortexa72_tunings): Likewise.
10197 (cortexa73_tunings): Likewise.
10198 (thunderx2t99_tunings): Likewise.
10200 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10202 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10203 (sparc_option_override): Honour MASK_FSMULD.
10204 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10205 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10206 * config/sparc/sparc.opt (mfsmuld): New option.
10207 * doc/invoke.texi (mfsmuld): Document option.
10209 2017-07-26 Marek Polacek <polacek@redhat.com>
10211 PR middle-end/70992
10212 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10214 2017-07-26 Richard Biener <rguenther@suse.de>
10216 * gimple-match-head.c (do_valueize): Return OP if valueize
10218 (get_def): New helper to get at the def stmt of a SSA name
10219 if valueize allows.
10220 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10221 do_valueize to get at the def stmt.
10222 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10224 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10226 PR middle-end/46932
10227 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10229 2017-07-26 Martin Liska <mliska@suse.cz>
10232 * function.c (expand_function_start): Make expansion of
10233 nonlocal_goto_save_area after parm_birth_insn.
10235 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10237 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10238 from all CPU target flags enable members.
10240 2017-07-26 Richard Biener <rguenther@suse.de>
10242 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10243 (decision_tree::gen): Make 'type' const.
10244 (write_predicate): Likewise.
10246 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10248 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10249 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10250 (rs6000_option_override_internal): Likewise.
10251 (rs6000_expand_vector_set): Likewise.
10252 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10253 (TARGET_UPPER_REGS_SF): Likewise.
10254 (TARGET_UPPER_REGS_DI): Likewise.
10255 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10256 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10257 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10258 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10259 (Splitters for DI constants in Altivec registers): Likewise.
10260 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10261 (vsx_set_v4sf_p9): Likewise.
10262 (vsx_set_v4sf_p9_zero): Likewise.
10263 (vsx_insert_extract_v4sf_p9): Likewise.
10264 (vsx_insert_extract_v4sf_p9_2): Likewise.
10266 2017-07-25 Carl Love <cel@us.ibm.com>
10268 * doc/extend.texi: Update the built-in documentation file for the
10269 existing built-in functions
10270 vector signed char vec_cnttz (vector signed char);
10271 vector unsigned char vec_cnttz (vector unsigned char);
10272 vector signed short vec_cnttz (vector signed short);
10273 vector unsigned short vec_cnttz (vector unsigned short);
10274 vector signed int vec_cnttz (vector signed int);
10275 vector unsigned int vec_cnttz (vector unsigned int);
10276 vector signed long long vec_cnttz (vector signed long long);
10277 vector unsigned long long vec_cnttz (vector unsigned long long);
10279 2017-07-25 Andrew Pinski <apinski@cavium.com>
10281 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10282 accesses where the use is for the first operand of a BIT_INSERT.
10284 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10287 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10288 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10290 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10292 * config/i386/gstabs.h: Delete.
10293 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10295 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10297 * config/i386/i386.c (ix86_decompose_address): Do not check for
10298 register RTX when looking at index_reg or base_reg.
10299 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10301 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10303 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10304 to update EH info here.
10306 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10308 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10310 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10312 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10314 2017-07-25 Torsten Duwe <duwe@suse.de>
10316 * common.opt: Introduce -fpatchable-function-entry
10317 command line option, and its variables function_entry_patch_area_size
10318 and function_entry_patch_area_start.
10319 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10320 including a two-value parser.
10321 * target.def (print_patchable_function_entry): New target hook.
10322 * targhooks.h (default_print_patchable_function_entry): New function.
10323 * targhooks.c (default_print_patchable_function_entry): Likewise.
10324 * toplev.c (process_options): Switch off IPA-RA if
10325 patchable function entries are being generated.
10326 * varasm.c (assemble_start_function): Look at the
10327 patchable-function-entry command line switch and current
10328 function attributes and maybe generate NOP instructions by
10329 calling the print_patchable_function_entry hook.
10330 * doc/extend.texi: Document patchable_function_entry attribute.
10331 * doc/invoke.texi: Document -fpatchable_function_entry
10332 command line option.
10333 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10335 * doc/tm.texi: Re-generate.
10337 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10340 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10341 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10343 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10345 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10347 2017-07-25 Richard Biener <rguenther@suse.de>
10349 PR tree-optimization/81455
10350 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10351 not walk in cycles when looking for guards.
10353 2017-07-25 Richard Biener <rguenther@suse.de>
10355 PR tree-optimization/81529
10356 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10357 when optimizing backedge uses.
10359 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10361 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10363 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10364 to dl_section_ref. On AIX, append an expression to subtract
10365 the size of the section length to dl_section_ref.
10367 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10369 * configure.ac: If any of the config.* scripts fail, exit 1.
10370 * configure: Regenerate.
10372 2017-07-25 Richard Biener <rguenther@suse.de>
10374 PR middle-end/81546
10375 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10376 of immediate uses, be more verbose on errors.
10378 2017-07-25 Richard Biener <rguenther@suse.de>
10380 PR tree-optimization/81510
10381 * tree-vect-loop.c (vect_is_simple_reduction): When the
10382 reduction stmt is not inside the loop bail out.
10384 2017-07-25 Richard Biener <rguenther@suse.de>
10386 PR tree-optimization/81303
10387 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10388 profitability check against LOOP_VINFO_NITERSM1.
10390 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10392 * domwalk.c (cmp_bb_postorder): Simplify.
10393 (sort_bbs_postorder): New function. Use it...
10394 (dom_walker::walk): ...here to optimize common cases.
10396 2017-07-25 Martin Liska <mliska@suse.cz>
10399 * ipa-visibility.c (function_and_variable_visibility): Make the
10400 redirection just on target that supports aliasing.
10401 Fix GNU coding style.
10403 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10406 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10408 * config/arm/rtems.h: Likewise.
10409 * config/bfin/rtems.h: Likewise.
10410 * config/i386/rtemself.h: Likewise.
10411 * config/lm32/rtems.h: Likewise.
10412 * config/m32c/rtems.h: Likewise.
10413 * config/m68k/rtemself.h: Likewise.
10414 * config/microblaze/rtems.h: Likewise.
10415 * config/mips/rtems.h: Likewise.
10416 * config/moxie/rtems.h: Likewise.
10417 * config/nios2/rtems.h: Likewise.
10418 * config/powerpcspe/rtems.h: Likewise.
10419 * config/rs6000/rtems.h: Likewise.
10420 * config/rtems.h: Likewise.
10421 * config/sh/rtems.h: Likewise.
10422 * config/sh/rtemself.h: Likewise.
10423 * config/sparc/rtemself.h: Likewise.
10425 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10428 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10429 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10430 * tree-ssa-structalias.c (alias_get_name): Same.
10432 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10435 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10436 instructions if no du chain is found.
10438 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10440 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10442 2017-07-25 Richard Biener <rguenther@suse.de>
10444 PR middle-end/81505
10445 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10448 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10450 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10451 upper-regs options.
10452 (ISA_2_7_MASKS_SERVER): Likewise.
10453 (ISA_3_0_MASKS_IEEE): Likewise.
10454 (OTHER_P8_VECTOR_MASKS): Likewise.
10455 (OTHER_VSX_VECTOR_MASKS): Likewise.
10456 (POWERPC_MASKS): Likewise.
10457 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10458 duplicate list of options.
10459 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10460 explicit -mupper-regs options.
10461 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10462 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10463 alias for -mupper-regs-df.
10464 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10465 (rs6000_init_hard_regno_mode_ok): Likewise.
10466 (rs6000_option_override_internal): Likewise.
10467 (rs6000_opt_masks): Likewise.
10468 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10469 options in terms of whether -mvsx or -mpower8-vector was used.
10470 (TARGET_UPPER_REGS_DI): Likewise.
10471 (TARGET_UPPER_REGS_SF): Likewise.
10472 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10473 -mupper-regs-* options.
10475 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10477 * passes.c (emergency_dump_function): Print some empty lines and a
10478 header before the RTL dump.
10480 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10482 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10484 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10487 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10488 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10490 2017-07-24 Carl Love <cel@us.ibm.com>
10492 * config/rs6000/rs6000-c.c: Add support for built-in functions
10493 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10494 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10495 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10496 vec_extract_fp_from_shortl): Add defines for the two builtins.
10497 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10498 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10500 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10501 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10502 * doc/extend.texi: Update the built-in documentation file for the
10503 new built-in function.
10505 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10508 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10510 * doc/generic.texi: Likewise.
10511 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10512 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10514 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10516 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10517 (aarch64_mls_elt_merge<mode>): Likewise.
10519 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10521 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10522 having __cxa_atexit.
10524 2017-07-23 Michael Collison <michael.collison@arm.com>
10526 * config/arm/arm.c (arm_option_override): Deprecate
10527 use of -mstructure-size-boundary.
10528 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10529 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10531 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10534 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10535 Reduce cost estimate for direct moves.
10537 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10540 * config/i386/i386.c (ix86_option_override_internal): Disable
10541 BMI, BMI2 and TBM instructions for -m16.
10543 2017-07-21 Carl Love <cel@us.ibm.com>
10545 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10546 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10547 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10549 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10550 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10551 VMULOSW): New enum "unspec" values.
10552 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10553 altivec_vmulosw): New patterns.
10554 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10555 VMULOSW): Add definitions.
10557 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10559 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10560 (qdf24xx): Likewise.
10561 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10562 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10563 (AARCH64_FL_V8_1): Renumber.
10564 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10565 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10566 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10567 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10568 rdma to feature modifiers list.
10570 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10572 PR middle-end/56727
10573 * ipa-visibility (function_and_variable_visibility): Convert
10574 recursive PLT call to direct call if appropriate.
10576 2017-07-21 Andrew Pinski <apinski@cavium.com>
10578 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10579 operand 1 to see if the types precision matches.
10580 * fold-const.c (operand_equal_p): Likewise.
10582 2017-07-21 Richard Biener <rguenther@suse.de>
10584 PR tree-optimization/81303
10585 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10586 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10587 (vect_peeling_hash_get_lowest_cost): Adjust.
10588 (vect_enhance_data_refs_alignment): Likewise. Use
10589 vect_get_peeling_costs_all_drs to compute the penalty for no
10590 peeling to match up costs.
10592 2017-07-21 Richard Biener <rguenther@suse.de>
10594 PR tree-optimization/81500
10595 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10596 we didn't identify a reduction path.
10598 2017-07-21 Tom de Vries <tom@codesourcery.com>
10599 Cesar Philippidis <cesar@codesourcery.com>
10601 PR gcov-profile/81442
10602 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10605 2017-07-21 Tom de Vries <tom@codesourcery.com>
10608 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10610 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10611 nvptx_override_options_after_change.
10613 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10615 * dwarf2out.c (output_file_names): Avoid double testing for
10616 dwarf_version >= 5.
10618 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10620 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10622 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10624 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10626 (try_crossjump_to_edge): Do not punt on partitioned functions.
10628 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10630 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10631 Put all BBs reachable only via paths crossing cold region to cold
10633 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10635 2016-07-21 Richard Biener <rguenther@suse.de>
10637 PR tree-optimization/81303
10638 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10639 into account prologue and epilogue iterations when raising
10640 min_profitable_iters to sth at least covering one vector iteration.
10642 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10644 * config/arm/arm.c (arm_test_cpu_arch_dat):
10647 2017-07-20 Nathan Sidwell <nathan@acm.org>
10649 Remove TYPE_METHODS.
10650 * tree.h (TYPE_METHODS): Delete.
10651 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10652 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10653 (dbxout_type_methods): Scan TYPE_FIELDS.
10654 (dbxout_type): Don't check TYPE_METHODS here.
10655 * function.c (use_register_for_decl): Always ignore register for
10656 class types when not optimizing.
10657 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10658 * tree.c (free_lang_data_in_type): Stitch out member functions and
10659 templates from TYPE_FIELDS.
10660 (build_distinct_type_copy, verify_type_variant,
10661 verify_type): Member fns are on TYPE_FIELDS.
10662 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10663 * tree-pretty-print.c (dump_generic_node): Likewise.
10665 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10668 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10669 V2TImode and V4TImode.
10670 (ix86_expand_vector_extract): Likewise.
10671 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10672 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10673 (ssescalarmode): Handle V4TImode and V2TImode.
10674 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10675 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10676 (VEXTRACTI128_MODE): New mode iterator.
10677 (splitter for *vec_extractv?ti first element): New.
10678 (VEC_INIT_MODE): New mode iterator.
10679 (vec_init<mode>): Consolidate 3 expanders into one using
10680 VEC_INIT_MODE mode iterator.
10682 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10684 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10685 non_spilled_static_chain_regno_p.
10687 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10689 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10691 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10693 * bb-reorder.c (connect_traces): Allow copying of blocks within
10696 2017-07-20 Richard Biener <rguenther@suse.de>
10698 * gimple.h (gimple_phi_result): Add gphi * overload.
10699 (gimple_phi_result_ptr): Likewise.
10700 (gimple_phi_arg): Likewise. Adjust index assert to only
10701 allow actual argument accesses rather than all slots available
10703 (gimple_phi_arg_def): Add gphi * overload.
10704 * tree-phinodes.c (make_phi_node): Initialize only actual
10706 (resize_phi_node): Clear memory not covered by old node,
10707 do not initialize excess argument slots.
10708 (reserve_phi_args_for_new_edge): Initialize new argument slot
10711 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10713 PR tree-optimization/81388
10715 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10717 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10718 by removing computation of may_be_zero.
10720 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10721 Tom de Vries <tom@codesourcery.com>
10723 PR middle-end/81030
10724 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10725 when gimple level profile disagrees with what RTL expander did.
10727 2017-07-20 Richard Biener <rguenther@suse.de>
10729 PR tree-optimization/61171
10730 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10731 (vect_analyze_stmt): Add slp instance parameter.
10732 (vectorizable_reduction): Likewise.
10733 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10734 (vect_is_simple_reduction): Deal with chains not detected
10735 as SLP reduction chain, specifically not properly associated
10736 chains containing a mix of plus/minus.
10737 (get_reduction_op): Remove.
10738 (get_initial_defs_for_reduction): Simplify, pass in whether
10739 this is a reduction chain, pass in the SLP node for the PHIs.
10740 (vect_create_epilog_for_reduction): Get the SLP instance as
10742 (vectorizable_reduction): Get the SLP instance as arg.
10743 During analysis remember the SLP node with the PHIs in the
10744 instance. Simplify getting at the vectorized reduction PHIs.
10745 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10746 through SLP instance.
10747 (vect_slp_analyze_operations): Likewise.
10748 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10749 (vect_transform_stmt): Likewise.
10751 2017-07-20 Tom de Vries <tom@codesourcery.com>
10753 PR tree-optimization/81489
10754 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10755 read of phi arg location to before loop that modifies phi.
10757 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10759 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10762 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10764 PR middle-end/81331
10765 * except.c (execute): Fix ordering issue.
10767 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10769 PR rtl-optimization/81423
10770 * combine.c (make_compound_operation_int): Don't try to optimize
10771 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10773 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10775 PR rtl-optimization/81423
10776 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10777 with a constant that is -1 in the truncated to mode.
10779 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10781 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10782 (determine_unlikely_bbs): ... here.
10783 * predict.h (propagate_unlikely_bbs_forward): Declare.
10784 * cfgexpand.c (pass_expand::execute): Use it.
10785 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10787 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10788 propagate_unlikely_bbs_forward.
10790 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10792 PR middle-end/81331
10793 * except.c (maybe_add_nop_after_section_switch): New function.
10796 2017-07-19 Tom de Vries <tom@codesourcery.com>
10798 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10800 2017-07-19 Tom de Vries <tom@codesourcery.com>
10802 * gimple.h (gimple_phi_arg): Make assert more strict.
10804 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10806 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10807 * config/rs6000/mmintrin.h: New file.
10808 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10810 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10812 PR tree-optimization/81346
10813 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10815 2017-07-19 Tom de Vries <tom@codesourcery.com>
10817 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10819 2017-07-19 Tom de Vries <tom@codesourcery.com>
10821 * config/nvptx/nvptx-modes.def: Add V2DImode.
10822 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10823 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10824 (nvptx_output_mov_insn): Handle lack of mov.b128.
10825 (nvptx_print_operand): Handle 'H' and 'L' codes.
10826 (nvptx_vector_mode_supported): Allow V2DImode.
10827 (nvptx_preferred_simd_mode): New function.
10828 (nvptx_data_alignment): New function.
10829 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10830 nvptx_preferred_simd_mode.
10831 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10833 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10835 2017-07-19 Tom de Vries <tom@codesourcery.com>
10837 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10838 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10839 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10840 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10841 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10842 (mov<VECIM>_insn): New define_insn.
10843 (define_expand "mov<VECIM>): New define_expand.
10845 2017-07-19 Tom de Vries <tom@codesourcery.com>
10847 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10849 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10851 PR tree-optimization/81346
10852 * fold-const.h (fold_div_compare, range_check_type): Declare.
10853 * fold-const.c (range_check_type): New function.
10854 (build_range_check): Use range_check_type.
10855 (fold_div_compare): No longer static, rewritten into
10856 a match.pd helper function.
10857 (fold_comparison): Don't call fold_div_compare here.
10858 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10859 as helper function.
10861 2017-07-19 Nathan Sidwell <nathan@acm.org>
10863 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10864 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10865 * tree.c (find_decls_types_r, verify_type): Use
10866 TYPE_{MIN,MAX}_VALUE_RAW.
10867 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10868 (hash_tree): Likewise.
10869 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10871 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10874 2017-07-18 Tom de Vries <tom@codesourcery.com>
10876 PR middle-end/81464
10877 * omp-expand.c (expand_omp_for_static_chunk): Handle
10878 equal-argument loop exit phi.
10880 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10883 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10884 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10885 operand 2 predicate.
10886 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10887 operand 2 predicate.
10888 (ror,rol -> rorx splitters): Use const_int_operand as
10889 operand 2 predicate.
10891 2017-06-18 Richard Biener <rguenther@suse.de>
10893 PR tree-optimization/81410
10894 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10895 the gap in the ! slp_perm SLP case after each group.
10897 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10899 PR middle-end/81463
10900 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10903 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10905 PR middle-end/81462
10906 * predict.c (set_even_probabilities): Cleanup; do not affect
10907 probabilities that are already known.
10908 (combine_predictions_for_bb): Call even when count is set.
10910 2017-07-18 Nathan Sidwell <nathan@acm.org>
10912 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10915 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10918 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10919 optimization for loop niter analysis.
10921 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10924 * config/avr/avr.c (avr_optimize_casesi): Don't use
10925 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10927 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10929 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10930 body_cost_vec from _vect_peel_extended_info.
10931 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10932 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10935 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10937 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10939 2017-07-18 Richard Biener <rguenther@suse.de>
10941 PR tree-optimization/80620
10942 PR tree-optimization/81403
10943 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10944 info when re-using a VN table entry.
10946 2017-07-18 Richard Biener <rguenther@suse.de>
10948 PR tree-optimization/81418
10949 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10950 vectype_in. Verify that with lane-reducing reduction operations
10951 we have a single def-use cycle.
10953 2017-07-17 Carl Love <cel@us.ibm.com>
10955 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10957 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10958 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10959 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10960 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10961 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10962 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10963 VMULOSW): New enum "unspec" values.
10964 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10965 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10966 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10967 altivec_vmulosw): New patterns.
10968 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10969 VMULOSW): Add definitions.
10970 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10972 * config/alpha/alpha.c: Include predict.h.
10974 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10976 * tree-vrp.c (compare_assert_loc): Fix comparison function
10977 to return predictable results.
10979 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10981 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10983 (subdi3): Likewise.
10984 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10985 * doc/invoke.texi (mexpand-adddi): Update text.
10987 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10989 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10990 that also clobbers the CC register. The old expand code is moved
10992 (*arc_clzsi2): ... here.
10993 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10994 the CC register. The old expand code is moved to ...
10995 (arc_ctzsi2): ... here.
10997 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10999 * config/arc/arc.opt (mindexed-loads): Use initial value
11000 TARGET_INDEXED_LOADS_DEFAULT.
11001 (mauto-modify-reg): Use initial value
11002 TARGET_AUTO_MODIFY_REG_DEFAULT.
11003 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11004 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11005 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11006 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11008 2017-07-17 Martin Liska <mliska@suse.cz>
11011 * opts.c (finish_options): Do not allow -fgnu-tm
11012 w/ -fsanitize={kernel-,}address. Say sorry.
11014 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11017 * tree-loop-distribution.c (classify_partition): Only assert on
11018 numer of iterations.
11019 (merge_dep_scc_partitions): Delete prameter. Update function call.
11020 (distribute_loop): Remove code handling loop with unknown niters.
11021 (pass_loop_distribution::execute): Skip loop with unknown niters.
11023 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11026 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11027 function sort_partitions_by_post_order.
11029 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11031 PR tree-optimization/81374
11032 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11033 the max index of basic blocks, rather than number of basic blocks.
11035 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11037 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11039 (arc_legitimate_pic_operand_p): Likewise.
11040 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11042 (arc_needs_pcl_p): Likewise.
11043 (arc_legitimate_pc_offset_p): Likewise.
11044 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11045 function is also used in constrains.md.
11046 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11047 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11048 PLUS. Only return true/false in known cases, otherwise assert.
11049 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11050 is already called in arc_legitimate_constant_p.
11051 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11053 (LEGITIMATE_PIC_OPERAND_P): Use
11054 arc_raw_symbolic_reference_mentioned_p function.
11055 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11060 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11061 Andrew Burgess <andrew.burgess@embecosm.com>
11063 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11064 (arc_return_address_register): New function.
11065 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11066 (arc_handle_fndecl_attribute): Add naked attribute.
11067 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11068 (TARGET_WARN_FUNC_RETURN): Likewise.
11069 (arc_allocate_stack_slots_for_args): New function.
11070 (arc_warn_func_return): Likewise.
11071 (machine_function): Change type fn_type.
11072 (arc_compute_function_type): Consider new naked function type,
11073 change function return type.
11074 (arc_must_save_register): Adapt to handle new
11075 arc_compute_function_type's return type.
11076 (arc_expand_prologue): Likewise.
11077 (arc_expand_epilogue): Likewise.
11078 (arc_return_address_regs): Delete.
11079 (arc_return_address_register): New function.
11080 (arc_epilogue_uses): Use above function.
11081 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11082 (arc_function_type): Change encoding, add naked type.
11083 (ARC_INTERRUPT_P): Change to handle the new encoding.
11084 (ARC_FAST_INTERRUPT_P): Likewise.
11085 (ARC_NORMAL_P): Define.
11086 (ARC_NAKED_P): Likewise.
11087 (arc_compute_function_type): Delete prototype.
11088 * config/arc/arc.md (in_ret_delay_slot): Use
11089 arc_return_address_register function.
11090 (simple_return): Likewise.
11091 (p_return_i): Likewise.
11093 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11095 PR tree-optimization/81428
11096 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11097 can't be built for those types.
11099 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11101 Remove stuff dead since r239246.
11103 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11104 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11105 (avr_inform_devices): Remove dead stuff.
11107 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11109 * config/arm/arm_neon.h: Fix softp typo.
11111 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11113 PR tree-optimization/81365
11114 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11115 aggregate moves onto bb predecessor edges, make sure there are no
11116 loads that could alias the lhs in between the start of bb and the
11119 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11122 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11123 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11124 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11126 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11128 PR tree-optimization/81396
11129 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11130 (init_symbolic_number): Initialize it to 1.
11131 (perform_symbolic_merge): Add n_ops from both operands into the new
11133 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11134 without base_addr as useless if they need more than one operation.
11135 (bswap_replace): Handle !bswap case for NULL base_addr.
11137 2017-07-17 Tom de Vries <tom@codesourcery.com>
11140 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11143 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11145 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11146 conditional builtin define __FIX_LEON3FT_B2BST.
11148 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11150 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11151 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11154 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11156 PR rtl-optimization/81424
11157 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11158 to remove potential trapping from operands if -fnon-call-exceptions.
11160 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11162 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11163 profile_proability for scalling.
11164 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11166 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11168 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11170 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11172 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11173 fixpoint arithmetics.
11175 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11177 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11178 fixpoint arithmetics.
11180 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11182 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11183 fixpoint arithmetics.
11185 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11187 * profile-count.h (profile_probability::from_reg_br_prob_note,
11188 profile_probability::to_reg_br_prob_note): New functions.
11189 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11190 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11191 * predict.c (probability_reliable_p): Update.
11192 (edge_probability_reliable_p): Update.
11193 (br_prob_note_reliable_p): Update.
11194 (invert_br_probabilities): Update.
11195 (add_reg_br_prob_note): New function.
11196 (combine_predictions_for_insn): Update.
11197 * asan.c (asan_clear_shadow): Update.
11198 * cfgbuild.c (compute_outgoing_frequencies): Update.
11199 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11200 (update_br_prob_note): Update.
11201 (rtl_verify_edges): Update.
11202 (purge_dead_edges): Update.
11203 (fixup_reorder_chain): Update.
11204 * emit-rtl.c (try_split): Update.
11205 * ifcvt.c (cond_exec_process_insns): Update.
11206 (cond_exec_process_if_block): Update.
11207 (dead_or_predicable): Update.
11208 * internal-fn.c (expand_addsub_overflow): Update.
11209 (expand_neg_overflow): Update.
11210 (expand_mul_overflow): Update.
11211 * loop-doloop.c (doloop_modify): Update.
11212 * loop-unroll.c (compare_and_jump_seq): Update.
11213 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11214 * predict.h: Update.
11215 * reorg.c (mostly_true_jump): Update.
11217 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11218 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11219 * config/arc/arc.c: (emit_unlikely_jump): Update.
11220 * config/arm/arm.c: (emit_unlikely_jump): Update.
11221 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11222 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11223 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11224 (ix86_print_operand): Update.
11225 (ix86_split_fp_branch): Update.
11226 (predict_jump): Update.
11227 * config/ia64/ia64.c (ia64_print_operand): Update.
11228 * config/mmix/mmix.c (mmix_print_operand): Update.
11229 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11230 (rs6000_expand_split_stack_prologue): Update.
11231 * config/rs6000/rs6000.c: Update.
11232 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11233 (s390_expand_vec_movstr): Update.
11234 (s390_expand_cs_tdsi): Update.
11235 (s390_expand_split_stack_prologue): Update.
11236 * config/sh/sh.c (sh_print_operand): Update.
11237 (expand_cbranchsi4): Update.
11238 (expand_cbranchdi4): Update.
11239 * config/sparc/sparc.c (output_v9branch): Update.
11240 * config/spu/spu.c (get_branch_target): Update.
11241 (ea_load_store_inline): Update.
11242 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11243 * config/tilepro/tilepro.c: Update.
11245 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11247 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11248 (gimplify_save_expr): Likewise.
11250 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11252 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11254 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11256 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11258 * timevar.def (TV_IPA_FNSUMMARY): Define.
11260 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11262 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11263 to back store errata sensitive sequence from being generated.
11264 (sqrtdf2_fix): Likewise.
11266 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11268 * tree-ssa-threadupdate.c (compute_path_counts,
11269 update_joiner_offpath_counts): Use profile_probability.
11271 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11274 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11276 * config/arm/arm-c.c (arm_cpu_builtins): Define
11277 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11279 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11281 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11282 array entries to represent __ieee128 versions of the
11283 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11284 scalar_extract_sig, and scalar_insert_exp built-in functions.
11285 (altivec_resolve_overloaded_builtin): Add special case handling
11286 for the __builtin_scalar_insert_exp function, as represented by
11287 the P9V_BUILTIN_VEC_VSIEDP constant.
11288 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11289 exponent support for __ieee128 argument.
11290 (VSESQP): Add scalar extract signature support for __ieee128
11292 (VSTDCNQP): Add scalar test negative support for __ieee128
11294 (VSIEQP): Add scalar insert exponent support for __int128 argument
11295 with __ieee128 result.
11296 (VSIEQPF): Add scalar insert exponent support for __ieee128
11297 argument with __ieee128 result.
11298 (VSTDCQP): Add scalar test data class support for __ieee128
11300 (VSTDCNQP): Add overload support for scalar test negative with
11301 __ieee128 argument.
11302 (VSTDCQP): Add overload support for scalar test data class
11303 __ieee128 argument.
11304 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11305 UNSPEC_VSX_SXSIGDP.
11306 (UNSPEC_VSX_SIEXPQP): New constant.
11307 (xsxexpqp): New insn for VSX scalar extract exponent quad
11309 (xsxsigqp): New insn for VSX scalar extract significand quad
11311 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11312 precision with floating point argument.
11313 (xststdcqp): New expand for VSX scalar test data class quad
11315 (xststdcnegqp): New expand for VSX scalar test negative quad
11317 (xststdcqp): New insn to match expansions for VSX scalar test data
11318 class quad precision and VSX scalar test negative quad precision.
11319 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11320 special case operand checking to enforce that second operand of
11321 VSX scalar test data class with quad precision argument is a 7-bit
11323 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11324 prototypes and descriptions of __ieee128 versions of
11325 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11326 scalar_test_data_class, and scalar_test_neg built-in functions.
11328 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11330 PR tree-optimization/81162
11331 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11332 replace a negate with an add.
11334 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11336 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11338 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11340 * config/arm/arm-c.c (arm_cpu_builtins): Define
11341 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11343 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11345 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11346 (armv8-r): Set ARM Cortex-R52 as default CPU.
11347 * config/arm/arm-tables.opt: Regenerate.
11348 * config/arm/arm-tune.md: Regenerate.
11349 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11351 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11352 extension for -mcpu=cortex-r52.
11354 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11356 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11357 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11358 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11359 (fp-armv8): Define it as FP_ARMv8 only.
11360 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11361 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11363 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11364 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11365 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11366 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11367 than TARGET_FPU_ARMV8.
11368 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11369 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11370 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11372 * config/arm/neon.md (neon_vrint): Likewise.
11373 (neon_vcvt): Likewise.
11374 (neon_<fmaxmin_op><mode>): Likewise.
11375 (<fmaxmin><mode>3): Likewise.
11376 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11377 * config/arm/predicates.md (arm_cond_move_operator): Check against
11378 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11380 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11382 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11383 to top of function.
11385 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11387 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11388 loop in comment with memset.
11390 2017-07-14 Martin Liska <mliska@suse.cz>
11392 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11393 * dwarf2out.c (is_java): Remove the function.
11394 (output_pubname): Remove usage of the function.
11395 (lower_bound_default): Remove usage of DW_LANG_Java.
11396 (gen_compile_unit_die): Likewise.
11397 * gcc.c: Remove compiler defaults for .java and .zip files.
11398 * gimple-expr.c (remove_suffix): Change as there's no longer
11399 extension than 4-letter one.
11400 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11401 (gimplify_save_expr): Likewise.
11402 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11403 as it's possible even for other languages than Java.
11404 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11405 * lto-opts.c (lto_write_options): Remove reference to Java.
11406 * opts.c (strip_off_ending): Update file extension handling.
11407 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11408 * tree-eh.c (lower_resx): Likewise.
11409 * tree.c (free_lang_data_in_type): Remove dead code.
11410 (find_decls_types_r): Likewise.
11411 (build_common_builtin_nodes): Remove Java from a comment.
11412 (verify_type): Remove dead code.
11413 * varasm.c (assemble_external): Remove Java from a comment.
11415 2017-07-14 Martin Liska <mliska@suse.cz>
11417 * opts.c (finish_options): Add quotes.
11418 (common_handle_option): Likewise.
11420 2017-07-14 Martin Liska <mliska@suse.cz>
11422 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11423 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11424 Remove N_SO_PASCAL.
11425 * dwarf2out.c (lower_bound_default): Do not handle
11427 (gen_compile_unit_die): Likewise.
11428 * gcc.c: Remove default extension binding for GNU Pascal.
11429 * stmt.c: Remove Pascal language from a comment.
11430 * xcoffout.c: Likewise.
11432 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11435 * diagnostic-show-locus.c (fixit_cmp): New function.
11436 (layout::layout): Sort m_fixit_hints.
11437 (column_range::column_range): Assert that the values are valid.
11438 (struct char_span): New struct.
11439 (correction::overwrite): New method.
11440 (struct source_line): New struct.
11441 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11442 calls in terms of classes source_line and char_span, and
11443 correction::overwrite.
11444 (selftest::test_overlapped_fixit_printing_2): New function.
11445 (selftest::diagnostic_show_locus_c_tests): Call it.
11447 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11449 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11450 early if there is no lhs.
11452 2017-07-13 Martin Liska <mliska@suse.cz>
11454 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11455 (gen_reference_type_die): Likewise.
11456 * stor-layout.c: Remove Pascal-related comment.
11458 2017-07-13 Martin Liska <mliska@suse.cz>
11460 * opts.c (finish_options): Add quotes to error messages.
11461 (parse_sanitizer_options): Likewise.
11463 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11465 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11467 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11469 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11471 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11473 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11475 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11477 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11480 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11481 provides the hardware capability bits, define the macro
11482 __BUILTIN_CPU_SUPPORTS__.
11483 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11484 if GLIBC does not provide the hardware capability bits. Add a
11485 gcc_unreachable call if the built-in cpu function is neither
11486 __builtin_cpu_is nor __builtin_cpu_supports.
11487 (rs6000_get_function_versions_dispatcher): Change the warning
11488 that an old GLIBC is used which does not export the capability
11489 bits to be an error.
11490 * doc/extend.texi (target_clones attribute): Document the
11491 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11492 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11493 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11494 the macros defined by GCC if the newer GLIBC is available.
11496 2017-07-12 Jeff Law <law@redhat.com>
11498 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11499 remaining includes slightly.
11500 * config/riscv/riscv-builtins.c: Include profile-count.h.
11502 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11505 * config/avr/avr.c (avr_set_current_function): In diagnostic
11506 messages: Quote keywords and (parts of) identifiers.
11507 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11510 2017-07-12 Carl Love <cel@us.ibm.com>
11512 * config/rs6000/rs6000-c.c: Add support for built-in functions
11513 vector bool char vec_revb (vector bool char);
11514 vector bool short vec_revb (vector short char);
11515 vector bool int vec_revb (vector bool int);
11516 vector bool long long vec_revb (vector bool long long);
11517 * doc/extend.texi: Update the built-in documentation file for the
11518 new built-in functions.
11520 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11522 * config/s390/s390.md: Remove movcc splitter.
11524 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11526 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11527 load/store on condition.
11529 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11532 * config/avr/avr.c (avr_encode_section_info)
11533 [progmem && !TREE_READONLY]: Error if progmem object needs
11536 2017-07-11 Michael Collison <michael.collison@arm.com>
11538 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11541 2017-07-11 Carl Love <cel@us.ibm.com>
11543 * config/rs6000/rs6000-c.c: Add support for builtins
11544 vector unsigned int vec_parity_lsbb (vector signed int);
11545 vector unsigned int vec_parity_lsbb (vector unsigned int);
11546 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11547 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11548 vector unsigned long long vec_parity_lsbb (vector signed long long);
11549 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11550 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11551 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11552 * doc/extend.texi: Update the built-in documentation file for the
11553 new built-in functions.
11555 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11557 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11558 (layout::m_primary_loc): New field.
11559 (layout::layout): Initialize new field. Move location filtering
11560 logic from here to...
11561 (layout::maybe_add_location_range): ...this new method. Add
11562 support for filtering to just the lines already specified by other
11564 (layout::will_show_line_p): New method.
11565 (gcc_rich_location::add_location_if_nearby): New method.
11566 (selftest::test_add_location_if_nearby): New test function.
11567 (selftest::diagnostic_show_locus_c_tests): Call it.
11568 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11571 2017-07-11 Tom de Vries <tom@codesourcery.com>
11573 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11574 (bb_first_real_insn): New function.
11575 (nvptx_single): Add extra initialization of broadcasted condition
11578 2017-07-11 Nathan Sidwell <nathan@acm.org>
11580 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11582 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11584 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11585 Binutils doc as TEXI will mess them up.
11586 * doc/invoke.texi (AVR Options): Same here.
11588 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11590 * config/sparc/sparc.opt (mfix-ut700): New option.
11591 (mfix-gr712rc): Likewise.
11592 (sparc_fix_b2bst): New variable.
11593 * doc/invoke.texi (SPARC options): Document them.
11594 (ARM options): Fix warnings.
11595 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11596 instructions to prevent sequences that can trigger the store-store
11597 errata for certain LEON3FT processors.
11598 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11599 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11600 * config/sparc/sparc.md (fix_b2bst): New attribute.
11601 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11603 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11606 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11608 (*rsqrtsf2_sse): Ditto.
11610 (div<mode>3): Macroize insn from divdf3 and divsf3
11611 using MODEF mode iterator.
11613 2017-07-10 Martin Sebor <msebor@redhat.com>
11615 PR tree-optimization/80397
11616 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11617 instead of testing for equality to INTEGER_TYPE.
11619 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11621 * config.gcc: Remove uclibc from arc target spec.
11623 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11625 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11627 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11630 * lto-wrapper.c (remove_option): New function.
11631 (merge_and_complain): Merge PIC/PIE options more realistically.
11633 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11635 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11639 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11640 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11641 * config.in: Regenerate.
11642 * configure: Regenerate.
11643 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11644 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11645 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11646 (TARGET_GASISR_PROLOGUES): ...target mask.
11647 * common/config/avr/avr-common.c
11648 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11649 Set -mgas-isr-prologues.
11650 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11651 INSERT_PASS_BEFORE for it.
11652 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11653 * config/avr/avr.c (avr_option_override)
11654 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11655 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11656 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11657 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11658 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11660 (make_avr_pass_pre_proep): New function.
11661 (emit_push_sfr) <treg>: Add argument to function and use it
11662 instead of TMP_REG.
11663 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11664 and set machine->gasisr.yes.
11665 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11666 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11667 __gcc_isr.n_pushed to .L__stack_usage.
11668 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11669 (avr_asm_final_postscan_insn): ...this new static function.
11670 * config/avr/avr.h (machine_function)
11671 <is_no_gccisr, use_L__stack_usage>: New fields.
11672 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11673 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11674 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11675 (gasisr, *gasisr): New expander and insn.
11676 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11677 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11678 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11680 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11682 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11685 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11687 Move jump-tables out of .text again.
11690 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11691 (ASM_OUTPUT_ADDR_VEC): New function.
11692 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11693 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11694 INSN_ADDRESSes as asm comment.
11695 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11696 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11697 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11698 * config/avr/avr.md (*tablejump): Adjust comment.
11699 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11700 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11702 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11703 (avr_output_addr_vec): New proto.
11704 (avr_log_t) <insn_addresses>: New field.
11706 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11709 * config/i386/i386.c (ix86_function_arg_advance): Set
11710 outgoing_args_on_stack to true if there are outgoing arguments
11712 (ix86_function_arg): Likewise.
11713 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11714 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11715 * config/i386/i386.h (machine_function): Add
11716 outgoing_args_on_stack.
11718 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11720 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11721 supporting pthreds.
11722 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11724 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11726 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11727 (REAL_H): Remove $(MACHMODE_H).
11728 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11730 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11731 $(MACHMODE_H) and double-int.h.
11732 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11734 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11737 2017-07-07 Andrew Pinski <apinski@cavium.com>
11739 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11740 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11742 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11744 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11745 Add warning if GCC was not configured to link against a GLIBC that
11746 exports the hardware capability bits.
11747 (make_resolver_func): Make resolver function private and not a
11748 COMDAT function. Create the name with clone_function_name instead
11749 of make_unique_name.
11752 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11753 correct operand in doing the split.
11755 2017-07-07 Carl Love <cel@us.ibm.com>
11757 * config/rs6000/rs6000-c: Add support for built-in function
11758 vector unsigned short vec_pack_to_short_fp32 (vector float,
11760 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11761 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11762 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11763 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11764 (convert_4f32_8i16): Add define_expand.
11765 * doc/extend.texi: Update the built-in documentation file for the
11766 new built-in function.
11768 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11770 * config/sparc/m8.md: New file.
11771 * config/sparc/sparc.md: Include m8.md.
11773 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11775 * config/sparc/sparc.opt: New option -mvis4b.
11776 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11777 (sparc_option_override): Handle VIS4B.
11778 (enum sparc_builtins): Define
11779 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11780 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11781 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11782 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11783 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11784 (check_constant_argument): New function.
11785 (sparc_vis_init_builtins): Define builtins
11786 __builtin_vis_dictunpack{8,16,32},
11787 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11788 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11789 __builtin_vis_fpcmpde{8,16,32}shl and
11790 __builtin_vis_fpcmpur{8,16,32}shl.
11791 (sparc_expand_builtin): Check that the constant operands to
11792 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11793 constant and in range.
11794 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11796 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11797 (SPARC_IMM5_P): Likewise.
11798 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11799 (enabled): Handle vis4b.
11800 (UNSPEC_DICTUNPACK): New unspec.
11801 (UNSPEC_FPCMPSHL): Likewise.
11802 (UNSPEC_FPUCMPSHL): Likewise.
11803 (UNSPEC_FPCMPDESHL): Likewise.
11804 (UNSPEC_FPCMPURSHL): Likewise.
11805 (cpu_feature): New CPU feature `vis4b'.
11806 (dictunpack{8,16,32}): New insns.
11807 (FPCSMODE): New mode iterator.
11808 (fpcscond): New code iterator.
11809 (fpcsucond): Likewise.
11810 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11811 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11812 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11813 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11814 * config/sparc/constraints.md: Define constraints `q' for unsigned
11815 2-bit integer constants and `t' for unsigned 5-bit integer
11817 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11819 (imm5_operand_dictunpack16): Likewise.
11820 (imm5_operand_dictunpack32): Likewise.
11821 (imm2_operand): Likewise.
11822 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11823 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11824 ditunpack* and fpcmp*shl builtins.
11826 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11828 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11829 * config.in: Add HAVE_AS_SPARC6 define.
11830 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11832 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11834 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11835 (CPP_CPU_SPEC): Handle m8.
11836 (ASM_CPU_SPEC): Likewise.
11837 * config/sparc/sparc-opts.h (enum processor_type): Add
11839 * config/sparc/sparc.c (m8_costs): New struct.
11840 (sparc_option_override): Handle TARGET_CPU_m8.
11841 (sparc32_initialize_trampoline): Likewise.
11842 (sparc64_initialize_trampoline): Likewise.
11843 (sparc_issue_rate): Likewise.
11844 (sparc_register_move_cost): Likewise.
11845 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11846 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11847 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11848 (CPP_CPU_SPEC): Handle M8.
11849 (ASM_CPU_SPEC): Likewise.
11850 (AS_M8_FLAG): Define.
11851 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11852 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11853 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11855 * configure: Regenerate.
11856 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11859 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11861 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11863 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11864 ("*movdi_insn_sp32"): Do not set v3pipe.
11865 ("*movsi_insn"): Likewise.
11866 ("*movdi_insn_sp64"): Likewise.
11867 ("*movsf_insn"): Likewise.
11868 ("*movdf_insn_sp32"): Likewise.
11869 ("*movdf_insn_sp64"): Likewise.
11870 ("*zero_extendsidi2_insn_sp64"): Likewise.
11871 ("*sign_extendsidi2_insn"): Likewise.
11872 ("*mov<VM32:mode>_insn"): Likewise.
11873 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11874 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11875 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11876 ("<vlop:code><VL:mode>3"): Likewise.
11877 ("*not_<vlop:code><VL:mode>3"): Likewise.
11878 ("*nand<VL:mode>_vis"): Likewise.
11879 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11880 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11881 ("one_cmpl<VL:mode>2"): Likewise.
11882 ("faligndata<VM64:mode>_vis"): Likewise.
11883 ("alignaddrsi_vis"): Likewise.
11884 ("alignaddrdi_vis"): Likweise.
11885 ("alignaddrlsi_vis"): Likewise.
11886 ("alignaddrldi_vis"): Likewise.
11887 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11888 ("bmaskdi_vis"): Likewise.
11889 ("bmasksi_vis"): Likewise.
11890 ("bshuffle<VM64:mode>_vis"): Likewise.
11891 ("cmask8<P:mode>_vis"): Likewise.
11892 ("cmask16<P:mode>_vis"): Likewise.
11893 ("cmask32<P:mode>_vis"): Likewise.
11894 ("pdistn<P:mode>_vis"): Likewise.
11895 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11897 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11899 * config/sparc/sparc.md ("subtype"): New insn attribute.
11900 ("*wrgsr_sp64"): Set insn subtype.
11901 ("*rdgsr_sp64"): Likewise.
11902 ("alignaddrsi_vis"): Likewise.
11903 ("alignaddrdi_vis"): Likewise.
11904 ("alignaddrlsi_vis"): Likewise.
11905 ("alignaddrldi_vis"): Likewise.
11906 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11907 ("fexpand_vis"): Likewise.
11908 ("fpmerge_vis"): Likewise.
11909 ("faligndata<VM64:mode>_vis"): Likewise.
11910 ("bshuffle<VM64:mode>_vis"): Likewise.
11911 ("cmask8<P:mode>_vis"): Likewise.
11912 ("cmask16<P:mode>_vis"): Likewise.
11913 ("cmask32<P:mode>_vis"): Likewise.
11914 ("fchksm16_vis"): Likewise.
11915 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11916 ("fmean16_vis"): Likewise.
11917 ("fp<plusminus_insn>64_vis"): Likewise.
11918 ("<plusminus_insn>v8qi3"): Likewise.
11919 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11920 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11921 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11922 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11923 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11924 ("*movqi_insn"): Likewise.
11925 ("*movhi_insn"): Likewise.
11926 ("*movsi_insn"): Likewise.
11927 ("movsi_pic_gotdata_op"): Likewise.
11928 ("*movdi_insn_sp32"): Likewise.
11929 ("*movdi_insn_sp64"): Likewise.
11930 ("movdi_pic_gotdata_op"): Likewise.
11931 ("*movsf_insn"): Likewise.
11932 ("*movdf_insn_sp32"): Likewise.
11933 ("*movdf_insn_sp64"): Likewise.
11934 ("*zero_extendhisi2_insn"): Likewise.
11935 ("*zero_extendqihi2_insn"): Likewise.
11936 ("*zero_extendqisi2_insn"): Likewise.
11937 ("*zero_extendqidi2_insn"): Likewise.
11938 ("*zero_extendhidi2_insn"): Likewise.
11939 ("*zero_extendsidi2_insn_sp64"): Likewise.
11940 ("ldfsr"): Likewise.
11941 ("prefetch_64"): Likewise.
11942 ("prefetch_32"): Likewise.
11943 ("tie_ld32"): Likewise.
11944 ("tie_ld64"): Likewise.
11945 ("*tldo_ldub_sp32"): Likewise.
11946 ("*tldo_ldub1_sp32"): Likewise.
11947 ("*tldo_ldub2_sp32"): Likewise.
11948 ("*tldo_ldub_sp64"): Likewise.
11949 ("*tldo_ldub1_sp64"): Likewise.
11950 ("*tldo_ldub2_sp64"): Likewise.
11951 ("*tldo_ldub3_sp64"): Likewise.
11952 ("*tldo_lduh_sp32"): Likewise.
11953 ("*tldo_lduh1_sp32"): Likewise.
11954 ("*tldo_lduh_sp64"): Likewise.
11955 ("*tldo_lduh1_sp64"): Likewise.
11956 ("*tldo_lduh2_sp64"): Likewise.
11957 ("*tldo_lduw_sp32"): Likewise.
11958 ("*tldo_lduw_sp64"): Likewise.
11959 ("*tldo_lduw1_sp64"): Likewise.
11960 ("*tldo_ldx_sp64"): Likewise.
11961 ("*mov<VM32:mode>_insn"): Likewise.
11962 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11963 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11965 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11967 * config/sparc/sparc.md ("type"): New insn type viscmp.
11968 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11970 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11971 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11972 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11973 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11975 ("n7_vis_logical_11cycle"): Likewise.
11976 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11977 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11978 * config/sparc/niagara.md ("niag_vis"): Likewise.
11979 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11980 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11982 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11984 * config/sparc/sparc.md: New instruction type `bmask'.
11985 (bmaskdi_vis): Use the `bmask' type.
11986 (bmasksi_vis): Likewise.
11987 * config/sparc/ultra3.md (us3_array): Likewise.
11988 * config/sparc/niagara7.md (n7_array): Likewise.
11989 * config/sparc/niagara4.md (n4_array): Likewise.
11990 * config/sparc/niagara2.md (niag2_vis): Likewise.
11991 (niag3_vis): Likewise.
11992 * config/sparc/niagara.md (niag_vis): Likewise.
11994 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11996 * ipa-comdats.c: Remove optimize check from gate.
11997 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11998 for functions not optimized.
11999 (ipa_fn_summary_read): Skip optimize check.
12000 (ipa_fn_summary_write): Likewise.
12001 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12003 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12005 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12006 (check_callers): Give up on caller which is not optimized.
12007 (inline_small_functions): Likewise.
12008 (ipa_inline): Do not give up when not optimizing.
12009 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12010 away unoptimizes cdtors.
12011 (whole_program_function_and_variable_visibility): Do
12012 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12013 * ipa.c (process_references): Do not check optimize.
12014 (symbol_table::remove_unreachable_nodes): Update optimize check.
12015 (set_writeonly_bit): Update optimize check.
12016 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12017 (pass_ipa_single_use::gate): Remove.
12019 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12021 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12022 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12023 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12024 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12025 permute_load, permute_store, adjust_extract, adjust_splat,
12026 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12027 replace_swap_with_copy, dump_swap_insn_table,
12028 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12029 recombine_lvx_pattern, recombine_stvx_pattern,
12030 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12031 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12032 to file rs6000-p8swap.c.
12033 * config/rs6000/rs6000-p8swap.c: New file.
12034 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12035 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12036 and rs6000*-*-* targets.
12038 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12040 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12042 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12044 * lto-wrapper.c (merge_and_complain): Do not merge
12045 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12046 fsigned_zeros, ftrapping_math, fwrapv.
12047 (append_compiler_options): Do not track these options.
12048 (append_linker_options): Likewie
12050 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12052 * cgraphunit.c (cgraph_node::finalize_function): When
12053 !flag_toplevel_reorde set no_reorder flag.
12054 (varpool_node::finalize_decl): Likewise.
12055 (symbol_table::compile): Drop no toplevel reorder path.
12057 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12059 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12060 edges; zero probability is not better than uninitialized.
12062 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12064 * asan.h (asan_sanitize_allocas_p): Declare.
12065 * asan.c (asan_sanitize_allocas_p): New function.
12066 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12067 (handle_builtin_alloca): Likewise.
12068 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12069 if !asan_sanitize_allocas_p.
12070 * params.def (asan-instrument-allocas): Add new option.
12071 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12072 * opts.c (common_handle_option): Disable allocas sanitization for
12075 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12077 * asan.c: Include gimple-fold.h.
12078 (get_last_alloca_addr): New function.
12079 (handle_builtin_stackrestore): Likewise.
12080 (handle_builtin_alloca): Likewise.
12081 (asan_emit_allocas_unpoison): Likewise.
12082 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12083 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12084 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12085 (instrument_builtin_call): Pass gimple iterator to
12086 get_mem_refs_of_builtin_call.
12087 (last_alloca_addr): New global.
12088 * asan.h (asan_emit_allocas_unpoison): Declare.
12089 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12090 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12091 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12092 if function calls alloca.
12093 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12094 * gimple-fold.h (replace_call_with_value): Declare.
12095 * internal-fn.c: Include asan.h.
12096 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12097 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12099 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12101 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12102 (C_SELFTEST_FLAGS): New.
12103 (CPP_SELFTEST_FLAGS): New.
12104 (SELFTEST_DEPS): New, from deps of s-selftest.
12105 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12106 (CPP_SELFTEST_DEPS): New.
12107 (selftest): Add dependency on s-selftest-c++.
12108 (s-selftest): Rename to...
12109 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12110 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12111 than SELFTEST_FLAGS.
12112 (selftest-gdb): Rename to...
12113 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12115 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12116 (selftest-valgrind): Rename to...
12117 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12119 (selftest-valgrind): Reintroduce as an alias for
12120 selftest-c-valgrind.
12121 (s-selftest-c++): New.
12122 (selftest-c++-gdb): New.
12123 (selftest-c++-valgrind): New.
12125 2017-07-06 Olivier Hainque <hainque@adacore.com>
12127 * gcc.c (process_command): When deciding if undefined variables
12128 should be ignored when processing specs, accept "gcc -v" as well.
12130 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12132 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12133 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12135 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12137 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12138 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12139 * config/arm/arm-tables.opt: Regenerate.
12140 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12142 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12144 2017-07-06 Carl Love <cel@us.ibm.com>
12146 * ChangeLog: Clean up from mid air collision
12148 2017-07-06 Carl Love <cel@us.ibm.com>
12150 * config/rs6000/rs6000-c.c: Add support for built-in functions
12151 vector signed int vec_subc (vector signed int, vector signed int);
12152 vector signed __int128 vec_subc (vector signed __int128,
12153 vector signed __int128);
12154 vector unsigned __int128 vec_subc (vector unsigned __int128,
12155 vector unsigned __int128);
12156 vector signed int vec_sube (vector signed int, vector signed int,
12157 vector signed int);
12158 vector unsigned int vec_sube (vector unsigned int,
12159 vector unsigned int,
12160 vector unsigned int);
12161 vector signed __int128 vec_sube (vector signed __int128,
12162 vector signed __int128,
12163 vector signed__int128);
12164 vector unsigned __int128 vec_sube (vector unsigned __int128,
12165 vector unsigned __int128,
12166 vector unsigned __int128);
12167 vector signed int vec_subec (vector signed int, vector signed int,
12168 vector signed int);
12169 vector unsigned int vec_subec (vector unsigned int,
12170 vector unsigned int,
12171 vector unsigned int);
12172 vector signed __int128 vec_subec (vector signed __int128,
12173 vector signed __int128,
12174 vector signed__int128);
12175 vector unsigned __int128 vec_subec (vector unsigned __int128,
12176 vector unsigned __int128,
12177 vector unsigned __int128);
12178 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12179 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12180 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12181 BU_ALTIVEC_OVERLOAD_X definitions.
12182 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12183 * doc/extend.texi: Update the built-in documentation file for the new
12184 built-in functions.
12186 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12189 * diagnostic-show-locus.c (layout::layout): Use start and finish
12190 spelling location for the start and finish of each range.
12191 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12192 Add unused aspect param.
12193 * input.c (expand_location_1): Add "aspect" param, and use it
12194 to access the correct part of the location.
12195 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12197 (expand_location_to_spelling_point): Likewise.
12198 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12199 param, and pass it to expand_location_1.
12201 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12203 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12204 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12205 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12206 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12207 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12208 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12209 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12210 _mm_maskz_getmant_ss): New intrinsics.
12211 (__builtin_ia32_getexpss128_mask): Changed to ...
12212 __builtin_ia32_getexpss128_round ... this.
12213 (__builtin_ia32_getexpsd128_mask): Changed to ...
12214 __builtin_ia32_getexpsd128_round ... this.
12215 * config/i386/i386-builtin-types.def
12216 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12217 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12218 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12219 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12220 __builtin_ia32_getmantss_mask_round): New builtins.
12221 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12222 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12223 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12224 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12225 * config/i386/sse.md
12226 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12227 avx512f_sgetexp<mode><mask_scalar_name>
12228 <round_saeonly_scalar_name> ... this.
12229 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12230 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12231 vgetexp<ssescalarmodesuffix>
12232 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12233 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12234 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12235 avx512f_vgetmant<mode><mask_scalar_name>
12236 <round_saeonly_scalar_name> ... this.
12237 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12238 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12239 vgetmant<ssescalarmodesuffix>
12240 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12241 %0<mask_scalar_operand4>, %1, %2
12242 <round_saeonly_scalar_mask_op4>, %3} ... this.
12243 * config/i386/subst.md (mask_scalar_operand4,
12244 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12245 round_saeonly_scalar_nimm_predicate): New subst attributes.
12247 2017-07-06 Julia Koval <julia.koval@intel.com>
12249 * config/i386/i386.c (ix86_erase_embedded_rounding):
12250 Remove code for old rounding pattern.
12252 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12254 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12256 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12258 * doc/sourcebuild.texi (Test Directives, Variants of
12259 dg-require-support): Add documentation for dg-require-stack-check.
12261 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12263 * config/i386/subst.md (mask_scalar, round_scalar,
12264 round_saeonly_scalar): New meta-templates.
12265 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12266 round_scalar_mask_operand3, round_scalar_mask_op3,
12267 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12268 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12269 round_saeonly_scalar_constraint,
12270 round_saeonly_scalar_prefix): New subst attribute.
12271 * config/i386/sse.md
12272 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12273 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12274 <round_scalar_name> ... this.
12275 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12276 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12277 <round_scalar_name> ... this.
12278 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12279 <sse>_vm<code><mode>3<mask_scalar_name>
12280 <round_saeonly_scalar_name> ... this.
12281 (v<plusminus_mnemonic><ssescalarmodesuffix>
12282 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12283 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12284 v<plusminus_mnemonic><ssescalarmodesuffix>
12285 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12286 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12287 (v<multdiv_mnemonic><ssescalarmodesuffix>
12288 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12289 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12290 v<multdiv_mnemonic><ssescalarmodesuffix>
12291 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12292 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12293 (v<maxmin_float><ssescalarmodesuffix>
12294 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12295 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12296 v<maxmin_float><ssescalarmodesuffix>
12297 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12298 %0<mask_scalar_operand3>, %1, %<iptr>2
12299 <round_saeonly_scalar_mask_op3>} ... this.
12301 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12303 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12304 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12306 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12307 Alan Hayward <alan.hayward@arm.com>
12308 David Sherwood <david.sherwood@arm.com>
12310 * combine.c (simplify_if_then_else): Remove "enum" before
12312 * compare-elim.c (can_eliminate_compare): Likewise.
12313 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12315 (aarch64_lookup_simd_builtin_type): Likewise.
12316 (aarch64_simd_builtin_type): Likewise.
12317 (aarch64_init_simd_builtin_types): Likewise.
12318 (aarch64_simd_expand_args): Likewise.
12319 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12321 (aarch64_reverse_mask): Likewise.
12322 (aarch64_simd_emit_reg_reg_move): Likewise.
12323 (aarch64_gen_adjusted_ldpstp): Likewise.
12324 (aarch64_ccmp_mode_to_code): Likewise.
12325 (aarch64_operands_ok_for_ldpstp): Likewise.
12326 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12327 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12329 (aarch64_min_divisions_for_recip_mul): Likewise.
12330 (aarch64_reassociation_width): Likewise.
12331 (aarch64_get_condition_code_1): Likewise.
12332 (aarch64_simd_emit_reg_reg_move): Likewise.
12333 (aarch64_simd_attr_length_rglist): Likewise.
12334 (aarch64_reverse_mask): Likewise.
12335 (aarch64_operands_ok_for_ldpstp): Likewise.
12336 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12337 (aarch64_gen_adjusted_ldpstp): Likewise.
12338 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12340 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12341 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12342 (arm_lookup_simd_builtin_type): Likewise.
12343 (arm_simd_builtin_type): Likewise.
12344 (arm_init_simd_builtin_types): Likewise.
12345 (arm_expand_builtin_args): Likewise.
12346 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12347 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12348 (ft32_setup_incoming_varargs): Likewise.
12349 (ft32_function_arg): Likewise.
12350 (ft32_function_arg_advance): Likewise.
12351 (ft32_pass_by_reference): Likewise.
12352 (ft32_arg_partial_bytes): Likewise.
12353 (ft32_valid_pointer_mode): Likewise.
12354 (ft32_addr_space_pointer_mode): Likewise.
12355 (ft32_addr_space_legitimate_address_p): Likewise.
12356 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12358 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12359 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12360 (iamcu_alignment): Likewise.
12361 (canonicalize_vector_int_perm): Likewise.
12362 (ix86_noce_conversion_profitable_p): Likewise.
12363 (ix86_mpx_bound_mode): Likewise.
12364 (ix86_operands_ok_for_move_multiple): Likewise.
12365 * config/microblaze/microblaze-protos.h
12366 (microblaze_expand_conditional_branch_reg): Likewise.
12367 * config/microblaze/microblaze.c
12368 (microblaze_expand_conditional_branch_reg): Likewise.
12369 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12371 (rs6000_reassociation_width): Likewise.
12372 (rs6000_invalid_binary_op): Likewise.
12373 (fusion_p9_p): Likewise.
12374 (emit_fusion_p9_load): Likewise.
12375 (emit_fusion_p9_store): Likewise.
12376 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12378 (riscv_hard_regno_mode_ok_p): Likewise.
12379 (riscv_address_insns): Likewise.
12380 (riscv_split_symbol): Likewise.
12381 (riscv_legitimize_move): Likewise.
12382 (riscv_function_value): Likewise.
12383 (riscv_hard_regno_nregs): Likewise.
12384 (riscv_expand_builtin): Likewise.
12385 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12386 (riscv_build_integer): Likewise.
12387 (riscv_split_integer): Likewise.
12388 (riscv_legitimate_constant_p): Likewise.
12389 (riscv_cannot_force_const_mem): Likewise.
12390 (riscv_regno_mode_ok_for_base_p): Likewise.
12391 (riscv_valid_base_register_p): Likewise.
12392 (riscv_valid_offset_p): Likewise.
12393 (riscv_valid_lo_sum_p): Likewise.
12394 (riscv_classify_address): Likewise.
12395 (riscv_legitimate_address_p): Likewise.
12396 (riscv_address_insns): Likewise.
12397 (riscv_load_store_insns): Likewise.
12398 (riscv_force_binary): Likewise.
12399 (riscv_split_symbol): Likewise.
12400 (riscv_force_address): Likewise.
12401 (riscv_legitimize_address): Likewise.
12402 (riscv_move_integer): Likewise.
12403 (riscv_legitimize_const_move): Likewise.
12404 (riscv_legitimize_move): Likewise.
12405 (riscv_address_cost): Likewise.
12406 (riscv_subword): Likewise.
12407 (riscv_output_move): Likewise.
12408 (riscv_canonicalize_int_order_test): Likewise.
12409 (riscv_emit_int_order_test): Likewise.
12410 (riscv_function_arg_boundary): Likewise.
12411 (riscv_pass_mode_in_fpr_p): Likewise.
12412 (riscv_pass_fpr_single): Likewise.
12413 (riscv_pass_fpr_pair): Likewise.
12414 (riscv_get_arg_info): Likewise.
12415 (riscv_function_arg): Likewise.
12416 (riscv_function_arg_advance): Likewise.
12417 (riscv_arg_partial_bytes): Likewise.
12418 (riscv_function_value): Likewise.
12419 (riscv_pass_by_reference): Likewise.
12420 (riscv_setup_incoming_varargs): Likewise.
12421 (riscv_print_operand): Likewise.
12422 (riscv_elf_select_rtx_section): Likewise.
12423 (riscv_save_restore_reg): Likewise.
12424 (riscv_for_each_saved_reg): Likewise.
12425 (riscv_register_move_cost): Likewise.
12426 (riscv_hard_regno_mode_ok_p): Likewise.
12427 (riscv_hard_regno_nregs): Likewise.
12428 (riscv_class_max_nregs): Likewise.
12429 (riscv_memory_move_cost): Likewise.
12430 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12431 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12432 (rl78_addr_space_address_mode): Likewise.
12433 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12435 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12436 (rs6000_reassociation_width): Likewise.
12437 (rs6000_invalid_binary_op): Likewise.
12438 (fusion_p9_p): Likewise.
12439 (emit_fusion_p9_load): Likewise.
12440 (emit_fusion_p9_store): Likewise.
12441 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12442 (ok_for_simple_move_operands): Likewise.
12443 (ok_for_simple_move_strict_operands): Likewise.
12444 (ok_for_simple_arith_logic_operands): Likewise.
12445 (visium_legitimize_reload_address): Likewise.
12446 (visium_select_cc_mode): Likewise.
12447 (output_cbranch): Likewise.
12448 (visium_split_double_move): Likewise.
12449 (visium_expand_copysign): Likewise.
12450 (visium_expand_int_cstore): Likewise.
12451 (visium_expand_fp_cstore): Likewise.
12452 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12453 (visium_function_arg): Likewise.
12454 (visium_function_arg_advance): Likewise.
12455 (visium_libcall_value): Likewise.
12456 (visium_setup_incoming_varargs): Likewise.
12457 (visium_legitimate_constant_p): Likewise.
12458 (visium_legitimate_address_p): Likewise.
12459 (visium_legitimize_address): Likewise.
12460 (visium_secondary_reload): Likewise.
12461 (visium_register_move_cost): Likewise.
12462 (visium_memory_move_cost): Likewise.
12463 (prepare_move_operands): Likewise.
12464 (ok_for_simple_move_operands): Likewise.
12465 (ok_for_simple_move_strict_operands): Likewise.
12466 (ok_for_simple_arith_logic_operands): Likewise.
12467 (visium_function_value_1): Likewise.
12468 (rtx_ok_for_offset_p): Likewise.
12469 (visium_legitimize_reload_address): Likewise.
12470 (visium_split_double_move): Likewise.
12471 (visium_expand_copysign): Likewise.
12472 (visium_expand_int_cstore): Likewise.
12473 (visium_expand_fp_cstore): Likewise.
12474 (visium_split_cstore): Likewise.
12475 (visium_select_cc_mode): Likewise.
12476 (visium_split_cbranch): Likewise.
12477 (output_cbranch): Likewise.
12478 (visium_print_operand_address): Likewise.
12479 * expmed.c (flip_storage_order): Likewise.
12480 * expmed.h (emit_cstore): Likewise.
12481 (flip_storage_order): Likewise.
12482 * genrecog.c (validate_pattern): Likewise.
12483 * hsa-gen.c (gen_hsa_addr): Likewise.
12484 * internal-fn.c (expand_arith_overflow): Likewise.
12485 * ira-color.c (allocno_copy_cost_saving): Likewise.
12486 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12487 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12488 (process_invariant_for_inheritance): Likewise.
12489 * lra-eliminations.c (move_plus_up): Likewise.
12490 * omp-low.c (lower_oacc_reductions): Likewise.
12491 * simplify-rtx.c (simplify_subreg): Likewise.
12492 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12493 (TARGET_CHKP_BOUND_MODE): Likewise..
12494 * targhooks.c (default_chkp_bound_mode): Likewise.
12495 (default_setup_incoming_vararg_bounds): Likewise.
12496 * targhooks.h (default_chkp_bound_mode): Likewise.
12497 (default_setup_incoming_vararg_bounds): Likewise.
12498 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12499 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12500 (have_whole_vector_shift): Likewise.
12501 * tree-vect-stmts.c (vectorizable_load): Likewise.
12502 * doc/tm.texi: Regenerate.
12504 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12506 Graceful degrade if Binutils PR21472 is not available.
12509 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12510 .rodata in flash test fails.
12511 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12512 * confgure: Regenerate.
12513 * config.in: Regenerate.
12514 * config/avr/avr.c (avr_asm_named_section)
12515 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12516 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12517 (avr_asm_init_sections): Same.
12519 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12521 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12522 (fma<VH:mode>4_intrinsic): Likewise.
12523 (*fmsub<VCVTF:mode>4): Likewise.
12524 (*fmsub<VH:mode>4_intrinsic): Likewise.
12526 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12529 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12530 Don't depend on "optimize > 0".
12531 (out_movhi_r_mr, out_movqi_mr_r): Same.
12532 (out_movhi_mr_r, out_movqi_r_mr): Same.
12533 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12534 io_address_operand on "optimize > 0".
12536 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12538 * tree-loop-distribution.c: Add general explanantion on the pass.
12539 (generate_loops_for_partition): Mark distributed loop.
12540 (pg_add_dependence_edges): New parameter. Handle alias data
12541 dependence specially and record it in the parameter if asked.
12542 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12543 (init_partition_graph_vertices, add_partition_graph_edge): New.
12544 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12545 (free_partition_graph_vdata, build_partition_graph): New.
12546 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12547 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12548 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12549 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12550 (version_for_distribution_p, finalize_partitions): New.
12551 (distribute_loop): Handle alias data dependence specially. Factor
12552 out loop fusion code as functions and call these functions.
12554 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12556 * tree-loop-distribution.c (classify_partition): New parameter and
12557 better handle reduction statement.
12558 (rdg_build_partitions): Revise comment.
12559 (distribute_loop): Compute statements in all partitions and pass it
12560 to classify_partition.
12562 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12564 * tree-loop-distribution.c (enum partition_type): New.
12565 (struct partition): New field type.
12566 (partition_merge_into): Add parameter. Update partition type.
12567 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12568 (build_rdg_partition_for_vertex): Compute partition type.
12569 (rdg_build_partitions): Dump partition type.
12570 (distribute_loop): Update calls to partition_merge_into.
12572 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12574 * tree-loop-distribution.c (struct ddr_hasher): New.
12575 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12577 (classify_partition): Call get_data_dependence.
12578 (pg_add_dependence_edges): Ditto.
12579 (distribute_loop): Release data dependence hash table.
12581 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12583 * tree-loop-distribution.c (ref_base_address): Delete.
12584 (similar_memory_accesses): Rename ...
12585 (share_memory_accesses): ... to this. Check if partitions access
12586 the same memory reference.
12587 (distribute_loop): Call share_memory_accesses.
12589 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12591 * tree-loop-distribution.c (struct partition): New field recording
12592 its data reference.
12593 (partition_alloc, partition_free): Init and release data refs.
12594 (partition_merge_into): Merge data refs.
12595 (build_rdg_partition_for_vertex): Collect data refs for partition.
12596 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12598 (distribute_loop): Remve data refs from vertice data of partition
12601 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12603 * tree-loop-distribution.c (params.h): Include header file.
12604 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12605 (datarefs_vec): New global var.
12606 (create_rdg_vertices): Use datarefs_vec directly.
12607 (free_rdg): Don't free data references.
12608 (build_rdg): Update use. Don't free data references.
12609 (distribute_loop): Compute global variable for data references.
12610 Bail out if there are too many data references.
12612 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12614 * tree-loop-distribution.c (loop_nest): New global var.
12615 (build_rdg): Use loop directly, rather than loop nest.
12616 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12618 (distribute_loop): Compute global variable loop nest. Update use.
12620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12622 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12623 (partition_merge_into): New parameter. Dump reason for fusion.
12624 (distribute_loop): Update use of partition_merge_into.
12626 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12628 * tree-loop-distribution.c (bb_top_order_index): New.
12629 (bb_top_order_index_size, bb_top_order_cmp): New.
12630 (stmts_from_loop): Use topological order.
12631 (pass_loop_distribution::execute): Compute and release topological
12632 order for basic blocks.
12634 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12636 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12639 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12641 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12642 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12643 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12644 * internal-fn.def (LOOP_DIST_ALIAS): New.
12645 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12646 (fold_loop_internal_call): ... this.
12647 (vect_loop_dist_alias_call): New function.
12648 (set_uid_loop_bbs): Call fold_loop_internal_call.
12649 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12652 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12655 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12656 Require dead FLAGS_REG at the beginning of a peephole.
12658 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12661 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12662 arguments in the call to __builtin_ia32_sbb_u32.
12663 (_subborrow_u64): Swap _X and _Y arguments in the call to
12664 __builtin_ia32_sbb_u64.
12666 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12669 * tree-vrp.c (compare_assert_loc): Turn into a function template
12670 with stable template parameter. Only test if a->e is NULL,
12671 !a->e == !b->e has been verified already. Use e == NULL or
12672 e != NULL instead of e or ! e tests. If stable is true, don't use
12673 iterative_hash_expr, on the other side allow a or b or both NULL
12674 and sort the NULLs last.
12675 (process_assert_insertions): Sort using compare_assert_loc<false>
12676 instead of compare_assert_loc, later sort using
12677 compare_assert_loc<true> before calling process_assert_insertions_for
12678 in a loop. Use break instead of continue once seen NULL pointer.
12680 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12682 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12683 Cortex-R7 and Cortex-R8 processors.
12685 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12687 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12688 uninitialized while src is not.
12690 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12692 * common/config/arm/arm-common.c: Adjust include path for
12694 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12695 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12696 (arm-cpu-data.h): Likewise.
12697 (arm-cpu-cdata.h): Likewise.
12698 * config/arm/arm-cpu.h: Delete.
12699 * config/arm/arm-cpu-cdata.h: Delete.
12700 * config/arm/arm-cpu-data.h: Delete.
12702 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12704 * config/arm/arm-cpus.in (cortex-a55): New.
12705 (cortex-a75): Likewise.
12706 (cortex-a75.cortex-a55): Likewise.
12707 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12709 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12710 * config/arm/arm-cpu-cdata.h: Regenerate.
12711 * config/arm/arm-cpu-data.h: Regenerate.
12712 * config/arm/arm-cpu.h: Regenerate.
12713 * config/arm/arm-tables.opt: Regenerate.
12714 * config/arm/arm-tune.md: Regenerate.
12716 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12718 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12720 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12722 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12725 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12727 PR tree-optimization/81292
12728 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12729 full_string_p, also call adjust_related_strinfos if the adjustment
12730 is simple, otherwise invalidate related strinfos.
12732 2017-07-04 Martin Liska <mliska@suse.cz>
12735 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12736 newly created variable as DECL_IGNORED_P.
12738 2017-07-04 Martin Liska <mliska@suse.cz>
12741 * ipa-inline.c (inline_small_functions):
12742 Use xstrdup_for_dump.
12744 2017-07-04 Tom de Vries <tom@codesourcery.com>
12746 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12748 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12751 * config/darwin.c (darwin_function_switched_text_sections):
12754 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12756 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12758 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12760 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12762 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12764 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12765 min_profitable_iters, and th as inclusive lower bounds.
12766 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12767 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12768 for min_profitable_iters and min_profitable_estimate.
12769 (vect_transform_loop): Treat th as an inclusive lower bound.
12770 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12772 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12775 * config/darwin.c (darwin_function_switched_text_sections):
12776 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12777 in two pieces, and suppress the use of buf.
12779 2017-07-03 Nathan Sidwell <nathan@acm.org>
12781 * hash-table.h (hash_table_mod1): Fix indentation.
12783 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12785 PR middle-end/81290
12786 * predict.c (force_edge_cold): Be more careful about propagation
12788 * profile-count.h (profile_probability::guessed,
12789 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12791 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12793 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12795 * doc/invoke.texi (rcpc architecture extension): Document it.
12797 2017-07-03 Richard Biener <rguenther@suse.de>
12799 PR tree-optimization/60510
12800 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12801 the scalar reduction PHI and use it.
12802 (vectorizable_reduction): Properly guard the single_defuse_cycle
12803 path for non-SLP reduction chains where we cannot use it.
12804 Rework reduc_def/index and vector type deduction. Rework
12805 vector operand gathering during reduction op code-gen.
12806 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12807 chains dissolve the chain and leave it to non-SLP reduction
12810 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12812 * tree-data-ref.h (dr_alignment): Declare.
12813 * tree-data-ref.c (dr_alignment): New function.
12814 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12815 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12817 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12819 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12821 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12822 and base_misalignment fields.
12823 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12824 * tree-data-ref.c: Include builtins.h.
12825 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12826 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12827 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12828 * tree-vect-data-refs.c: Include tree-cfg.h.
12829 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12830 fields instead of calculating an alignment here.
12831 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12832 innermost_loop_behavior fields.
12834 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12836 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12838 (DR_STEP_ALIGNMENT): New macro.
12839 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12840 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12841 (create_data_ref): Print it.
12842 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12843 to tell whether the step preserves vector (mis)alignment.
12844 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12845 Move the check for an integer step and generalise to all INTEGER_CST.
12846 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12847 Print the outer step alignment.
12849 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12851 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12852 with offset_alignment.
12853 (DR_ALIGNED_TO): Delete.
12854 (DR_OFFSET_ALIGNMENT): New macro.
12855 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12856 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12857 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12858 (create_data_ref): Likewise.
12859 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12860 (vect_analyze_data_refs): Likewise.
12861 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12862 creating dummy innermost behavior.
12864 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12866 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12867 with a "innermost_loop_behavior *" and refeence tree.
12868 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12869 (create_data_ref): Update call accordingly.
12870 * tree-predcom.c (find_looparound_phi): Likewise.
12872 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12874 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12875 fields with dr_wrt_vec_loop.
12876 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12877 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12878 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12879 (vect_dr_behavior): New function.
12880 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12881 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12882 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12883 track whether the step preserves the misalignment.
12884 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12885 Use vect_dr_behavior.
12886 (vect_setup_realignment): Update call accordingly.
12887 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12888 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12889 call to vect_create_addr_base_for_vector_ref.
12890 (vect_create_cond_for_align_checks): Likewise.
12891 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12892 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12893 (vect_recog_mask_conversion_pattern): Likewise.
12894 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12895 (new_stmt_vec_info): Remove redundant zeroing.
12897 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12899 * common/config/arm/arm-common.c (arm_be8_option): New function.
12900 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12901 (ISA_ARMv6): Add isa_bit_be8.
12902 * config/arm/arm.h (arm_be8_option): Add prototype.
12903 (BE8_SPEC_FUNCTION): New define.
12904 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12905 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12906 (mlittle-endian): Similarly.
12907 (mbe8, mbe32): New options.
12908 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12909 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12911 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12913 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12915 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12917 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12918 (cleanup_tree_cfg_bb): Use it.
12919 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12921 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12923 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12926 * loop-doloop.c (add_test): Update profile.
12928 2017-07-03 Martin Liska <mliska@suse.cz>
12931 * sanopt.c (rewrite_usage_of_param): New function.
12932 (sanitize_rewrite_addressable_params): Likewise.
12933 (pass_sanopt::execute): Call rewrite_usage_of_param.
12935 2017-07-03 Richard Biener <rguenther@suse.de>
12937 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12938 back to using VIEW_CONVERT_EXPR.
12940 2017-07-03 Martin Liska <mliska@suse.cz>
12943 * doc/extend.texi: Document when a resolver function is
12944 generated for target_clones.
12946 2017-07-03 Martin Liska <mliska@suse.cz>
12948 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12949 and shadow memory of auto variables which are subject of
12950 use-after-scope sanitization.
12951 (asan_expand_mark_ifn): Add do set only when is_poison.
12953 2016-07-03 Richard Biener <rguenther@suse.de>
12955 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12957 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12958 (vectorizable_reduction): Perform reduction PHI creation when
12959 visiting a reduction PHI and adjust and simplify code generation
12960 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12961 (vect_transform_loop): Visit reduction PHIs.
12962 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12963 defs into the SLP tree.
12964 (vect_build_slp_tree): Reduction defs terminate the recursion.
12965 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12967 (vect_get_vec_defs_for_stmt_copy): Export.
12968 (vect_get_vec_defs): Likewise.
12969 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12971 (vect_get_vec_defs_for_stmt_copy): Declare.
12972 (vect_get_vec_defs): Likewise.
12974 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12976 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12977 parameter with a "loop" parameter and use it instead of the
12978 loop containing DR_STMT. Don't check simple_iv when doing
12979 BB analysis. Describe the two analysis modes in the comment.
12981 2017-07-03 Tom de Vries <tom@codesourcery.com>
12983 PR tree-optimization/69468
12984 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12985 (find_same_succ_bb): Handle ignore_edge_flags.
12987 2017-07-03 Tom de Vries <tom@codesourcery.com>
12989 PR tree-optimization/81192
12990 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12992 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12994 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12996 2017-07-03 Tom de Vries <tom@codesourcery.com>
12998 PR tree-optimization/81192
12999 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13000 BB_SAME_SUCC (bb) == NULL.
13002 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13004 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13007 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13009 * dumpfile.c: Include profile-count.h
13010 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13012 (insert_cond_bb): Update profile.
13013 * tree-cfg.h (insert_cond_bb): Update prototype.
13014 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13015 * tree-dump.c: Do not include tree-cfg.
13017 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13019 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13021 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13023 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13026 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13028 * tree-complex.c (expand_complex_div_wide): update profile.
13030 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13031 Alan Hayward <alan.hayward@arm.com>
13032 David Sherwood <david.sherwood@arm.com>
13034 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13035 (CORETYPES_H): New define.
13036 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13037 (insn-modes-inline.h, s-modes-inline-h): New rules.
13038 (generated_files): Add insn-modes-inline.h.
13039 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13040 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13041 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13042 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13043 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13044 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13045 (build/gencodes.o, build/genconditions.o): Likewise.
13046 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13047 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13048 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13049 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13050 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13051 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13052 * coretypes.h: Include everything up to real.h for generators.
13053 Include insn-modes.h first. Include wide-int-print.h after
13054 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13055 * machmode.h: Don't include insn-modes.h here.
13056 * function-tests.c: Remove includes of signop.h, machmode.h,
13057 double-int.h and wide-int.h.
13059 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13061 * optc-save-gen.awk: Likewise.
13062 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13063 * godump.c: Remove include of wide-int-print.h.
13064 * pretty-print.h: Likewise.
13065 * wide-int-print.cc: Likewise.
13066 * wide-int.cc: Likewise.
13067 * hash-map-tests.c: Remove include of signop.h.
13068 * hash-set-tests.c: Likewise.
13069 * rtl-tests.c: Likewise.
13070 * mkconfig.sh: Remove include of machmode.h.
13071 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13073 (emit_insn_modes_inline_h): ...this new function. Emit the code
13074 into an insn-modes-inline.h header file, adding appropriate
13075 include guards and end comments.
13076 (emit_insn_modes_c_header): Remove include of machmode.h.
13077 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13079 (main): Handle -i flag and call emit_insn_modes_inline_h when
13082 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13084 * tree-ssa-strlen.c (strinfo): Rename the length field to
13085 nonzero_chars. Add a full_string_p field.
13086 (compare_nonzero_chars, zero_length_string_p): New functions.
13087 (get_addr_stridx): Add an offset_out parameter.
13088 Use compare_nonzero_chars.
13089 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13090 (new_strinfo): Update after above changes to strinfo.
13091 (set_endptr_and_length): Set full_string_p.
13092 (get_string_length): Update after above changes to strinfo.
13093 (unshare_strinfo): Update call to new_strinfo.
13094 (maybe_invalidate): Likewise.
13095 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13096 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13097 as a uhwi instead of an shwi. Update after above changes to
13098 strinfo and new_strinfo.
13099 (zero_length_string): Assert that chainsi contains full strings.
13100 Use zero_length_string_p. Update call to new_strinfo.
13101 (adjust_related_strinfos): Update after above changes to strinfo.
13102 Copy full_string_p from origsi.
13103 (adjust_last_stmt): Use zero_length_string_p.
13104 (handle_builtin_strlen): Update after above changes to strinfo and
13105 new_strinfo. Install the lhs as the string length if the previous
13106 entry didn't describe a full string.
13107 (handle_builtin_strchr): Update after above changes to strinfo
13109 (handle_builtin_strcpy): Likewise.
13110 (handle_builtin_strcat): Likewise.
13111 (handle_builtin_malloc): Likewise.
13112 (handle_pointer_plus): Likewise.
13113 (handle_builtin_memcpy): Likewise. Track nonzero characters
13114 that aren't necessarily followed by a nul terminator.
13115 (handle_char_store): Likewise.
13117 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13119 PR tree-optimization/80769
13120 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13121 for malloc and calloc. Document the new invariant that all related
13122 strinfos have delayed lengths or none do.
13123 (verify_related_strinfos): Move earlier in file.
13124 (set_endptr_and_length): New function, split out from...
13125 (get_string_length): ...here. Also set the lengths of related
13127 (zero_length_string): Assert that chainsi has known (rather than
13129 (adjust_related_strinfos): Likewise.
13131 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13133 PR tree-optimization/81136
13134 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13135 assert that two references with the same misalignment have the same
13136 compile-time misalignment if those compile-time misalignments
13139 2017-07-01 Andi Kleen <ak@linux.intel.com>
13141 * print-tree.c (print_node): Print all attributes.
13143 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13145 * cfg.c (scale_bbs_frequencies): New function.
13146 * cfg.h (scale_bbs_frequencies): Declare it.
13147 * cfgloopanal.c (single_likely_exit): Cleanup.
13148 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13150 (scale_loop_profile): Likewise.
13151 (loop_version): Likewise.
13152 (create_empty_loop_on_edge): Update.
13153 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13154 scale_loop_frequencies, scale_loop_profile, loopify,
13155 loop_version): Update prototypes.
13156 * modulo-sched.c (sms_schedule): Update.
13157 * predict.c (unlikely_executed_edge_p): Also check probability.
13158 (probably_never_executed_edge_p): Fix typo.
13159 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13160 * tree-parloops.c (gen_parallel_loop): Update.
13161 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13162 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13163 * tree-ssa-loop-split.c (split_loop): Update.
13164 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13165 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13166 (vect_loop_versioning): Update.
13167 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13169 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13171 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13173 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13175 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13176 to keep profile consistent.
13178 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13180 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13181 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13182 * profile-count.h (max_safe_multiplier): Make unsigned.
13183 (profile_count::guessed_zero): New.
13185 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13187 * bb-reorder.c (fix_up_crossing_landing_pad,
13188 fix_crossing_conditional_branches): Use make_single_succ_edge
13189 to keep profile consistent.
13191 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13193 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13196 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13199 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13200 the right scopes, make sure cond_jump isn't preserved between multiple
13201 iterations. Search for fallthru edge whenever there are 3+ edges and
13202 use find_fallthru_edge for it.
13204 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13206 Patch by Alexander Monakov <amonakov@ispras.ru>
13207 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13208 probabilities consistently.
13210 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13212 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13213 emit_cmp_and_jump_insns.
13215 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13218 * tree-inline.c (expand_call_inline): Combine profile statuses.
13220 2017-06-30 Andrew Pinski <apinski@cavium.com>
13222 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13223 fold_stmt returned true.
13225 2017-06-30 Nathan Sidwell <nathan@acm.org>
13227 * ggc.h (empty_string): Delete.
13228 * cfgexpand.c (expand_asm_stmt): Use plain "".
13229 * optabs.c (expand_asm_memory_barrier): Likewise.
13230 * stringpool.c (empty_string): Delete.
13231 (digit_vector, digit_string): Delete.
13232 (ggc_alloc_string): Use plain "", don't optimize single digit
13233 strings. Use ggc_alloc_atomic.
13235 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13237 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13238 comparison set and one other set, use the cost of the non-comparison
13241 2017-06-30 Nathan Sidwell <nathan@acm.org>
13243 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13246 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13248 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13249 loops. Remove now unneeded calls to gimple_switch_set_label() that
13250 just set removed labels to NULL_TREE.
13252 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13254 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13255 (set_range_info): ...here. Only call set_range_info_raw if domain
13257 (set_nonzero_bits): Call set_range_info_raw.
13258 * tree-ssanames.h (set_range_info_raw): New.
13260 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13263 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13264 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13265 of nonimmediate_operand and <store_mask_constraint> instead of m
13266 for the input operand. For V8FI iterator, always split if input
13267 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13268 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13269 <store_mask_predicate> instead of register_operand and
13270 <store_mask_constraint> instead of v for the input operand. Make
13271 sure both operands aren't MEMs for if not <mask_applied>.
13273 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13275 * lto-wrapper.c (copy_file) Close both file descriptors before
13278 2017-06-30 Martin Liska <mliska@suse.cz>
13281 * multiple_target.c (create_dispatcher_calls): Make ifunc
13282 also for function that don't have calls or are not referenced.
13284 2017-06-30 Richard Biener <rguenther@suse.de>
13286 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13287 analyze the first scalar stmt. Move vector type computation
13288 for the BB case here from ...
13289 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13290 live operation processing in the SLP case properly.
13292 2017-06-30 Richard Biener <rguenther@suse.de>
13294 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13296 2017-06-30 Martin Liska <mliska@suse.cz>
13299 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13300 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13302 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13304 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13307 2017-06-30 Richard Biener <rguenther@suse.de>
13309 PR tree-optimization/81249
13310 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13311 condition reduction result to original scalar type.
13313 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13315 * profile-count.h (enum profile_quality): Fix typos and whitespace
13318 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13320 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13321 type for branch probabilities.
13323 2017-06-29 Julian Brown <julian@codesourcery.com>
13324 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13326 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13327 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13328 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13329 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13331 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13333 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13334 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13335 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13336 CC usage from generic code to here.
13337 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13338 CC usage into the target macros.
13340 2017-06-29 Maya Rashish <coypu@sdf.org>
13342 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13345 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13347 * arm/arm-builtins.c: Include profile-count.h
13348 * except.c (sjlj_emit_function_enter): Use
13349 profile_probability::unlikely.
13351 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13353 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13354 and tocrel_offset be pointer args rather than implicitly using
13356 (legitimate_constant_pool_address_p, rs6000_emit_move,
13357 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13358 tocrel_offset and use in toc_relative_expr_p call.
13359 (print_operand, print_operand_address): Use static tocrel_base_oac
13360 and tocrel_offset_oac.
13361 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13364 2017-06-29 Maya Rashish <coypu@sdf.org>
13366 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13368 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13370 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13371 objects, take into account only the alignment of 'op0' and 'mode1' if
13374 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13376 * ccmp.c (ccmp_tree_comparison_p): New function.
13377 (ccmp_candidate_p): Update to use above function.
13378 (get_compare_parts): New function.
13379 (expand_ccmp_next): Update to use new functions.
13380 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13382 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13383 take mode as argument.
13384 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13385 * expr.c (expand_expr_real_1): Pass mode as argument.
13387 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13389 * combine.c (combine_instructions): Print insns to dump_file, together
13392 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13394 * asan.c (asan_emit_stack_protection): Update.
13395 (create_cond_insert_point): Update.
13396 * auto-profile.c (afdo_propagate_circuit): Update.
13397 * basic-block.h (struct edge_def): Turn probability to
13398 profile_probability.
13399 (EDGE_FREQUENCY): Update.
13400 * bb-reorder.c (find_traces_1_round): Update.
13401 (better_edge_p): Update.
13402 (sanitize_hot_paths): Update.
13403 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13404 (make_single_succ_edge): Update.
13405 (check_bb_profile): Update.
13406 (dump_edge_info): Update.
13407 (update_bb_profile_for_threading): Update.
13408 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13409 probabilitycount to 0.
13410 * cfgbuild.c (compute_outgoing_frequencies): Update.
13411 * cfgcleanup.c (try_forward_edges): Update.
13412 (outgoing_edges_match): Update.
13413 (try_crossjump_to_edge): Update.
13414 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13415 (expand_gimple_tailcall): Update.
13416 (construct_init_block): Use make_single_succ_edge.
13417 (construct_exit_block): Use make_single_succ_edge.
13418 * cfghooks.c (verify_flow_info): Update.
13419 (redirect_edge_succ_nodup): Update.
13420 (split_edge): Update.
13421 (account_profile_record): Update.
13422 * cfgloopanal.c (single_likely_exit): Update.
13423 * cfgloopmanip.c (scale_loop_profile): Update.
13424 (set_zero_probability): Remove.
13425 (duplicate_loop_to_header_edge): Update.
13426 * cfgloopmanip.h (loop_version): Update prototype.
13427 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13428 (force_nonfallthru_and_redirect): Update.
13429 (update_br_prob_note): Update.
13430 (rtl_verify_edges): Update.
13431 (purge_dead_edges): Update.
13432 (rtl_lv_add_condition_to_bb): Update.
13433 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13434 * cgraphunit.c (init_lowered_empty_function): Update.
13435 (cgraph_node::expand_thunk): Update.
13436 * cilk-common.c: Include profile-count.h
13437 * dojump.c (inv): Remove.
13438 (jumpifnot): Update.
13439 (jumpifnot_1): Update.
13440 (do_jump_1): Update.
13442 (do_jump_by_parts_greater_rtx): Update.
13443 (do_compare_rtx_and_jump): Update.
13444 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13445 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13446 * dwarf2cfi.c: Include profile-count.h
13447 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13448 (sjlj_emit_dispatch_table): Likewise.
13449 * explow.c: Include profile-count.h
13450 * expmed.c (emit_store_flag_force): Update.
13451 (do_cmp_and_jump): Update.
13452 * expr.c (compare_by_pieces_d::generate): Update.
13453 (compare_by_pieces_d::finish_mode): Update.
13454 (emit_block_move_via_loop): Update.
13455 (store_expr_with_bounds): Update.
13456 (store_constructor): Update.
13457 (expand_expr_real_2): Update.
13458 (expand_expr_real_1): Update.
13459 * expr.h (try_casesi, try_tablejump): Update prototypes.
13460 * gimple-pretty-print.c (dump_probability): Update.
13461 (dump_profile): New.
13462 (dump_gimple_label): Update.
13463 (dump_gimple_bb_header): Update.
13464 * graph.c (draw_cfg_node_succ_edges): Update.
13465 * hsa-gen.c (convert_switch_statements): Update.
13466 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13467 (find_if_case_1): Update.
13468 (find_if_case_2): Update.
13469 * internal-fn.c (expand_arith_overflow_result_store): Update.
13470 (expand_addsub_overflow): Update.
13471 (expand_neg_overflow): Update.
13472 (expand_mul_overflow): Update.
13473 (expand_vector_ubsan_overflow): Update.
13474 * ipa-cp.c (good_cloning_opportunity_p): Update.
13475 * ipa-split.c (split_function): Use make_single_succ_edge.
13476 * ipa-utils.c (ipa_merge_profiles): Update.
13477 * loop-doloop.c (add_test): Update.
13478 (doloop_modify): Update.
13479 * loop-unroll.c (compare_and_jump_seq): Update.
13480 (unroll_loop_runtime_iterations): Update.
13481 * lra-constraints.c (lra_inheritance): Update.
13482 * lto-streamer-in.c (input_cfg): Update.
13483 * lto-streamer-out.c (output_cfg): Update.
13484 * mcf.c (adjust_cfg_counts): Update.
13485 * modulo-sched.c (sms_schedule): Update.
13486 * omp-expand.c (expand_omp_for_init_counts): Update.
13487 (extract_omp_for_update_vars): Update.
13488 (expand_omp_ordered_sink): Update.
13489 (expand_omp_for_ordered_loops): Update.
13490 (expand_omp_for_generic): Update.
13491 (expand_omp_for_static_nochunk): Update.
13492 (expand_omp_for_static_chunk): Update.
13493 (expand_cilk_for): Update.
13494 (expand_omp_simd): Update.
13495 (expand_omp_taskloop_for_outer): Update.
13496 (expand_omp_taskloop_for_inner): Update.
13497 * omp-simd-clone.c (simd_clone_adjust): Update.
13498 * optabs.c (expand_doubleword_shift): Update.
13499 (expand_abs): Update.
13500 (emit_cmp_and_jump_insn_1): Update.
13501 (expand_compare_and_swap_loop): Update.
13502 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13503 * predict.c (predictable_edge_p): Update.
13504 (edge_probability_reliable_p): Update.
13505 (set_even_probabilities): Update.
13506 (combine_predictions_for_insn): Update.
13507 (combine_predictions_for_bb): Update.
13508 (propagate_freq): Update.
13509 (estimate_bb_frequencies): Update.
13510 (force_edge_cold): Update.
13511 * profile-count.c (profile_count::dump): Add missing space into dump.
13512 (profile_count::debug): Add newline.
13513 (profile_count::differs_from_p): Explicitly convert to unsigned.
13514 (profile_count::stream_in): Update.
13515 (profile_probability::dump): New member function.
13516 (profile_probability::debug): New member function.
13517 (profile_probability::differs_from_p): New member function.
13518 (profile_probability::differs_lot_from_p): New member function.
13519 (profile_probability::stream_in): New member function.
13520 (profile_probability::stream_out): New member function.
13521 * profile-count.h (profile_count_quality): Rename to ...
13522 (profile_quality): ... this one.
13523 (profile_probability): New.
13524 (profile_count): Update.
13525 * profile.c (compute_branch_probabilities): Update.
13526 * recog.c (peep2_attempt): Update.
13527 * sched-ebb.c (schedule_ebbs): Update.
13528 * sched-rgn.c (find_single_block_region): Update.
13529 (compute_dom_prob_ps): Update.
13530 (schedule_region): Update.
13531 * sel-sched-ir.c (compute_succs_info): Update.
13532 * stmt.c (struct case_node): Update.
13533 (do_jump_if_equal): Update.
13534 (get_outgoing_edge_probs): Update.
13535 (conditional_probability): Update.
13536 (emit_case_dispatch_table): Update.
13537 (expand_case): Update.
13538 (expand_sjlj_dispatch_table): Update.
13539 (emit_case_nodes): Update.
13540 * targhooks.c: Update.
13541 * tracer.c (better_p): Update.
13542 (find_best_successor): Update.
13543 * trans-mem.c (expand_transaction): Update.
13544 * tree-call-cdce.c: Update.
13545 * tree-cfg.c (gimple_split_edge): Upate.
13546 (move_sese_region_to_fn): Upate.
13547 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13548 * tree-eh.c (lower_resx): Upate.
13549 (cleanup_empty_eh_move_lp): Upate.
13550 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13551 * tree-inline.c (copy_edges_for_bb): Update.
13552 (copy_cfg_body): Update.
13553 * tree-parloops.c (gen_parallel_loop): Update.
13554 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13555 (gimple_gen_time_profiler): Update.
13556 * tree-ssa-dce.c (remove_dead_stmt): Update.
13557 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13558 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13559 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13560 (unloop_loops): Update.
13561 (try_peel_loop): Update.
13562 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13563 * tree-ssa-loop-split.c (connect_loops): Update.
13564 (split_loop): Update.
13565 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13566 (hoist_guard): Update.
13567 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13568 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13569 (value_replacement): Update.
13570 * tree-ssa-reassoc.c (branch_fixup): Update.
13571 * tree-ssa-tail-merge.c (replace_block_by): Update.
13572 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13573 (create_edge_and_update_destination_phis): Update.
13574 (compute_path_counts): Update.
13575 (recompute_probabilities): Update.
13576 (update_joiner_offpath_counts): Update.
13577 (freqs_to_counts_path): Update.
13578 (duplicate_thread_path): Update.
13579 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13580 (struct switch_conv_info): Update.
13581 (gen_inbound_check): Update.
13582 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13583 (vect_do_peeling): Update.
13584 (vect_loop_versioning): Update.
13585 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13586 (optimize_mask_stores): Update.
13587 * ubsan.c (ubsan_expand_null_ifn): Update.
13588 * value-prof.c (gimple_divmod_fixed_value): Update.
13589 (gimple_divmod_fixed_value_transform): Update.
13590 (gimple_mod_pow2): Update.
13591 (gimple_mod_pow2_value_transform): Update.
13592 (gimple_mod_subtract): Update.
13593 (gimple_mod_subtract_transform): Update.
13594 (gimple_ic): Update.
13595 (gimple_stringop_fixed_value): Update.
13596 (gimple_stringops_transform): Update.
13597 * value-prof.h: Update.
13599 2017-06-29 Carl Love <cel@us.ibm.com>
13601 * config/rs6000/rs6000-c.c: Add support for built-in functions
13602 vector signed int vec_signed (vector float);
13603 vector signed long long vec_signed (vector double);
13604 vector signed int vec_signed2 (vector double, vector double);
13605 vector signed int vec_signede (vector double);
13606 vector signed int vec_signedo (vector double);
13607 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13608 instruction generator.
13609 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13610 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13611 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13613 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13614 vunsignede_v2df): Add define_expands.
13615 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13616 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13617 VEC_UNSIGNEDO): Add definitions.
13618 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13619 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13620 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13621 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13622 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13623 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13624 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13625 vec_unsignede, vec_unsignedo): Add builtin defines.
13626 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13628 * doc/extend.texi: Update the built-in documentation file for the
13629 new built-in functions.
13631 2017-06-29 Richard Biener <rguenther@suse.de>
13633 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13634 reduction chains to LOOP_VINFO_REDUCTIONS.
13635 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13636 SLP reductions after processing reduction chains.
13638 2017-06-29 Nathan Sidwell <nathan@acm.org>
13640 * builtins.c (fold_builtin_FUNCTION): Use
13641 lang_hooks.decl_printable_name.
13643 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13645 PR middle-end/81194
13646 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13647 with only one label.
13648 * stmt.c (expand_case): Assert NCASES is greater than one.
13650 2017-06-29 Richard Biener <rguenther@suse.de>
13652 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13654 (group_case_labels): Likewise.
13655 (find_taken_edge): Push sanity checking on val to workers...
13656 (find_taken_edge_cond_expr): ... here
13657 (find_taken_edge_switch_expr): ... and here, handle cases
13658 with just a default label.
13659 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13660 (group_case_labels): Likewise.
13661 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13662 group_case_labels does anything cleanup the CFG again.
13664 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13666 PR tree-optimization/81196
13667 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13668 exit condition comparing two IVs.
13670 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13672 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13673 profile to the dummy entry at the end of the list of architectures.
13674 * config/arm/arm-cpu-cdata.h: Regenerated.
13676 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13677 Michael Collison <michael.collison@arm.com>
13680 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13682 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13683 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13684 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13685 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13686 of shift when the shift amount is masked with constant equal to
13687 the size of the mode.
13688 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13691 2017-06-29 Martin Liska <mliska@suse.cz>
13693 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13695 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13698 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13701 (edge_growth_cache_entry::edge_growth_cache_entry): New
13703 (reset_edge_growth_cache): Update to use constructor.
13705 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13707 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13708 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13709 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13711 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13713 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13714 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13716 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13718 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13719 (*-linux-uclibc*): Add t-uclibc tmake_file.
13720 * config/t-musl: New.
13721 * config/t-uclibc: New.
13723 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13725 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13727 (gen_comm_data): Emit architectural setting of arch_prof.
13728 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13730 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13731 (armv8-m.base, armv8-m.main): Likewise.
13732 * arm-protos.h (arm_build_target): Add profile field.
13733 (arch_option): Likewise.
13734 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13736 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13737 arm_active_target.profile.
13739 2017-06-28 Richard Biener <rguenther@suse.de>
13741 PR middle-end/81227
13742 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13743 TYPE_OVERFLOW_WRAPS.
13744 * match.pd (negate_expr_p): Likewise.
13745 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13746 fold_build2, not fold_binary.
13748 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13750 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13751 Convert memory address to Pmode.
13752 (aarch64_print_operand): Assert MEM operands are always Pmode.
13754 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13757 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13758 Remove redundant if.
13759 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13760 * config/arm/aarch-common-protos.h
13761 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13762 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13764 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13767 * multiple_target.c (create_dispatcher_calls): Set the default
13768 clone to be static, not public.
13770 2017-06-28 Richard Biener <rguenther@suse.de>
13772 * tree-vect-loop.c (vectorizable_reduction): Move special
13773 cond reduction IV var creation ...
13774 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13775 parameter. Use STMT_VINFO_VECTYPE.
13776 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13779 2017-06-28 Martin Liska <mliska@suse.cz>
13782 * ipa-visibility.c (non_local_p): Handle visibility.
13784 2017-06-28 Martin Liska <mliska@suse.cz>
13787 * common.opt: Add IntegerRange to various options.
13788 * opt-functions.awk (integer_range_info): New function.
13789 * optc-gen.awk: Add integer_range_info to cl_options struct.
13790 * opts-common.c (decode_cmdline_option): Handle
13791 CL_ERR_INT_RANGE_ARG.
13792 (cmdline_handle_error): Likewise.
13793 * opts.c (print_filtered_help): Show valid interval in
13794 when --help is provided.
13795 * opts.h (struct cl_option): Add range_min and range_max fields.
13796 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13798 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13800 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13801 (x * C EQ/NE y * C): New transformation.
13803 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13805 * genmultilib (combination_space): Accept '+' in option names.
13807 2017-06-28 Martin Liska <mliska@suse.cz>
13810 * asan.c (instrument_derefs): Bail out inner references
13811 that are hard register variables.
13813 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13816 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13817 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13819 2017-06-28 Richard Biener <rguenther@suse.de>
13821 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13822 (vect_get_slp_defs): Adjust.
13823 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13825 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13827 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13828 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13829 (vectorizable_reduction): Adjust.
13830 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13832 (vect_get_slp_defs): Likewise.
13833 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13834 (vectorizable_bswap): Adjust.
13835 (vectorizable_call): Likewise.
13836 (vectorizable_conversion): Likewise.
13837 (vectorizable_assignment): Likewise.
13838 (vectorizable_shift): Likewise.
13839 (vectorizable_operation): Likewise.
13840 (vectorizable_store): Likewise.
13841 (vectorizable_condition): Likewise.
13842 (vectorizable_comparison): Likewise.
13844 2017-06-28 Michael Collison <michael.collison@arm.com>
13847 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13849 (arm_gen_movmemqi): Removed unused variable 'i'.
13850 Convert 'for' loop into 'while' loop.
13851 (arm_expand_prologue): Remove last unnecessary set of insn.
13852 (thumb_pop): Remove unused variable 'pushed_words'.
13853 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13855 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13857 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13858 * config/s390/s390-protos.h: Add prototype of
13859 s390_rel_address_ok_p.
13860 * config/s390/s390.c (s390_got_symbol): New function.
13861 (s390_rel_address_ok_p): New function.
13862 (legitimize_pic_address): Use s390_rel_address_ok_p.
13863 (s390_load_got): Use s390_got_symbol.
13864 (s390_option_override): Issue error if
13865 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13866 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13868 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13870 2017-06-27 Andrew Pinski <apinski@cavium.com>
13872 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13873 (X * copysign (1.0, X)): New pattern.
13874 (X * copysign (1.0, -X)): New pattern.
13875 (copysign (-1.0, CST)): New pattern.
13877 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13879 * genmultilib (combination_space): Remove variable.
13880 Validate reuse rules against regular expression for any sequence
13881 of multilib options in any order.
13883 2017-06-27 Michael Collison <michael.collison@arm.com>
13885 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13886 call aarch64_split_simd_combine.
13887 * (aarch64_combine_internal<mode>): Delete pattern.
13888 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13889 Allow register and subreg operands.
13891 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13893 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13894 specific need, just fallback on defaults.
13895 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13897 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13898 Olivier Hainque <hainque@adacore.com>
13900 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13902 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13903 targets. Pick a default if no particular attempt applied.
13904 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13907 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13909 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13910 (x86_64-wrs-vxworks7): Likewise.
13912 2017-06-27 Marek Polacek <polacek@redhat.com>
13915 * ubsan.c (instrument_null): Check get_base_address's result for null.
13917 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13919 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13921 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13923 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13924 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13925 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13926 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13927 New function types.
13928 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13929 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13930 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13931 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13932 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13933 BUILT_IN_FEUPDATEENV): New builtins.
13934 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13935 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13936 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13937 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13939 (builtin_structptr_types): Adjust size.
13940 * tree.c (builtin_structptr_types): Add four entries.
13942 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13943 Olivier Hainque <hainque@adacore.com>
13945 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13946 (TLS_SYM): New local macro, forcing reference to __tls__ on
13947 link command lines for VxWorks 7 RTPs, triggering initialization
13949 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13950 OS features TLS support, true for RTPs on VxWorks 7.
13951 * config/vxworks.c (vxworks_override_options): Setup emutls
13954 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13956 * predict.c (test_prediction_value_range): Use -1U instead of -1
13957 to avoid narrowing conversion warning.
13958 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13959 to avoid narrowing conversion warning.
13960 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13962 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13964 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13966 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13967 64bit configurations.
13968 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13969 (SIZE_TYPE): Likewise.
13970 * config/vxworks.c (vxworks_emutls_var_fields): Use
13971 long_unsigned_type_node instead of unsigned_type_node as the offset
13972 field type, which is "pointer" mode in emutls.c.
13974 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13977 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13979 PR middle-end/81207
13980 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13981 gimple_vuse copying separately from gimple_vdef copying.
13983 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13985 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13988 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13989 Olivier Hainque <hainque@adacore.com>
13991 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13992 for all vxworks7 targets.
13993 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13994 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13995 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13996 variations for VX6/VX7 and 32/64bits later on in ...
13997 (VXWORKS_LIB_SPEC): Leverage new macros.
13998 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13999 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14001 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14003 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14004 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14006 2017-06-26 Carl Love <cel@us.ibm.com>
14008 * config/rs6000/rs6000-c.c: Add support for built-in functions
14009 vector bool char vec_reve (vector bool char);
14010 vector signed char vec_reve (vector signed char);
14011 vector unsigned char vec_reve (vector unsigned char);
14012 vector bool int vec_reve (vector bool int);
14013 vector signed int vec_reve (vector signed int);
14014 vector unsigned int vec_reve (vector unsigned int);
14015 vector bool long long vec_reve (vector bool long long);
14016 vector signed long long vec_reve (vector signed long long);
14017 vector unsigned long long vec_reve (vector unsigned long long);
14018 vector bool short vec_reve (vector bool short);
14019 vector signed short vec_reve (vector signed short);
14020 vector double vec_reve (vector double);
14021 vector float vec_reve (vector float);
14022 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14023 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14024 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14025 (altivec_vreve): New pattern.
14026 * config/rs6000/altivec.h (vec_reve): New define.
14027 * doc/extend.texi (vec_rev): Update the built-in documentation file
14028 for the new built-in functions.
14030 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14032 PR tree-optimization/71815
14033 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14035 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14037 (slsr_process_phi): Likewise.
14038 (replace_uncond_cands_and_profitable_phis): Don't replace a
14039 multiply candidate with a stride of 1 (copy or cast).
14040 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14042 (lowest_cost_path): Likewise.
14043 (total_savings): Likewise.
14045 2017-06-26 Richard Biener <rguenther@suse.de>
14048 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14049 Use def_builtin_pure for all gather builtins.
14051 2017-06-26 Richard Biener <rguenther@suse.de>
14053 PR tree-optimization/81203
14054 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14055 non-dominating BBs.
14057 2017-06-26 Marek Polacek <polacek@redhat.com>
14060 * doc/invoke.texi: Document -Wmultistatement-macros.
14062 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14064 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14065 arm_neon_ok_no_float_abi effective target.
14067 2017-06-26 Richard Biener <rguenther@suse.de>
14069 PR tree-optimization/80928
14070 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14071 (copy_bbs): Set BB_DUPLICATED flag early.
14072 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14074 (execute_on_shrinking_pred): Likewise.
14075 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14076 BB_DUPLICATED blocks.
14077 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14078 iterate over all PHIs considering removal of *gsi.
14080 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14082 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14085 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14087 * config/rs6000/rs6000-string.c: (expand_block_clear,
14088 do_load_for_compare, select_block_compare_mode,
14089 compute_current_alignment, expand_block_compare,
14090 expand_strncmp_align_check, expand_strn_compare,
14091 expand_block_move, rs6000_output_load_multiple)
14092 Move functions related to string/block move/compare
14093 to a separate file.
14094 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14095 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14096 for this function which is now used in two files.
14097 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14098 * config.gcc: Add rs6000-string.o to extra_objs for
14099 targets powerpc*-*-* and rs6000*-*-*.
14101 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14104 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14105 32-bit, since indexed is not valid for DImode.
14106 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14107 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14108 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14109 (define_peephole2 for Altivec d-form store): Likewise.
14112 * multiple_target.c (create_dispatcher_calls): Only create the
14113 dispatcher call if the function is the default clone of a
14114 versioned function.
14116 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14118 PR middle-end/80902
14119 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14120 a call, force the call to not be a tail call.
14122 2017-06-23 Jeff Law <law@redhat.com>
14124 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14127 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14129 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14130 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14131 handling for early expansion of vector shifts (sl,sr,sra,rl).
14132 (builtin_function_type): Add vector shift right instructions
14133 to the unsigned argument list.
14135 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14137 rtl-optimizatoin/79286
14138 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14139 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14140 trap. PIC register plus a const unspec without offset can never trap.
14142 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14144 * tree.h (builtin_structptr_type): New type.
14145 (builtin_structptr_types): Declare new array.
14146 * tree.c (builtin_structptr_types): New array.
14147 (free_lang_data, build_common_tree_nodes): Use it.
14149 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14152 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14155 2017-06-22 Matt Turner <mattst88@gmail.com>
14157 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14158 Lake models to skylake case. Assume skylake for unknown
14159 models with clflushopt.
14161 2017-06-22 Jeff Law <law@redhat.com>
14163 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14164 frame sizes that do not satisfy aarch64_uimm12_shift.
14166 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14168 * profile-count.h (apply_probability,
14169 apply_scale, probability_in): Fix checks for zero.
14171 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14173 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14174 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14176 2016-06-22 Richard Biener <rguenther@suse.de>
14178 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14179 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14180 REDUC_MAX_EXPR support.
14181 (vectorizable_reduction): Likewise.
14182 (vect_create_epilog_for_reduction): Likewise.
14184 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14186 * match.pd (A / (1 << B) -> A >> B): New.
14187 * generic-match-head.c: Include optabs-tree.h.
14188 * gimple-match-head.c: Likewise.
14189 * optabs-tree.h (target_supports_op_p): New.
14190 * optabs-tree.c (target_supports_op_p): New.
14192 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14194 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14195 $gcc_cv_ld --help output.
14196 (gcc_cv_ld_demangle): Likewise.
14197 (gcc_cv_ld_eh_frame_hdr): Likewise.
14198 (gcc_cv_ld_pie): Likewise.
14199 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14200 (gcc_cv_ld_buildid): Likewise.
14201 (gcc_cv_ld_sysroot): Likewise.
14202 (ld_bndplt_support): Likewise.
14203 (ld_pushpopstate_support): Likewise.
14204 * configure: Regenerate.
14205 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14207 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14210 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14211 operands indexes to avoid gap between operands and match_dups.
14213 2017-06-21 Andrew Pinski <apinski@cavium.com>
14215 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14216 Increment Arith_shift and Arith_shift_reg by 1.
14217 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14219 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14220 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14221 (aarch64_strip_extend): Add new argument and test for it.
14222 (aarch64_cheap_mult_shift_p): New function.
14223 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14224 add a cost if it is true.
14225 Update calls to aarch64_strip_extend.
14226 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14228 2017-06-21 Andrew Pinski <apinski@cavium.com>
14230 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14232 (thunderxt88): Likewise.
14233 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14234 (thunderx_prefetch_tune): New variable.
14235 (thunderx2t99_prefetch_tune): Update for the correct values.
14236 (thunderxt88_tunings): New variable.
14237 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14238 generic_prefetch_tune.
14239 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14241 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14243 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14244 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14245 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14246 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14247 (aarch64_atomic_cas<mode>, GPI): Likewise.
14249 2017-06-21 Martin Liska <mliska@suse.cz>
14251 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14252 statements on cold and hot labels.
14253 * predict.c (tree_estimate_probability_bb): Remove the
14254 prediction from this place.
14256 2017-06-21 Martin Liska <mliska@suse.cz>
14258 PR tree-optimization/79489
14259 * gimplify.c (maybe_add_early_return_predict_stmt): New
14261 (gimplify_return_expr): Call the function.
14262 * predict.c (tree_estimate_probability_bb): Remove handling
14264 * predict.def: Update comment about early return predictor.
14265 * gimple-predict.h (is_gimple_predict): New function.
14266 * predict.def: Change default value of early return to 66.
14267 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14269 * passes.def: Put pass_strip_predict_hints to the beginning of
14272 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14274 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14275 FUNCTION_DECL declarations.
14276 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14278 (dwaf2out_decl): Likewise.
14279 * godump.c (go_early_global_decl): Skip call to the real debug hook
14280 for FUNCTION_DECL declarations.
14281 * passes.c (rest_of_decl_compilation): Skip call to the
14282 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14283 -fdump-go-spec is passed.
14285 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14287 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14288 Reorder for compactness.
14289 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14290 (def_builtin_pure, def_builtin_pure2): New functions.
14291 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14293 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14295 * match.pd (nop_convert): New predicate.
14296 ((A +- CST1) +- CST2): Allow some NOP conversions.
14298 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14301 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14302 with ctors/dtors if GOVD_SHARED is set.
14304 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14306 * config/aarch64/aarch64.md (movti_aarch64):
14307 Emit mov rather than orr.
14308 (movtf_aarch64): Likewise.
14309 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14310 Emit mov rather than orr.
14312 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14314 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14315 Swap alternatives, make integer dup more expensive.
14317 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14319 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14320 Return true for non-tls symbols.
14322 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14324 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14325 (cortex-a75): Likewise.
14326 (cortex-a75.cortex-a55): Likewise.
14327 * config/aarch64/aarch64-tune.md: Regenerate.
14328 * doc/invoke.texi (-mtune): Document new values for -mtune.
14330 2017-06-21 Tom de Vries <tom@codesourcery.com>
14332 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14333 stack_size feature.
14334 (Effective-Target Keywords, Other attributes): Suggest using
14335 dg-add-options stack_size feature to get stack limit in stack_size
14336 effective target documentation.
14338 2017-06-21 Julian Brown <julian@codesourcery.com>
14339 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14341 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14342 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14343 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14345 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14346 attribute type list for neon_multiply.
14347 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14348 attribute type list for neon_multiply.
14349 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14350 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14351 attribute type list for neon_multiply.
14352 * config/arm/types.md (crypto_pmull): Add.
14353 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14354 attribute type list.
14356 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14358 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14361 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14363 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14364 to make sure not to dereference a NULL cost_classes_ptr pointer.
14366 2017-06-20 Carl Love <cel@us.ibm.com>
14368 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14369 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14370 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14371 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14372 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14373 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14374 VMULOSW): New enum "unspec" values.
14375 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14376 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14377 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14378 altivec_vmulosw): New patterns.
14379 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14380 VMULOSW): Add definitions.
14382 2017-06-20 Julia Koval <julia.koval@intel.com>
14384 * config/i386/i386.c: Fix rounding expand for new pattern.
14385 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14387 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14389 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14390 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14392 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14394 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14397 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14399 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14400 architecture, then by implementer ID.
14401 * config/aarch64/aarch64-tune.md: Regenerate.
14403 2017-06-20 Richard Biener <rguenther@suse.de>
14405 PR middle-end/81097
14406 * fold-const.c (split_tree): Fold to type before negating.
14408 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14410 * diagnostic-show-locus.c
14411 (selftest::test_fixit_deletion_affecting_newline): New function.
14412 (selftest::diagnostic_show_locus_c_tests): Call it.
14414 2017-06-20 Andreas Schwab <schwab@suse.de>
14417 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14420 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14422 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14423 __ARM_FEATURE_COPROC according to support.
14425 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14427 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14428 Rewritten to avoid overflow for > 32-bit pointers.
14431 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14432 by removing enum keyword.
14433 (ubsan_type_descriptor): Likewise. Formatting fix.
14436 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14437 splitter): Require TARGET_SSE2 in the condition.
14439 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14442 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14443 for doing vector set of SFmode on ISA 3.0.
14444 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14445 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14447 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14448 SFmode value into a V4SF variable that was extracted from another
14449 V4SF variable without converting the element to double precision
14450 and back to single precision vector format.
14451 (vsx_insert_extract_v4sf_p9_2): Likewise.
14453 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14455 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14456 in UWHI to avoid undefined overflow.
14459 * ubsan.h (enum ubsan_encode_value_phase): New.
14460 (ubsan_encode_value): Change second argument to
14461 enum ubsan_encode_value_phase with default value of
14462 UBSAN_ENCODE_VALUE_GENERIC.
14463 * ubsan.c (ubsan_encode_value): Change second argument to
14464 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14465 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14466 create_tmp_var_raw instead of create_tmp_var and use a
14468 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14469 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14470 ubsan_encode_value callers.
14473 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14474 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14475 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14477 2017-06-19 Richard Biener <rguenther@suse.de>
14479 PR middle-end/81118
14480 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14481 estimates if we changed anything.
14483 2017-06-19 Richard Biener <rguenther@suse.de>
14485 PR tree-optimization/80887
14486 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14487 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14488 simplified lookups, then reset mprts_hook.
14489 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14491 (try_to_simplify): Likewise.
14493 2017-06-19 Martin Liska <mliska@suse.cz>
14496 * gimplify.c (gimplify_switch_expr):
14497 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14499 2017-06-19 Martin Liska <mliska@suse.cz>
14501 * doc/install.texi: Document that PGO runs in 4 stages.
14503 2017-06-19 Martin Liska <mliska@suse.cz>
14506 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14507 to dispatcher function name.
14508 * multiple_target.c (replace_function_decl): New function.
14509 (create_dispatcher_calls): Redirect both edges and references.
14511 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14513 * profile-count.c (profile_count::dump): Dump quality.
14514 (profile_count::differs_from_p): Update for unsigned val.
14515 * profile-count.h (profile_count_quality): New enum.
14516 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14518 2017-06-19 Richard Biener <rguenther@suse.de>
14520 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14521 struct function as arg.
14522 (estimate_numbers_of_iterations): Export overload with loop arg.
14523 (free_numbers_of_iterations_estimates_loop): Use an overload of
14524 free_numbers_of_iterations_estimates instead.
14525 * tree-cfg.c (remove_bb): Adjust.
14526 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14527 * tree-parloops.c (gen_parallel_loop): Likewise.
14528 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14530 (tree_unroll_loops_completely): Likewise.
14531 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14532 Use an overload instead and export.
14533 (estimated_loop_iterations): Adjust.
14534 (max_loop_iterations): Likewise.
14535 (likely_max_loop_iterations): Likewise.
14536 (estimate_numbers_of_iterations): Take struct function as arg
14538 (loop_exits_before_overflow): Adjust.
14539 (free_numbers_of_iterations_estimates_loop): Use an overload.
14540 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14541 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14543 2017-06-19 Richard Biener <rguenther@suse.de>
14546 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14547 RANGE_EXPR conservatively.
14549 2017-06-16 Carl Love <cel@us.ibm.com>
14551 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14552 definitions for vec_float, vec_float2, vec_floato,
14553 vec_floate built-ins.
14554 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14555 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14557 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14558 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14559 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14560 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14561 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14562 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14563 vec_floato): Add builtin defines.
14564 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14565 Update the built-in documentation file for the new built-in
14568 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14570 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14571 (mthumb): Mark as the negative of -marm.
14573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14575 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14577 (ARM Options, -mtune): Document that this accepts the same
14578 extension options as -mcpu.
14579 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14583 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14584 permitted extensions.
14586 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14588 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14589 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14590 (armv8-m.main): Add option +nodsp.
14591 * config/arm/arm-cpu-cdata.h: Regenerated.
14593 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14595 * config/arm/t-fuchsia: New file.
14596 * config.gcc (arm*-*-fuchsia*): Use it.
14598 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14600 * config/arm/t-symbian: Rewrite for new option infrastructure.
14602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14604 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14605 (MULTILIB_REQUIRED): Likewise.
14607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14609 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14610 (MULTILIB_RESUE): Likewise.
14611 (MULTILIB_MATCHES): Likewise.
14612 (MULTLIB_REQUIRED): Likewise.
14614 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14616 * config/arm/t-rtems: Rewrite for new option framework.
14618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14620 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14621 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14622 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14623 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14624 * config/arm/t-multilib: ... here.
14625 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14626 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14627 armv7-a and armv8*-a when A-profile libraries have not been built.
14628 * config/arm/t-rmprofile: Rewrite.
14630 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14632 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14633 with a backslash. Remove the backslash after substituting unescaped
14635 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14637 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14639 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14640 the driver to pass through the default setting of -mfloat-abi.
14641 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14643 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14644 (all_feat_combs): New rule.
14645 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14647 * config/arm/t-aprofile: Rewrite.
14649 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14651 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14652 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14653 fpu is not specified by the user/command-line.
14654 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14655 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14656 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14657 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14658 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14659 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14663 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14664 * config/arm/t-arm-elf: Rewritten.
14666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14668 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14669 have some floating-point instructions.
14670 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14671 (TARGET_MAYBE_HARD_FLOAT): New macro.
14672 * config/arm/arm-builtins.c (arm_init_builtins): Use
14673 TARGET_MAYBE_HARD_FLOAT.
14674 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14678 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14679 (configargs.h): Include it.
14680 (arm_print_hint_for_fpu_option): New function.
14681 (arm_parse_fpu_option): New function.
14682 (candidate_extension): New class.
14683 (arm_canon_for_multilib): New function.
14684 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14685 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14686 (ARCH_CANONICAL_SPECS): New macro.
14687 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14689 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14691 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14692 are set after handling multilib fragments. Set target_cpu_default2
14695 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14697 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14699 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14700 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14702 (check_arch): Likewise.
14703 * config/arm/arm.c (arm_configure_build_target): Handle
14704 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14705 options in the default.
14707 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14709 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14710 when an option is an alias of another.
14711 * config/arm/parsecpu.awk (optalias): New parser token.
14712 (gen_comm_data): Mark non-alias options as such. Emit entries
14713 for extension aliases.
14714 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14715 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14716 (armv6kz, armv6zk, armv6t2): Likewise.
14717 (armv7): Make vfpv3-d16 an alias.
14718 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14720 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14721 Sort in canonical order.
14722 (armv8-a): Sort in canonical order.
14723 (armv8.1-a, armv8.2-a): Likewise.
14724 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14726 (cortex-a9): Sort in canonical order.
14727 * config/arm/arm.c (selftests.h): Include it.
14728 (arm_test_cpu_arch_data): New function.
14729 (arm_run_self_tests): New function.
14730 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14731 (targetm): Move declaration to the end of the file.
14732 * arm-cpu-cdata.h: Regenerated.
14734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14736 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14737 call to target_mode_check describing the type of option passed.
14738 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14739 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14740 arm_parse_cpu_option_name to match parameters against list of
14742 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14743 arm_arch_core_flags data structure.
14744 * config/arm/arm-cpu_cdata.h: Regenerated.
14746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14748 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14750 (arm_print_hint_for_cpu_option): Likewise.
14751 (arm_print_hint_for_arch_option): Likewise.
14752 (arm_parse_cpu_option_name): Likewise.
14753 (arm_parse_arch_option_name): Likewise.
14754 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14755 of entries in the all_fpus list.
14756 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14757 (arm_parse_cpu_option_name): Declare.
14758 (arm_parse_arch_option_name): Declare.
14759 (arm_parse_option_features): Declare.
14760 (arm_intialize_isa): Declare.
14761 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14763 (gen_comm_data): ... here. Make definitions non-static.
14764 * config/arm/arm-cpu-data.h: Regenerated.
14765 * config/arm/arm-cpu-cdata.h: Regenerated.
14767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14769 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14770 (cpu_arch_extension): New structure.
14771 (cpu_arch_option, arch_option, cpu_option): New structures.
14772 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14773 architecture types.
14774 (gen_data): Generate new format data tables.
14775 * config/arm/arm.c (cpu_tune): New structure.
14776 (cpu_option, processors): Delete.
14777 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14778 (arm_print_hint_for_cpu_option): ... this and ...
14779 (arm_print_hint_for_arch_option): ... this.
14780 (arm_parse_arch_cpu_name): Delete. Replace with ...
14781 (arm_parse_cpu_option_name): ... this and ...
14782 (arm_parse_arch_option_name): ... this.
14783 (arm_unrecognized_feature): Change type of target parameter to
14785 (arm_parse_arch_cpu_features): Delete. Replace with ...
14786 (arm_parse_option_features): ... this.
14787 (arm_configure_build_target): Rework to use new configuration data
14789 (arm_print_tune_info): Rework for new configuration data tables.
14790 * config/arm/arm-cpu-data.h: Regenerated.
14791 * config/arm/arm-cpu.h: Regenerated.
14793 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14795 * Makefile.in (OBJS): Move sbitmap.o from here ...
14796 (OBJS-libcommon): ... to here.
14798 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14800 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14801 (ISA_ALL_CRYPTO): New macro.
14802 (ISA_ALL_SIMD): New macro
14803 (ISA_ALL_FP): New macro.
14804 * config/arm/arm.c (fpu_bitlist): Update initializer.
14805 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14807 (arm9e): Add fpu. Add option for nofp
14808 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14809 (arm926ej-s, arm1026ej-s): Likewise.
14810 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14811 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14812 neon-fp16, neon-vfpv4, nofp and nosimd.
14813 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14814 (cortex-a8): Add fpu. Add option for nofp.
14815 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14816 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14817 (cortex-r4f): Add fpu.
14818 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14819 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14821 (cortex-r8): Likewise.
14822 (cortex-m4): Add fpu. Add option for nofp.
14823 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14824 (cortex-a17.cortex-a7): Likewise.
14825 (cortex-a32): Add fpu. Add options for crypto and nofp.
14826 (cortex-a35, cortex-a53): Likewise.
14827 (cortex-a57): Add fpu. Add option for crypto.
14828 (cortex-a72, cortex-a73): Likewise.
14829 (exynos-m1): Likewise.
14830 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14831 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14832 (cortex-m33): Add fpu. Add option for nofp.
14833 * config/arm/arm-cpu-cdata.h: Regenerated
14834 * config/arm/arm-cpu-data.h: Regenerated.
14836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14838 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14839 (armv5te, armv5tej): Likewise.
14840 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14841 (armv7): Add options fp and vfpv3-d16.
14842 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14843 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14845 (armv7ve): Likewise.
14846 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14847 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14848 (armv8-a): Add nocrypto option.
14849 (armv8.1-a, armv8.2-a): Likewise.
14850 (armv8-m.main): add options fp, fp.dp and nofp.
14852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14854 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14856 (armv8-a+crc): Delete.
14857 (armv8.1-a): Add options simd, crypto and nofp.
14858 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14859 (armv8.2-a+fp16): Delete.
14860 (armv8-m.main): Add option dsp.
14861 (armv8-m.main+dsp): Delete.
14862 (cortex-a8): Add fpu. Add nofp option.
14863 (cortex-a9): Add fpu. Add nofp and nosimd options.
14864 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14865 link to main cpu and architecture data structures.
14866 (gen_comm_data): Only put isa attributes from the main architecture
14868 (option): New statement for architecture and CPU entries.
14869 * arm.c (struct cpu_option): New structure.
14870 (struct processors): Add entry for options.
14871 (arm_unrecognized_feature): New function.
14872 (arm_parse_arch_cpu_name): Ignore any characters after the first
14874 (arm_parse_arch_cpu_feature): New function.
14875 (arm_configure_build_target): Separate out any CPU and architecture
14876 features and parse separately. Don't error out if -mfpu=auto is
14877 used with only an architecture string.
14878 (arm_print_asm_arch_directives): New function.
14879 (arm_file_start): Call it.
14880 * config/arm/arm-cpu-cdata.h: Regenerated.
14881 * config/arm/arm-cpu-data.h: Likewise.
14882 * config/arm/arm-tables.opt: Likewise.
14884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14886 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14887 assembler when it is not -mfpu=auto.
14889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14891 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14892 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14893 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14894 (ASM_CPU_SPEC): Rewrite.
14895 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14896 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14897 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14898 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14899 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14900 copied string is NUL-terminated. Also strip any characters prefixed
14902 (arm_rewrite_selected_arch): New function.
14903 (arm_rewrite_march): New function.
14905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14907 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14908 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14909 (march, mcpu, mtune): Convert to string-based options.
14910 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14911 (arm_parse_arch_cpu_name): New function.
14912 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14913 identify selected architecture or CPU.
14914 (arm_option_save): New function.
14915 (TARGET_OPTION_SAVE): Redefine.
14916 (arm_option_restore): Restore string options.
14917 (arm_option_print): Print string options.
14919 2017-06-16 Martin Sebor <msebor@redhat.com>
14921 PR tree-optimization/80933
14922 PR tree-optimization/80934
14923 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14924 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14925 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14926 (gimple_fold_builtin): Call them.
14928 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14930 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14931 as unlikely; update profile.
14933 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14935 * predict.c (force_edge_cold): Handle declaring edges impossible
14938 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14940 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14942 (try_unroll_loop_completely): Fix reporting.
14944 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14946 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14948 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14951 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14952 if given a non-constant argument for an intrinsic which requires a
14955 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14957 * profile.c (compare_freqs): New function.
14958 (branch_prob): Sort edge list.
14959 (find_spanning_tree): Assume that the list is priority sorted.
14961 2017-06-16 Richard Biener <rguenther@suse.de>
14963 PR tree-optimization/81090
14964 * passes.def (pass_record_bounds): Remove.
14965 * tree-pass.h (make_pass_record_bounds): Likewise.
14966 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14967 make_pass_record_bounds): Likewise.
14968 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14969 not free niter estimates at the beginning but at the end.
14970 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14972 2017-06-16 Richard Biener <rguenther@suse.de>
14974 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14975 initializer to workaround ICE in host GCC 4.8.
14977 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14979 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14981 (clone_inlined_nodes): Update.
14983 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14985 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14986 prefetch settings, and enable prefetching by default at -O3.
14988 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14990 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14991 Set flag_prefetch_loop_arrays according to tuning data.
14993 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14995 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14996 New tune structure.
14997 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14998 [Unrelated to main purpose of the patch] Place the pointer field last
14999 to enable type checking errors when tune structure are wrongly merged.
15000 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15001 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15002 (thunderx2t99_prefetch_tune): New tune constants.
15003 (tune_params *_tunings): Update all tunings (no functional change).
15004 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15005 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15006 from tunings structures.
15008 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15011 * ubsan.c (instrument_null): Add T argument, use it instead
15012 of computing it based on IS_LHS.
15013 (instrument_object_size): Likewise.
15014 (pass_ubsan::execute): Adjust instrument_null and
15015 instrument_object_size callers to pass gimple_get_lhs or
15016 gimple_assign_rhs1 result to it. Use instrument_null instead of
15017 calling get_base_address and instrument_mem_ref. Handle
15018 aggregate call arguments for object-size sanitization.
15020 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15022 PR tree-optimization/81089
15023 * tree-vrp.c (is_masked_range_test): Validate operands of
15026 2017-06-15 Martin Sebor <msebor@redhat.com>
15029 * dumpfile.c (dump_register): Avoid calling memset to initialize
15030 a class with a default ctor.
15031 * gcc.c (struct compiler): Remove const qualification.
15032 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15033 * hash-table.h: Ditto.
15034 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15036 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15037 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15039 * params.h (struct param_info): Make struct members non-const.
15040 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15041 with default initialization.
15042 * vec.h (vec_copy_construct, vec_default_construct): New helper
15044 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15045 with vec_copy_construct.
15046 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15047 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15048 * doc/invoke.texi (-Wclass-memaccess): Document.
15050 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15052 * emit-rtl.h (is_leaf): Update comment about local
15053 register allocator.
15055 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15058 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15059 for a variable to have a section before checking if the section has a
15061 Set section to.persistent if persistent attribute is set.
15062 Warn if .persistent attribute is used on an automatic variable.
15064 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15066 PR rtl-optimization/80474
15067 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15069 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15071 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15074 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15076 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15077 (casesi): Emit bounds checking as RTL.
15078 (casesi_internal_mips16_<mode>): Remove bounds checking.
15080 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15082 * config/xtensa/xtensa.c (xtensa_option_override): Append
15083 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15084 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15085 xtensa_doloop_hooks): Define unconditionally.
15086 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15088 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15089 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15090 for it in xtensa_option_override.
15091 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15092 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15094 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15096 * doc/cppopts.texi: Document '-' special value to -MF.
15098 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15100 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15101 (cortex_a53_fconst): Likewise.
15102 (cortex_a53_fpmul): Likewise.
15103 (cortex_a53_f_load_64): Likewise.
15104 (cortex_a53_f_load_many): Likewise.
15105 (cortex_a53_advsimd_alu): Likewise.
15106 (cortex_a53_advsimd_alu_q): Likewise.
15107 (cortex_a53_advsimd_mul): Likewise.
15108 (cortex_a53_advsimd_mul_q): Likewise.
15109 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15110 Add missing fmul, r2f_cvt and fconst cases.
15112 2017-06-14 Richard Biener <rguenther@suse.de>
15114 PR middle-end/81088
15115 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15117 (fold_binary_loc): When associating do not treat pre-existing
15118 TREE_OVERFLOW on literal constants as a reason to allow
15119 TREE_OVERFLOW on associated literal constants.
15121 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15123 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15124 (MASK_FEATURES): New macro.
15125 * config/sparc/sparc.c (sparc_option_override): Remove the special
15126 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15128 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15130 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15131 a division of 0 if non-call exceptions are enabled.
15133 2017-06-14 Andrew Pinski <apinski@cavium.com>
15134 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15137 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15138 Improve vector initialization code gen for only variable case.
15140 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15142 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15144 2017-06-14 Richard Biener <rguenther@suse.de>
15146 PR tree-optimization/81083
15147 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15150 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15152 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15153 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15154 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15155 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15156 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15157 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15159 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15161 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15162 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15164 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15166 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15168 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15170 * config/rs6000/t-rtems: Don't handle SPE.
15172 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15174 * config/rs6000/t-linux: Don't handle SPE.
15176 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15178 * config/rs6000/eabispe.h: Delete file.
15180 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15182 * config/rs6000/t-spe: Delete file.
15184 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15186 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15187 (rs6000_legitimate_offset_address_p): Return false for anything in
15188 V2SImode or V2SFmode.
15190 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15192 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15193 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15194 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15196 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15197 (rs6000_legitimate_offset_address_p): Ditto.
15198 (rs6000_emit_move): Ditto.
15199 (rs6000_init_builtins): Remove V4HI_type_node.
15201 2017-06-13 Martin Liska <mliska@suse.cz>
15204 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15205 (gate_asan): Likewise.
15206 * asan.h (asan_no_sanitize_address_p): Remove the function.
15207 (sanitize_flags_p): New function.
15208 * builtins.def: Fix coding style.
15209 * common.opt: Use renamed enum value.
15210 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15211 * doc/extend.texi: Document no_sanitize attribute.
15212 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15213 to SANITIZE_UNDEFINED_NONDEFAULT.
15214 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15215 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15216 Use sanitize_flags_p.
15217 * gimplify.c (gimplify_function_tree): Likewise.
15218 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15219 * opts.c (parse_no_sanitize_attribute): New function.
15220 (common_handle_option): Use renamed enum value.
15221 * opts.h (parse_no_sanitize_attribute): Declare.
15222 * tree.c (sanitize_flags_p): New function.
15223 * tree.h: Declared here.
15224 * tsan.c: Use sanitize_flags_p.
15225 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15226 (instrument_mem_ref): Likewise.
15227 (instrument_bool_enum_load): Likewise.
15228 (do_ubsan_in_current_function): Remove the function.
15229 (pass_ubsan::execute): Use sanitize_flags_p.
15230 * ubsan.h: Remove do_ubsan_in_current_function
15231 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15232 (dump_function_to_file): Use it here.
15234 2017-06-13 Martin Jambor <mjambor@suse.cz>
15236 PR tree-optimization/80803
15237 PR tree-optimization/81063
15238 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15239 (propagate_subaccesses_across_link): Enqueue subtree whenever
15240 necessary instead of relying on the caller.
15242 2017-06-13 Martin Jambor <mjambor@suse.cz>
15244 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15245 that have a first_link.
15246 (sort_and_splice_var_accesses): Do not check first_link before
15248 (subtree_mark_written_and_enqueue): Likewise.
15249 (propagate_all_subaccesses): Likewise and do not stop at first
15250 parent with a first_link.
15252 2017-06-13 Martin Jambor <mjambor@suse.cz>
15254 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15257 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15259 * match.pd: New pattern.
15261 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15263 * tree-vrp.c (is_masked_range_test): New function.
15264 (register_edge_assert_for): Determine ranges for
15267 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15269 PR tree-optimization/67328
15270 * fold-const.c (maskable_range_p): New function.
15271 (build_range_check): Generate bittests if possible.
15273 2017-06-13 Martin Liska <mliska@suse.cz>
15275 * gimple-pretty-print.c (dump_probability): Add new argument.
15276 (dump_edge_probability): Dump both probability and count.
15277 (dump_gimple_label): Likewise.
15278 (dump_gimple_bb_header): Likewise.
15280 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15283 * config/avr/avr-devices.c: Fix indentation.
15284 * config/avr/gen-avr-mmcu-specs.c: Dito.
15286 2017-06-13 Richard Biener <rguenther@suse.de>
15288 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15289 instead get vector type from stmt_info.
15290 (vectorizable_reduction): Adjust. Remove dead code.
15292 2017-06-13 Richard Biener <rguenther@suse.de>
15294 PR middle-end/81065
15295 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15296 case of C * (x * C2 + C3).
15297 (fold_addr_of_array_ref_difference): Properly fold index difference.
15299 2017-06-12 David S. Miller <davem@davemloft.net>
15302 * config/sparc/sparc.md (return expander): Emit frame blockage if
15303 function uses alloca.
15305 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15307 * combine.c (make_field_assignment): Check len rather than the mode
15308 precision when calling force_to_mode.
15310 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15312 Support multilibs and devices that see flash in RAM address range.
15315 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15316 (avr_mcu_t) <flash_pm_offset>: New field.
15317 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15318 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15319 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15320 (AVR_TINY_PM_OFFSET): Remove macro.
15321 * config/avr/avr.opt (-mshort-calls): New option.
15322 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15323 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15324 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15325 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15326 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15327 instead of avr_arch->have_jmp_call.
15328 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15329 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15330 avr_arch->flash_pm_offset to define.
15331 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15332 new field flash_pm_offset. Add entry for avrxmega3.
15333 (avr_texinfo): Add entry for avrxmega3.
15334 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15335 attiny212, attiny214,
15336 attiny412, attiny414, attiny416, attiny417,
15337 attiny814, attiny816, attiny817,
15338 attiny1614, attiny1616, attiny1617,
15339 attiny3214, attiny3216, attiny3217.
15340 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15341 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15342 (avr_print_operand_address) [AVR_TINY]: Same.
15343 (avr_asm_init_sections) <readonly_data_section>: Only patch
15344 callback if avr_arch->flash_pm_offset = 0.
15345 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15346 for rodata if avr_arch->flash_pm_offset != 0.
15347 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15348 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15349 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15350 (m_options): Append opt_rcall.
15351 (m_dirnames): Append dir_rcall.
15352 * config/avr/t-multilib: Regenerate.
15354 * configure.ac [target=avr]: Check whether avrxmega3 default
15355 linker description file works as needed.
15356 * configure: Regenerate.
15357 * doc/avr-mmcu.texi: Regenerate.
15358 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15359 <__AVR_ARCH__>: Document avrxmega3 and 103.
15360 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15361 <__AVR_SHORT_CALLS__>: Document it.
15362 <__AVR_PM_BASE_ADDRESS__>: Document it.
15363 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15364 (AVR Variable Attributes) <progmem>: Document this is
15365 not needed for avrxmega3.
15366 (AVR Named Address Spaces) <__flash>: Dito.
15368 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15370 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15372 2017-06-12 Doug Rupp <rupp@adacore.com>
15374 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15375 Append vxworks-stdint.h to the tm_file list.
15376 * config/vxworks-stdint.h: New file.
15378 2017-06-12 Martin Liska <mliska@suse.cz>
15380 PR tree-optimization/81041
15381 * tree-profile.c (gimple_gen_ic_func_profiler):
15382 Create an extra BB in profile-generate
15383 (gimple_gen_time_profiler): Likewise.
15385 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15387 PR tree-optimization/81003
15388 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15389 (update_range_test): Use it instead of force_gimple_operand_gsi.
15391 2017-06-12 Richard Biener <rguenther@suse.de>
15393 PR tree-optimization/81053
15394 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15395 with backedge value not defined in loop. Simplify def stmt
15398 2017-06-11 Tom de Vries <tom@codesourcery.com>
15401 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15403 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15404 nvptx_cannot_force_const_mem.
15406 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15408 * opts.c (finish_options): Move test for flag_split_stack after
15409 it has been initialized.
15411 2017-06-11 Jason Merrill <jason@redhat.com>
15413 * tree.h (id_equal): New.
15414 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15415 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15416 instead of strcmp of IDENTIFIER_POINTER.
15418 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15420 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15421 (mark_all_inlined_calls_cdtor): Fix formating.
15422 (inline_transform): Rescale profile before inlining.
15424 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15426 * cgraph.h (cgraph_edge::clone): Update prototype.
15427 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15428 (cgraph_node::create_clone): Update.
15429 (cgraph_node::create_version_clone): Update.
15430 * tree-inline.c (copy_bb): Update.
15431 (expand_call_inline): Update.
15433 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15435 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15436 factored out from ...
15437 (rs6000_emit_prologue): ... here.
15439 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15441 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15442 factored out from ...
15443 (rs6000_emit_prologue): ... here.
15445 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15447 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15449 (handle_missing_profiles): Fix computation of tp_first_run.
15450 (counts_to_freqs): Do not touch freqs when count is 0.
15452 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15454 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15457 2017-06-10 Tom de Vries <tom@codesourcery.com>
15459 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15460 attributes): Document signal effective target.
15462 2017-06-10 Tom de Vries <tom@codesourcery.com>
15464 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15465 Document effective target stack_size.
15467 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15469 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15470 to the edit_context if they can be auto-applied.
15472 2017-06-9 Ian Lance Taylor <iant@golang.org>
15474 * opts.c (finish_options): If -fsplit-stack, disable implicit
15475 -forder-blocks-and-partition.
15476 * doc/invoke.texi (Optimize Options): Document that when using
15477 -fsplit-stack -forder-blocks-and-partition is not implicitly
15480 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15482 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15483 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15484 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15485 * builtins.def (abort, trap, unreachable): Declare cold.
15486 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15487 * tree-core.h (ECF_COLD): New.
15488 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15489 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15491 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15493 * predict.c (unlikely_executed_stmt_p): Cleanup.
15495 2017-06-09 Richard Biener <rguenther@suse.de>
15497 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15498 model if the ref is always written to.
15500 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15502 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15504 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15506 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15509 2017-06-09 Tom de Vries <tom@codesourcery.com>
15512 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15513 "target cannot support label values" when encountering LABEL_REF.
15515 2017-06-09 Martin Liska <mliska@suse.cz>
15517 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15518 (gimple_gen_ic_func_profiler): Emit direct comparison
15519 of __gcov_indirect_call_callee with NULL.
15520 (gimple_gen_time_profiler): Change probability from
15521 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15523 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15525 * profile.c (edge_gcov_counts): Turn to pointer.
15526 (compute_branch_probabilities, compute_branch_probabilities): Update.
15527 (branch_prob): Do not clear edge_gcov_count.
15528 * profile.h (edge_gcov_counts): Turn to pointer.
15529 (edge_gcov_count): Update.
15531 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15533 * gimple.h (gimple_check_failed): Mark cold.
15535 2017-06-09 Richard Biener <rguenther@suse.de>
15537 PR tree-optimization/66623
15538 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15539 refactor check_reduction into two parts, properly computing
15540 whether we have to check reduction validity for outer loop
15543 2017-06-09 Richard Biener <rguenther@suse.de>
15545 PR tree-optimization/79483
15546 * graphite-scop-detection.c (order): New global.
15547 (get_order): Compute bb to order mapping that satisfies code
15548 generation constraints.
15549 (cmp_pbbs): New helper.
15550 (build_scops): Start domwalk at entry block, sort generated
15553 2017-06-09 Richard Biener <rguenther@suse.de>
15555 PR middle-end/81007
15556 * ipa-polymorphic-call.c
15557 (ipa_polymorphic_call_context::restrict_to_inner_class):
15558 Skip FIELD_DECLs with error_mark_node type.
15559 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15562 2017-06-09 Martin Liska <mliska@suse.cz>
15564 * predict.c (struct branch_predictor): New struct.
15565 (test_prediction_value_range): New test.
15566 (predict_c_tests): New function.
15567 * selftest-run-tests.c (selftest::run_tests): Run the function.
15568 * selftest.h: Declare new tests.
15570 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15573 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15574 gen_add3_insn did not fail.
15575 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15576 r0, construct that number in a temporary reg and add that reg to r0.
15577 If asked to put the result in r0 as well, fail.
15579 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15581 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15582 for early expansion of vec_eqv.
15584 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15586 PR middle-end/81005
15587 * ubsan.c (instrument_null): Avoid pointless code temporary.
15588 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15590 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15594 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15597 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15600 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15602 * predict.c (unlikely_executed_edge_p): Move ahead.
15603 (probably_never_executed_edge_p): Use it.
15605 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15607 PR middle-end/79988
15608 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15609 gimple_call_builtin_p call.
15611 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15613 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15614 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15615 rtl_check_failed_type2, rtl_check_failed_code1,
15616 rtl_check_failed_code2, rtl_check_failed_code_mode,
15617 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15618 rtvec_check_failed_bounds, rtl_check_failed_flag,
15619 _fatal_insn_not_found, _fatal_insn): Likewise.
15620 * tree.h (tree_contains_struct_check_failed,
15621 tree_check_failed, tree_not_check_failed,
15622 tree_class_check_failed, tree_range_check_failed,
15623 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15624 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15625 tree_operand_check_failed, omp_clause_check_failed,
15626 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15629 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15631 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15632 flag_branch_probabilities.
15633 * ipa-inline.c (edge_badness): Likewise.
15634 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15635 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15636 * predict.c (maybe_hot_frequency_p): Likewise.
15637 (probably_never_executed): Likewise.
15638 * sched-ebb.c (schedule_ebbs): Likewise.
15639 * sched-rgn.c (find_single_block_region): Likewise.
15640 * tracer.c (tail_duplicate): Likewise.
15642 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15644 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15645 longer requires x_flag_profile_use.
15647 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15649 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15650 instead of flag_reorder_blocks_and_partition.
15651 * dbxout.c (dbxout_function_end): Likewise.
15652 * dwarf2out.c (gen_subprogram_die): Likewise.
15653 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15654 * hw-doloop.c (reorg_loops): Likewise.
15655 * varasm.c (assemble_start_function,
15656 assemble_end_function): Likewise.
15657 (decide_function_section): Do not check for
15658 flag_reorder_blocks_and_partition.
15660 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15662 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15664 (chkp_get_hard_register_fake_addr_expr): Ditto.
15665 (chkp_build_addr_expr): Add check for hard reg case.
15666 (chkp_parse_array_and_component_ref): Ditto.
15667 (chkp_find_bounds_1): Ditto.
15668 (chkp_process_stmt): Don't generate bounds store for
15671 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15673 * predict.c (maybe_hot_bb_p): Do not check profile status.
15674 (maybe_hot_edge_p): Likewise.
15675 (probably_never_executed): Check for zero counts even if profile
15677 (unlikely_executed_edge_p): New function.
15678 (unlikely_executed_stmt_p): New function.
15679 (unlikely_executed_bb_p): New function.
15680 (set_even_probabilities): Use unlikely predicates.
15681 (combine_predictions_for_bb): Likewise.
15682 (predict_paths_for_bb): Likewise.
15683 (predict_paths_leading_to_edge): Likewise.
15684 (determine_unlikely_bbs): New function.
15685 (estimate_bb_frequencies): Use it.
15686 (compute_function_frequency): Use zero counts even if profile is
15688 * profile-count.h: Fix typo.
15690 2017-08-08 Julia Koval <julia.koval@intel.com>
15692 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15693 _mm512_mask_cvtsepi16_storeu_epi8,
15694 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15695 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15696 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15697 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15698 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15699 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15700 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15701 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15702 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15703 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15704 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15705 __builtin_ia32_pmovuswb256mem_mask,
15706 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15707 __builtin_ia32_pmovwb512mem_mask): New builtins.
15709 2017-08-08 Julia Koval <julia.koval@intel.com>
15711 PR target/73350,80862
15712 * config/i386/subst.md (round): Fix round pattern.
15713 * config/i386/i386.c (ix86_erase_embedded_rounding):
15714 Fix erasing rounding for the fixed pattern.
15716 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15718 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15720 2017-06-08 Martin Liska <mliska@suse.cz>
15722 PR gcov-profile/80911
15723 * gcov.c (block_info::block_info): New constructor.
15725 2017-06-07 Carl Love <cel@us.ibm.com>
15727 * config/rs6000/rs6000-c: The return type of the following
15728 built-in functions was implemented as int not long long. Fix sign
15729 of return value for the unsigned version of vec_mulo and vec_mule.
15730 vector unsigned long long vec_bperm (vector unsigned long long,
15731 vector unsigned char)
15732 vector signed long long vec_mule (vector signed int,
15734 vector unsigned long long vec_mule (vector unsigned int,
15735 vector unsigned int)
15736 vector signed long long vec_mulo (vector signed int,
15738 vector unsigned long long vec_mulo (vector unsigned int,
15739 vector unsigned int)
15740 * doc/extend.texi: Fix the documentation for the built-in
15743 2017-06-07 Carl Love <cel@us.ibm.com>
15746 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15749 2017-06-07 Carl Love <cel@us.ibm.com>
15751 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15752 support, Generate doublehv for signed int/float for BE case only.
15754 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15756 * doc/invoke.texi (mcx16): Rewrite.
15758 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15760 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15761 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15762 *mov<mode>_softfloat, and an anonymous splitter): Use
15763 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15765 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15767 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15769 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15770 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15771 (rs6000_debug_reg_global): Adjust.
15772 (rs6000_init_hard_regno_mode_ok): Adjust.
15773 (rs6000_dbx_register_number): Adjust.
15774 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15775 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15776 Remove SPE_ACC and SPEFSCR.
15777 (REG_ALLOC_ORDER): Ditto.
15778 (FRAME_POINTER_REGNUM): Change to 111.
15779 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15780 (REG_CLASS_NAMES): Ditto.
15781 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15782 (REGISTER_NAMES): Ditto.
15783 (ADDITIONAL_REG_NAMES): Ditto.
15784 (rs6000_reg_names): Ditto.
15785 * config/rs6000/rs6000.md: Renumber some register number
15788 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15790 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15792 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15793 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15795 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15796 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15797 Delete the SPE high registers.
15798 (REG_ALLOC_ORDER): Ditto.
15799 (enum reg_class): Remove SPE_HIGH_REGS.
15800 (REG_CLASS_NAMES): Ditto.
15801 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15802 (REGISTER_NAMES): Ditto.
15803 (rs6000_reg_names): Ditto.
15804 * doc/tm.texi.in: Remove SPE as example.
15805 * doc/tm.texi: Regenerate.
15807 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15809 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15810 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15811 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15812 * config/rs6000/rs6000.md (type): Remove "brinc".
15814 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15816 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15817 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15818 * config/rs6000/linuxspe.h: Delete file.
15819 * config/rs6000/rs6000.md: Don't include spe.md.
15820 * config/rs6000/spe.h: Delete file.
15821 * config/rs6000/spe.md: Delete file.
15822 * config/rs6000/t-rs6000: Remove spe.md.
15824 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15826 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15827 (reg_or_none500mem_operand): Delete.
15828 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15829 instead of reg_or_none500mem_operand.
15831 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15833 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15834 handling of SPE flags.
15835 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15837 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15839 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15841 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15842 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15843 paired_divv2sf3): Similar.
15844 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15845 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15846 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15848 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15849 Rename the paired_* instruction patterns.
15850 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15852 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15853 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15854 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15855 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15856 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15857 PAIRED_VECTOR_MODE.
15858 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15859 (spe_func_has_64bit_regs_p): Delete.
15860 (spe_expand_predicate_builtin): Delete.
15861 (spe_expand_evsel_builtin): Delete.
15862 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15863 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15864 (invalid_e500_subreg): Delete.
15865 (rs6000_legitimize_address): Always force_reg op2 as well, for
15866 paired single memory accesses.
15867 (rs6000_member_type_forces_blk): Delete.
15868 (rs6000_spe_function_arg): Delete.
15869 (rs6000_expand_unop_builtin): Delete SPE handling.
15870 (rs6000_expand_binop_builtin): Ditto.
15871 (spe_expand_stv_builtin): Delete.
15872 (bdesc_2arg_spe): Delete.
15873 (spe_expand_builtin): Delete.
15874 (spe_expand_predicate_builtin): Delete.
15875 (spe_expand_evsel_builtin): Delete.
15876 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15877 (spe_init_builtins): Delete.
15878 (spe_func_has_64bit_regs_p): Delete.
15879 (savres_routine_name): Delete "info" parameter. Adjust callers.
15880 (rs6000_emit_stack_reset): Ditto.
15881 (rs6000_dwarf_register_span): Delete.
15882 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15883 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15884 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15886 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15888 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15889 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15890 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15891 mulv2sf3, divv2sf3): Delete expanders.
15893 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15895 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15897 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15899 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15900 * config/rs6000/rs6000.c: Ditto.
15902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15904 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15905 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15906 comparison_operator.
15908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15910 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15911 * config/rs6000/rs6000.opt: Ditto.
15912 * config/rs6000/t-rtems: Ditto.
15914 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15916 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15917 TARGET_E500_SINGLE by 0, simplify.
15918 * config/rs6000/rs6000.c: Ditto.
15919 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15920 (spe_build_register_parallel): Delete.
15921 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15922 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15923 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15924 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15925 (E500_CONVERT): Delete.
15926 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15928 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15930 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15931 * config/rs6000/dfp.md: Ditto.
15932 (negdd2, *negdd2_fpr): Merge.
15933 (absdd2, *absdd2_fpr): Merge.
15934 (negtd2, *negtd2_fpr): Merge.
15935 (abstd2, *abstd2_fpr): Merge.
15936 * config/rs6000/e500.h: Delete file.
15937 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15938 TARGET_FPRS by 1 and simplify.
15939 * config/rs6000/rs6000-c.c: Ditto.
15940 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15941 TARGET_DF_SPE by 0.
15942 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15944 * config/rs6000/rs6000.md: Ditto.
15945 (floatdidf2, *floatdidf2_fpr): Merge.
15946 (move_from_CR_gt_bit): Delete.
15947 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15948 (E500_CR_IOR_COMPARE): Delete.
15949 (All patterns that require !TARGET_FPRS): Delete.
15950 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15952 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15954 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15956 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15958 * graphds.c (add_edge): Intitialize edge's attached data.
15959 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15960 pointer parameter. Call pointed function on each edge during
15961 graph traversing. Skip traversing the edge when the function
15963 (graphds_dfs, graphds_scc): Ditto.
15964 (for_each_edge): New parameter. Pass the new parameter to callback
15966 * graphds.h (skip_edge_callback): New function pointer type.
15967 (graphds_dfs, graphds_scc): New function pointer parameter.
15968 (graphds_edge_callback, for_each_edge): New parameter.
15970 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15972 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15973 out code checking if runtime alias check is possible to below ...
15974 Call the new function.
15975 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15976 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15978 2017-06-07 Marek Polacek <polacek@redhat.com>
15981 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15982 TYPE_OVERFLOW_WRAPS check.
15984 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15986 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15987 if versioning is required.
15988 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15989 peeling with the check for versioning.
15991 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15993 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15994 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15995 Set true to new parameter if new ssa variable is defined.
15996 (vect_gen_vector_loop_niters): Refactor. Set range information
15997 for the new vector loop bound variable.
15998 (vect_do_peeling): Ditto.
16000 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16002 * tree-affine.c (ssa.h): Include header file.
16003 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16004 has wrapping overflow behavior.
16006 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16008 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16010 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16012 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16013 (tree_to_aff_combination): ... here.
16015 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16017 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16018 reg_pressure model function.
16019 (ivopts_global_cost_for_size): Delete.
16020 (determine_set_costs, iv_ca_recount_cost): Call new model function
16021 ivopts_estimate_reg_pressure.
16023 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16025 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16026 expensive than udiv. Remove floating point cases from mod.
16028 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16030 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16031 Increase idiv cost.
16033 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16035 * config/aarch64/aarch64.md
16036 (copysignsf3): Fix mask generation.
16038 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16040 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16042 (class dump_manager): Add register_dumps method.
16043 * dumpfile.c: Include langhooks.h.
16044 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16045 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16046 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16047 (dump_manager::dump_register): Start with 512 entries instead of 32.
16048 (dump_manager::register_dumps): New method.
16049 * toplev.c (general_init): Instead of invoking register_dumps
16050 langhook, invoke register_dumps method on the dump manager.
16051 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16054 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16056 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16057 Say that # requires an associated define_split to exist, and that
16058 the define_split must be suitable for use after register allocation.
16060 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16062 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16063 (compute_outgoing_frequencies): Also initialize zero counts.
16064 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16065 around loops; preserve more of profile when nothing changes.
16067 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16069 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16071 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16072 * config/arm/arm-cpu-cdata.h: Regenerate.
16073 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16074 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16075 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16076 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16077 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16079 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16081 * config/arm/t-rmprofile: Likewise.
16082 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16084 2017-06-06 David S. Miller <davem@davemloft.net>
16087 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16088 blockage if function uses alloca.
16090 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16092 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16093 New "uid" fields to hold pretty-print IDs of group and ref.
16094 Memory references are now identified as <group_id>:<ref_id>
16095 instead of using [random] addresses.
16096 (dump_mem_details): Simplify, no functional change.
16097 (dump_mem_ref): Simplify and make output more concise.
16098 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16099 (find_or_create_group): Initialize group uid.
16100 (record_ref): Initialize ref uid. Improve debug output.
16101 (prune_group_by_reuse, should_issue_prefetch_p,)
16102 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16103 (mark_nontemporal_store, determine_loop_nest_reuse):
16104 Improve debug output.
16106 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16108 * dbgcnt.def (prefetch): New debug counter.
16109 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16110 (schedule_prefetches): Stop issueing prefetches if debug counter
16113 2017-06-06 Tom de Vries <tom@codesourcery.com>
16115 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16116 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16118 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16120 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16121 Use aarch64_reg_or_zero predicate for operand 4.
16122 (aarch64_compare_and_swap<mode> define_insn_and_split):
16123 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16124 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16126 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16128 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16129 (arm_compute_save_core_reg_mask): This.
16130 (thumb1_compute_save_reg_mask): Rename into ...
16131 (thumb1_compute_save_core_reg_mask): This.
16132 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16133 (arm_compute_frame_layout): Likewise.
16135 2017-06-06 Richard Biener <rguenther@suse.de>
16137 PR tree-optimization/80974
16138 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16139 keep or clear leaders SSA info.
16141 2017-06-06 Tom de Vries <tom@codesourcery.com>
16143 * config/nvptx/nvptx.c (split_mode_p): New function.
16144 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16146 2017-06-06 Tom de Vries <tom@codesourcery.com>
16148 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16150 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16153 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16156 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16158 * shrink-wrap.c (handle_simple_exit): Update profile.
16159 (try_shrink_wrapping): Upate profile.
16161 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16163 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16164 (tree_guess_outgoing_edge_probabilities): New.
16165 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16166 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16168 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16170 * ipa-split.c (split_function): Initialize return bb profile.
16172 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16174 * profile.c (compute_branch_probabilities): Also initialize
16175 EXIT_BLOCK profile.
16177 2017-06-06 Richard Biener <rguenther@suse.de>
16179 PR tree-optimization/80928
16180 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16181 (vect_analyze_loop_operations): Properly guard analysis for
16183 (vect_transform_loop): Likewise.
16184 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16185 (vect_model_induction_cost): Do not cost for pure SLP.
16186 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16187 of induction in inner loop vectorization.
16188 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16189 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16190 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16192 (vect_analyze_slp_cost_1): Cost induction.
16193 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16194 (vect_get_slp_vect_defs): Likewise.
16195 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16196 (vect_transform_stmt): Handle SLP reductions.
16197 * tree-vectorizer.h (vectorizable_induction): Adjust.
16199 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16201 * config/rs6000/rs6000.c (make_resolver_func): Update
16202 init_lowered_empty_function call.
16204 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16206 * doc/invoke.texi: Document the -fprofile-abs-path option.
16207 * common.opt (fprofile-abs-path): New option.
16208 * gcov-io.h (gcov_write_filename): Declare.
16209 * gcov-io.c (gcov_write_filename): New function.
16210 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16211 * profile.c (output_location): Likewise.
16213 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16215 * shring-wrap.c: Revert accidental commit.
16217 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16219 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16221 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16223 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16225 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16226 profile in callgraph edge.
16227 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16228 (apply_scale): Likewise.
16229 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16230 Also scale profile when inlining function with zero profile.
16231 (initialize_cfun): Update exit block profile even when it is zero.
16232 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16233 when profile is read.
16235 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16237 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16238 (CLONE_*): New constants to define the processors we can generate
16239 code for with the target_clone attribute.
16240 (rs6000_clone_map): New array to identify which clone processors
16241 the current program is running on.
16242 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16243 target_clone attribute.
16244 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16245 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16246 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16247 (cpu_expand_builtin): Add support for target_clone attribute.
16248 (rs6000_valid_attribute_p): Allow "default" attribute.
16249 (get_decl_name): New debug function to simplify printing the
16250 current function name in debugging statements.
16251 (rs6000_clone_priority): New functions to support the target_clone
16252 attribute, and be able to generate code to switch between ISA 2.05
16253 through ISA 3.0 (power6 through power9).
16254 (rs6000_compare_version_priority): Likewise.
16255 (rs6000_get_function_versions_dispatcher): Likewise.
16256 (make_resolver_func): Likewise.
16257 (add_condition_to_bb): Likewise.
16258 (dispatch_function_versions): Likewise.
16259 (rs6000_generate_version_dispatcher_body): Likewise.
16260 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16261 (fusion_gpr_load_p): Fix a spacing issue.
16262 * doc/extend.texi (Common Function Attributes): Document that the
16263 PowerPC supports the target_clone attribute.
16265 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16267 * config/arm/arm.h: explain F symbol found in description of ARM
16268 register allocation in its legend.
16270 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16272 * config/mips/frame-header-opt.c: Include profile-count.h.
16273 * config/riscv/riscv.c: Include profile-count.h
16275 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16277 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16279 (sm_set_flag_if_changed): Add bbs field.
16280 (execute_sm_if_changed_flag_set): Pass BBS.
16281 (execute_sm): Update.
16283 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16285 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16288 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16290 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16291 (peephole2): New peephole2 to emit the above.
16292 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16294 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16296 * config/aarch64/aarch64.c (define_peephole2 above
16297 *sub_<shift>_<mode>): New peephole.
16299 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16301 * config/i386/i386.c (make_resolver_func): Update.
16302 * Makefile.in: Add profile-count.h and profile-count.o
16303 * auto-profile.c (afdo_indirect_call): Update to new API.
16304 (afdo_set_bb_count): Update.
16305 (afdo_propagate_edge): Update.
16306 (afdo_propagate_circuit): Update.
16307 (afdo_calculate_branch_prob): Update.
16308 (afdo_annotate_cfg): Update.
16309 * basic-block.h: Include profile-count.h
16310 (struct edge_def): Turn count to profile_count.
16311 (struct basic_block_def): Likewie.
16312 (REG_BR_PROB_BASE): Move to profile-count.h
16313 (RDIV): Move to profile-count.h
16314 * bb-reorder.c (max_entry_count): Turn to profile_count.
16315 (find_traces): Update.
16316 (rotate_loop):Update.
16317 (connect_traces):Update.
16318 (sanitize_hot_paths):Update.
16319 * bt-load.c (migrate_btr_defs): Update.
16320 * cfg.c (RDIV): Remove.
16321 (init_flow): Use alloc_block.
16322 (alloc_block): Uninitialize count.
16323 (unchecked_make_edge): Uninitialize count.
16324 (check_bb_profile): Update.
16325 (dump_edge_info): Update.
16326 (dump_bb_info): Update.
16327 (update_bb_profile_for_threading): Update.
16328 (scale_bbs_frequencies_int): Update.
16329 (scale_bbs_frequencies_gcov_type): Update.
16330 (scale_bbs_frequencies_profile_count): New.
16331 * cfg.h (update_bb_profile_for_threading): Update.
16332 (scale_bbs_frequencies_profile_count): Declare.
16333 * cfgbuild.c (compute_outgoing_frequencies): Update.
16334 (find_many_sub_basic_blocks): Update.
16335 * cfgcleanup.c (try_forward_edges): Update.
16336 (try_crossjump_to_edge): Update.
16337 * cfgexpand.c (expand_gimple_tailcall): Update.
16338 (construct_exit_block): Update.
16339 * cfghooks.c (verify_flow_info): Update.
16340 (dump_bb_for_graph): Update.
16341 (split_edge): Update.
16342 (make_forwarder_block): Update.
16343 (duplicate_block): Update.
16344 (account_profile_record): Update.
16345 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16346 (get_estimated_loop_iterations): Update.
16347 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16348 (single_likely_exit): Update.
16349 * cfgloopmanip.c (scale_loop_profile): Update.
16351 (set_zero_probability): Update.
16352 (lv_adjust_loop_entry_edge): Update.
16353 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16354 (purge_dead_edges): Update.
16355 (rtl_account_profile_record): Update.
16356 * cgraph.c (cgraph_node::create): Uninitialize count.
16357 (symbol_table::create_edge): Uninitialize count.
16358 (cgraph_update_edges_for_call_stmt_node): Update.
16359 (cgraph_edge::dump_edge_flags): Update.
16360 (cgraph_node::dump): Update.
16361 (cgraph_edge::maybe_hot_p): Update.
16362 * cgraph.h: Include profile-count.h
16363 (create_clone), create_edge, create_indirect_edge): Update.
16364 (cgraph_node): Turn count to profile_count.
16365 (cgraph_edge0: Likewise.
16366 (make_speculative, clone): Update.
16367 (create_edge): Update.
16368 (init_lowered_empty_function): Update.
16369 * cgraphclones.c (cgraph_edge::clone): Update.
16370 (duplicate_thunk_for_node): Update.
16371 (cgraph_node::create_clone): Update.
16372 * cgraphunit.c (cgraph_node::analyze): Update.
16373 (cgraph_node::expand_thunk): Update.
16374 * final.c (dump_basic_block_info): Update.
16375 * gimple-streamer-in.c (input_bb): Update.
16376 * gimple-streamer-out.c (output_bb): Update.
16377 * graphite.c (print_global_statistics): Update.
16378 (print_graphite_scop_statistics): Update.
16379 * hsa-brig.c: Include basic-block.h.
16380 * hsa-dump.c: Include basic-block.h.
16381 * hsa-gen.c (T sum_slice): Update.
16382 (convert_switch_statements):Update.
16383 * hsa-regalloc.c: Include basic-block.h.
16384 * ipa-chkp.c (chkp_produce_thunks): Update.
16385 * ipa-cp.c (struct caller_statistics): Update.
16386 (init_caller_stats): Update.
16387 (gather_caller_stats): Update.
16388 (ipcp_cloning_candidate_p): Update.
16389 (good_cloning_opportunity_p): Update.
16390 (get_info_about_necessary_edges): Update.
16391 (dump_profile_updates): Update.
16392 (update_profiling_info): Update.
16393 (update_specialized_profile): Update.
16394 (perhaps_add_new_callers): Update.
16395 (decide_about_value): Update.
16396 (ipa_cp_c_finalize): Update.
16397 * ipa-devirt.c (struct odr_type_warn_count): Update.
16398 (struct decl_warn_count): Update.
16399 (struct final_warning_record): Update.
16400 (possible_polymorphic_call_targets): Update.
16401 (ipa_devirt): Update.
16402 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16403 * ipa-icf.c (sem_function::merge): Update.
16404 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16405 * ipa-inline.c (compute_uninlined_call_time): Update.
16406 (compute_inlined_call_time): Update.
16407 (want_inline_small_function_p): Update.
16408 (want_inline_self_recursive_call_p): Update.
16409 (edge_badness): Update.
16410 (lookup_recursive_calls): Update.
16411 (recursive_inlining): Update.
16412 (inline_small_functions): Update.
16413 (dump_overall_stats): Update.
16414 (dump_inline_stats): Update.
16415 * ipa-profile.c (ipa_profile_generate_summary): Update.
16416 (ipa_propagate_frequency): Update.
16417 (ipa_profile): Update.
16418 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16419 * ipa-utils.c (ipa_merge_profiles): Update.
16420 * loop-doloop.c (doloop_modify): Update.
16421 * loop-unroll.c (report_unroll): Update.
16422 (unroll_loop_runtime_iterations): Update.
16423 * lto-cgraph.c (lto_output_edge): Update.
16424 (lto_output_node): Update.
16425 (input_node): Update.
16426 (input_edge): Update.
16427 (merge_profile_summaries): Update.
16428 * lto-streamer-in.c (input_cfg): Update.
16429 * lto-streamer-out.c (output_cfg): Update.
16430 * mcf.c (create_fixup_graph): Update.
16431 (adjust_cfg_counts): Update.
16432 (sum_edge_counts): Update.
16433 * modulo-sched.c (sms_schedule): Update.
16434 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16435 * predict.c (maybe_hot_count_p): Update.
16436 (probably_never_executed): Update.
16437 (dump_prediction): Update.
16438 (combine_predictions_for_bb): Update.
16439 (propagate_freq): Update.
16440 (handle_missing_profiles): Update.
16441 (counts_to_freqs): Update.
16442 (rebuild_frequencies): Update.
16443 (force_edge_cold): Update.
16444 * predict.h: Include profile-count.h
16445 (maybe_hot_count_p, counts_to_freqs): UPdate.
16446 * print-rtl-function.c: Do not include cfg.h
16447 * print-rtl.c: Include basic-block.h
16448 * profile-count.c: New file.
16449 * profile-count.h: New file.
16450 * profile.c (is_edge_inconsistent): Update.
16451 (correct_negative_edge_counts): Update.
16452 (is_inconsistent): Update.
16453 (set_bb_counts): Update.
16454 (read_profile_edge_counts): Update.
16455 (compute_frequency_overlap): Update.
16456 (compute_branch_probabilities): Update; Initialize and deinitialize
16458 (branch_prob): Update.
16459 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16460 (edge_gcov_count): New.
16461 (bb_gcov_count): New.
16462 * shrink-wrap.c (try_shrink_wrapping): Update.
16463 * tracer.c (better_p): Update.
16464 * trans-mem.c (expand_transaction): Update.
16465 (ipa_tm_insert_irr_call): Update.
16466 (ipa_tm_insert_gettmclone_call): Update.
16467 * tree-call-cdce.c: Update.
16468 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16469 (gimple_duplicate_sese_tail): Update.
16470 (gimple_account_profile_record): Update.
16471 (execute_fixup_cfg): Update.
16472 * tree-inline.c (copy_bb): Update.
16473 (copy_edges_for_bb): Update.
16474 (initialize_cfun): Update.
16475 (freqs_to_counts): Update.
16476 (copy_cfg_body): Update.
16477 (expand_call_inline): Update.
16478 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16479 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16480 (try_unroll_loop_completely): Update.
16481 (try_peel_loop): Update.
16482 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16483 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16484 * tree-ssa-loop-split.c (connect_loops): Update.
16485 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16486 * tree-ssa-reassoc.c (branch_fixup): Update.
16487 * tree-ssa-tail-merge.c (replace_block_by): Update.
16488 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16489 (compute_path_counts): Update.
16490 (update_profile): Update.
16491 (recompute_probabilities): Update.
16492 (update_joiner_offpath_counts): Update.
16493 (estimated_freqs_path): Update.
16494 (freqs_to_counts_path): Update.
16495 (clear_counts_path): Update.
16496 (ssa_fix_duplicate_block_edges): Update.
16497 (duplicate_thread_path): Update.
16498 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16499 (struct switch_conv_info): Update.
16500 * tree-tailcall.c (decrease_profile): Update.
16501 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16502 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16503 * value-prof.c (check_counter): Update.
16504 (gimple_divmod_fixed_value): Update.
16505 (gimple_mod_pow2): Update.
16506 (gimple_mod_subtract): Update.
16507 (gimple_ic_transform): Update.
16508 (gimple_stringop_fixed_value): Update.
16509 * value-prof.h (gimple_ic): Update.
16511 2017-06-02 Carl Love <cel@us.ibm.com>
16513 * config/rs6000/rs6000-c: Add support for built-in functions
16514 vector double vec_doublee (vector signed int);
16515 vector double vec_doublee (vector unsigned int);
16516 vector double vec_doublee (vector float);
16517 vector double vec_doubleh (vector signed int);
16518 vector double vec_doubleh (vector unsigned int);
16519 vector double vec_doubleh (vector float);
16520 vector double vec_doublel (vector signed int);
16521 vector double vec_doublel (vector unsigned int);
16522 vector double vec_doublel (vector float);
16523 vector double vec_doubleo (vector signed int);
16524 vector double vec_doubleo (vector unsigned int);
16525 vector double vec_doubleo (vector float);.
16526 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16527 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16529 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16530 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16531 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16533 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16534 vec_doublel, vec_doubleh.
16535 * doc/extend.texi: Update the built-in documentation file for the
16536 new built-in functions.
16538 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16541 * ipa-inline-analysis.c (free_growth_caches): Set
16542 edge_removal_hook_holder to NULL after removing it.
16544 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16546 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16547 comparision with zero.
16549 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16550 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16551 for early expansion of vec_min and vec_max builtins.
16552 (builtin_function_type): Add min/max unsigned variants to those
16553 identified as having unsigned arguments.
16555 2017-06-02 Olivier Hainque <hainque@adacore.com>
16557 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16559 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16561 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16562 Use VALL_F16 iterator rather than VALL.
16564 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16566 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16567 Emit CBNZ inside loop when doing a strong exchange and comparing
16568 against zero. Generate the CC flags after the loop.
16570 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16572 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16573 (dl_section_ref): New.
16574 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16575 On AIX, append an expression to subtract the size of the
16576 section length to dl_section_ref.
16578 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16580 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16581 for early expansion of vector absolute builtins.
16583 2017-06-02 Richard Biener <rguenther@suse.de>
16585 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16586 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16588 2017-06-02 Richard Biener <rguenther@suse.de>
16590 PR tree-optimization/80948
16591 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16592 stmt order as well.
16594 2017-06-02 Richard Biener <rguenther@suse.de>
16596 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16598 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16599 for inductions as relevant.
16601 2017-06-02 Richard Biener <rguenther@suse.de>
16603 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16604 (vectorizable_induction): ... this. Remove dead code.
16606 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16608 * builtins. (expand_builtin_alloca): Remove second parameter and
16609 infer its value from the first parameter instead.
16610 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16612 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16614 PR rtl-optimization/80903
16615 * loop-doloop.c (add_test): Unshare sequence.
16617 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16619 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16621 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16623 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16625 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16626 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16627 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16628 (xlogue_layout::get_stub_rtx): Make static.
16629 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16630 (xlogue_layout::compute_stub_managed_regs): Rename to...
16631 (xlogue_layout::count_stub_managed_regs): ...this.
16632 (xlogue_layout::is_stub_managed_reg): New function.
16633 (xlogue_layout::m_stub_names): Rename to...
16634 (xlogue_layout::s_stub_names): ...this, make static.
16635 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16636 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16637 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16638 xlogue_layout::s_stub_names): Instantiate statics.
16639 (stub_managed_regs): Remove.
16640 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16641 (disable_call_ms2sysv_xlogues): Rename to...
16642 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16643 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16645 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16646 change after reload_completed.
16647 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16649 (ix86_expand_prologue): Likewise.
16650 (ix86_expand_epilogue): Likewise.
16651 (ix86_expand_split_stack_prologue): Likewise.
16652 (ix86_compute_frame_layout): Remove frame parameter ...
16653 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16654 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16656 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16657 (ix86_frame): Move from here ...
16658 * config/i386/i386.h (ix86_frame): ... to here.
16659 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16660 complete ix86_frame data structure instead. Remove some_ld_name.
16662 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16664 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16665 symbols that hold a DECL_VALUE_EXPR.
16667 2017-06-01 Martin Jambor <mjambor@suse.cz>
16669 PR tree-optimization/80898
16670 * tree-sra.c (process_subtree_disqualification): Removed.
16671 (disqualify_candidate): Do not acll
16672 process_subtree_disqualification.
16673 (subtree_mark_written_and_enqueue): New function.
16674 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16675 RHS has been disqualified and re-queue LHS if necessary. Apart
16676 from that, ignore disqualified RHS.
16678 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16680 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16681 address fetch for z10 or later.
16683 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16685 * config/arc/arc.md (tst_movb): Add guard when splitting.
16687 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16689 * config/arc/arc.c (arc_can_eliminate): Test against
16690 arc_frame_pointer_needed.
16692 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16694 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16695 to prevent store reordering.
16696 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16697 (type): Add block type.
16698 (stack_tie): Define special instruction to be used in
16701 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16703 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16704 constraint. It is not valid for the pattern.
16705 (noncommutative_binary_comparison): Likewise.
16707 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16709 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16712 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16714 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16715 be used by the reg-alloc.
16717 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16719 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16720 reg-alloc when having mul64 or mul32x16 instructions.
16721 (mulsidi3): Likewise.
16722 (umulsidi3): Likewise.
16723 (mulsi32x16): New pattern.
16724 (mulsi64): Likewise.
16725 (mulsidi64): Likewise.
16726 (umulsidi64): Likewise.
16727 (MUL32x16_REG): Define.
16728 (mul64_600): Use MUL32x16_REG.
16729 (mac64_600): Likewise.
16730 (umul64_600): Likewise.
16731 (umac64_600): Likewise.
16733 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16735 * config/arc/arc.md (mulsi3_700): Make it commutative.
16737 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16739 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16741 (*sign_extendsidi2_insn): Likewise for movstosw.
16743 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16745 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16746 the type of the input discriminant value. Convert the
16747 discriminant value of signedness vary.
16749 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16751 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16752 Add to -Wall section.
16754 2017-06-01 Richard Biener <rguenther@suse.de>
16756 PR middle-end/66313
16757 * fold-const.c (fold_plusminus_mult_expr): If the factored
16758 factor may be zero use a wrapping type for the inner operation.
16759 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16760 and handle moved defs.
16761 (process_assignment): Properly guard the unary op case. Return a
16762 tri-state indicating that moving the stmt before the call may allow
16763 to continue. Pass through to_move.
16764 (find_tail_calls): Handle moving unrelated defs before
16767 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16770 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16771 splitter result in the canonical way.
16773 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16775 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16776 also for 32bit target. Update insn attributes.
16777 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16779 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16781 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16782 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16783 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16784 (_mm_maskz_min_ss): New intrinsics.
16786 2017-05-31 Martin Liska <mliska@suse.cz>
16788 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16789 Change comment style to one we normally use.
16790 (vectorizable_reduction): Likewise.
16791 (vectorizable_induction): Likewise.
16792 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16793 (vectorizable_call): Likewise.
16794 (vectorizable_simd_clone_call): Likewise.
16795 (vectorizable_conversion): Likewise.
16796 (vectorizable_assignment): Likewise.
16797 (vectorizable_shift): Likewise.
16798 (vectorizable_operation): Likewise.
16799 (vectorizable_store): Likewise.
16800 (vectorizable_load): Likewise.
16801 * tree-vectorizer.h: Likewise.
16803 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16805 * passes.c (emergency_dump_function): New.
16806 * tree-pass.h (emergency_dump_function): Declare.
16807 * plugin.c (plugins_internal_error_function): Remove.
16808 * plugin.h (plugins_internal_error_function): Remove declaration.
16809 * toplev.c (internal_error_function): New static function. Use it...
16810 (general_init): ...here.
16812 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16814 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16815 (arc_rtx_costs): Add costs for new patterns.
16816 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16817 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16819 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16821 * tree-ssa-strlen.c (get_next_strinfo): New function.
16822 (get_stridx_plus_constant): Use it.
16823 (zero_length_string): Likewise.
16824 (adjust_related_strinfos): Likewise.
16825 (adjust_last_stmt): Likewise.
16827 2017-05-31 Richard Biener <rguenther@suse.de>
16830 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16831 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16833 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16835 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16836 loop_vinfo argument and use of dependence distance vectors.
16837 Check instead whether the two references differ only in their
16838 initial value and assume that they have the same alignment if the
16839 difference is a multiple of the vector alignment.
16840 (vect_analyze_data_refs_alignment): Update call accordingly.
16842 2017-05-31 Martin Liska <mliska@suse.cz>
16845 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16847 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16849 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16850 (create_intersect_range_checks): Move from ...
16851 * tree-data-ref.c (create_intersect_range_checks_index)
16852 (create_intersect_range_checks): ... to here.
16853 (create_runtime_alias_checks): New function factored from ...
16854 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16855 here. Call above function.
16856 * tree-data-ref.h (create_runtime_alias_checks): New function.
16858 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16860 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16861 segment length for dr_b and compute it in wide_int.
16863 2017-05-31 Richard Biener <rguenther@suse.de>
16865 PR tree-optimization/80906
16866 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16867 and pass through iv_map.
16868 (copy_bb_and_scalar_dependences): Adjust.
16869 (translate_pending_phi_nodes): Likewise.
16870 (copy_loop_close_phi_args): Handle code-generating IVs instead
16873 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16875 * diagnostic-color.c (color_dict): Add "type-diff".
16876 (parse_gcc_colors): Update comment.
16877 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16878 -fdiagnostics-show-template-tree and -fno-elide-type.
16879 (GCC_COLORS): Add type-diff to example.
16881 (-fdiagnostics-show-template-tree): New.
16882 (-fno-elide-type): New.
16883 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16884 the pp_format_decoder callback. Call any m_format_postprocessor's
16886 (pretty_printer::pretty_printer): Initialize
16887 m_format_postprocessor.
16888 (pretty_printer::~pretty_printer): Delete any
16889 m_format_postprocessor.
16890 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16891 (class format_postprocessor): New class.
16892 (struct pretty_printer::format_decoder): Document the new parameters.
16893 (struct pretty_printer::m_format_postprocessor): New field.
16894 * tree-diagnostic.c (default_tree_printer): Update for new
16895 bool and const char ** params.
16896 * tree-diagnostic.h (default_tree_printer): Likewise.
16898 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16900 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16901 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16902 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16903 rs6000_gen_cell_microcode code.
16904 (rs6000_final_prescan_insn): Delete.
16905 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16906 "warn-cell-microcode" entries.
16907 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16908 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16909 throughout. Change cc_reg_not_micro_cr0_operand to
16910 cc_reg_not_cr0_operand throughout.
16911 (*extendhi<mode>2_noload): Delete.
16912 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16913 (mwarn-cell-microcode): Delete.
16914 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16915 -mgen-cell-microcode and -mwarn-cell-microcode.
16917 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16920 * config/i386/constraints.md (Yd): New constraint.
16922 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16923 and (?Yd, r) alternatives. Update insn attributes.
16924 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16925 and (?*Yd, r) alternatives. Update insn attributes.
16926 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16928 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16930 * gimplify.c (gimplify_modify_expr): Don't create a
16931 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16934 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16936 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16938 2017-05-30 Richard Biener <rguenther@suse.de>
16940 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16941 and reduc_def fields.
16942 (STMT_VINFO_REDUC_TYPE): New define.
16943 (STMT_VINFO_REDUC_DEF): Likewise.
16944 (vect_force_simple_reduction): Adjust prototype.
16945 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16946 (vect_is_simple_reduction): Remove check_reduction argument.
16947 (vect_force_simple_reduction): Adjust and set
16948 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16949 (vectorizable_reduction): Do not re-do reduction analysis
16950 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16951 * tree-parloops.c (gather_scalar_reductions): Adjust.
16953 2017-05-30 Richard Biener <rguenther@suse.de>
16955 PR middle-end/80901
16956 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16959 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16961 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16962 Introduce unknown_misalignment parameter and remove vf.
16963 (vect_peeling_hash_get_lowest_cost):
16964 Pass unknown_misalignment parameter.
16965 (vect_enhance_data_refs_alignment):
16966 Fix unsupportable data ref treatment.
16968 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16970 * tree-vect-data-refs.c (vect_get_data_access_cost):
16971 Workaround for SLP handling.
16972 (vect_enhance_data_refs_alignment):
16973 Compute costs for doing no peeling at all, compare to the best
16974 peeling costs so far and avoid peeling if cheaper.
16976 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16978 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16979 Return peeling info and set costs to zero for unlimited cost
16981 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16982 with unknown misalignment. Compute and costs for unknown
16983 misalignment, compare them to the costs for known misalignment
16984 and choose the cheapest for peeling.
16986 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16988 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16989 (vect_get_peeling_costs_all_drs): Create function.
16990 (vect_peeling_hash_get_lowest_cost):
16991 Use vect_get_peeling_costs_all_drs.
16992 (vect_peeling_supportable): Create function.
16993 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16995 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16997 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16998 DR_HAS_NEGATIVE_STEP.
16999 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17000 (vect_enhance_data_refs_alignment): Use.
17001 (vect_duplicate_ssa_name_ptr_info): Use.
17002 * tree-vectorizer.h (dr_misalignment): Use.
17003 (known_alignment_for_access_p): Use.
17005 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17008 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17010 (has_section_name): New function.
17012 2017-05-30 Martin Liska <mliska@suse.cz>
17015 * auto-profile.c (get_function_decl_from_block): Fix
17018 2017-05-30 Richard Biener <rguenther@suse.de>
17020 PR middle-end/80876
17021 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17023 2017-05-30 Martin Liska <mliska@suse.cz>
17025 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17026 * dumpfile.h (struct dump_file_info): Remove ctors.
17028 2017-05-30 Martin Liska <mliska@suse.cz>
17030 * predict.def: Fix GNU coding style.
17032 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17034 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17035 Mark 'to' argument with ATTRIBUTE_UNUSED.
17037 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17039 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17040 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17041 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17044 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17046 * doc/install.texi (Options specification): Restore entry of
17047 --enable-sjlj-exceptions.
17049 2017-05-27 Michael Eager <eager@eagercon.com>
17052 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17054 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17056 * config/microblaze/microblaze.h
17057 (FIXED_REGISTERS): Update in macro.
17058 (CALL_USED_REGISTERS): Update in macro.
17060 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17062 * doc/install.texi: Add links to macOS binary distributions.
17064 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17068 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17070 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17072 2017-05-26 Martin Liska <mliska@suse.cz>
17074 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17076 2017-05-26 Martin Liska <mliska@suse.cz>
17078 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17079 always leading ';; '.
17080 (dump_bb_info): Likewise.
17081 (brief_dump_cfg): Likewise.
17082 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17083 * dumpfile.c: Remove usage of TDF_VERBOSE.
17084 * dumpfile.h (enum dump_kind): Likewise.
17085 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17086 * print-tree.c (debug_verbose): Remove.
17087 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17088 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17089 * tree-diagnostic.c (default_tree_printer): Replace
17090 TDF_DIAGNOSTIC with TDF_SLIM.
17092 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17094 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17095 in parameter loop, rather than loop_vinfo.
17096 (create_intersect_range_checks): Ditto.
17097 (vect_create_cond_for_alias_checks): Update call to above functions.
17099 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17101 PR tree-optimization/80815
17102 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17103 for merging runtime alias checks. Handle negative DR_STEPs.
17105 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17107 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17109 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17110 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17111 out code pruning runtime alias checks.
17112 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17113 factored out from above.
17114 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17116 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17118 (prune_runtime_alias_test_list): New decalaration.
17120 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17122 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17123 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17124 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17125 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17126 (operator==, comp_dr_with_seg_len_pair): Ditto.
17127 (vect_prune_runtime_alias_test_list): Ditto.
17129 2017-05-26 Martin Liska <mliska@suse.cz>
17132 * params.def: Bound partial-inlining-entry-probability param.
17134 2017-05-26 Marek Polacek <polacek@redhat.com>
17137 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17140 2017-05-26 Richard Biener <rguenther@suse.de>
17142 PR tree-optimization/80842
17143 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17146 2017-05-26 Richard Biener <rguenther@suse.de>
17148 PR tree-optimization/80844
17149 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17151 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17153 * doc/md.texi (Machine Constraints): Update x86 family
17154 machine constraints section to match 'config/i386/constraints.md'.
17156 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17158 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17160 2017-05-25 Nathan Sidwell <nathan@acm.org>
17162 * doc/invoke.texi (--enable-languages): Update documentation.
17164 2017-05-25 Martin Liska <mliska@suse.cz>
17166 * dumpfile.c: Add TDF_FOLDING.
17167 * dumpfile.h (enum dump_kind): Likewise.
17168 * genmatch.c (dt_simplify::gen_1): Use it.
17170 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17172 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17174 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17176 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17177 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17179 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17181 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17182 * match.pd (X == C): Rewrite it here.
17183 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17184 with_certain_nonzero_bits2): New predicates.
17185 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17187 2017-05-24 Nathan Sidwell <nathan@acm.org>
17189 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17192 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17195 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17197 * config/powerpcspe: New port. Files are copied from the rs6000
17198 port, with "rs6000" in filenames replaced by "powerpcspe".
17200 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17202 PR rtl-optimization/80754
17203 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17205 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17207 * config/sparc/sparc.md (length): Return the correct value for -mflat
17208 sibcalls to match output_sibcall.
17210 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17214 * config/rs6000/rs6000.c (struct machine_function): Add new field
17216 (rs6000_get_separate_components): Init that field, use it.
17217 (rs6000_components_for_bb): Use the field.
17219 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17221 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17223 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17225 PR middle-end/80823
17226 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17228 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17231 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17232 address against address_operand predicate.
17233 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17235 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17237 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17238 parameters passed indirectly.
17240 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17242 * config/i386/i386.md (*movdi_internal): Remove SSE4
17243 alternative 18 (?r, *v). Update insn attributes.
17244 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17245 Update insn attributes.
17246 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17247 Update insn attributes.
17248 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17249 alternative 1 (r, v). Remove isa attribute.
17250 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17251 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17252 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17254 2017-05-23 Tom de Vries <tom@codesourcery.com>
17256 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17259 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17261 * cgraphunit.c (symbol_table::process_new_functions): Update.
17262 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17263 (inline_generate_summary): Rename to ...
17264 (ipa_fn_summary_generate): ... this one.
17265 (inline_read_summary): Rename to ...
17266 (ipa_fn_summary_read): ... this one.
17267 (inline_write_summary): Rename to ...
17268 (ipa_fn_summary_write): ... this one.
17269 (inline_free_summary): Rename to ...
17270 (ipa_free_fn_summary): ... this one.
17271 (pass_data_local_fn_summary, pass_local_fn_summary,
17272 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17273 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17274 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17275 make_pass_ipa_fn_summary): New.
17276 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17277 inline_write_summary, inline_free_summary): Remove.
17278 (ipa_free_fn_summary) : New.
17279 * ipa-inline.c (ipa_inline): Update.
17280 (pass_ipa_inline): Do not generate summaries.
17281 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17283 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17284 and add pass_ipa_fn_summary.
17285 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17287 (make_pass_inline_parameters): Remove.
17289 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17291 * omp-low.c (struct omp_context): Remove "default_kind" member.
17294 * omp-offload.c (execute_oacc_device_lower): Remove the
17295 parallelism dimensions function attributes for unparallelized
17296 OpenACC kernels constructs.
17298 2017-05-23 Martin Liska <mliska@suse.cz>
17300 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17302 (cgraph_edge::make_speculative): Likewise.
17303 (cgraph_edge::resolve_speculation): Likewise.
17304 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17305 (cgraph_node::dump): Likewise.
17306 * cgraph.h: Likewise.
17307 * cgraphunit.c (analyze_functions): Likewise.
17308 (symbol_table::compile): Likewise.
17309 * ipa-cp.c (print_all_lattices): Likewise.
17310 (determine_versionability): Likewise.
17311 (initialize_node_lattices): Likewise.
17312 (ipcp_verify_propagated_values): Likewise.
17313 (estimate_local_effects): Likewise.
17314 (update_profiling_info): Likewise.
17315 (create_specialized_node): Likewise.
17316 (perhaps_add_new_callers): Likewise.
17317 (decide_about_value): Likewise.
17318 (decide_whether_version_node): Likewise.
17319 (identify_dead_nodes): Likewise.
17320 (ipcp_store_bits_results): Likewise.
17321 * ipa-devirt.c (dump_targets): Likewise.
17322 (ipa_devirt): Likewise.
17323 * ipa-icf.c (sem_item::dump): Likewise.
17324 (sem_function::equals): Likewise.
17325 (sem_variable::equals): Likewise.
17326 (sem_item_optimizer::read_section): Likewise.
17327 (sem_item_optimizer::execute): Likewise.
17328 (congruence_class::dump): Likewise.
17329 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17330 (dump_inline_summary): Likewise.
17331 (estimate_node_size_and_time): Likewise.
17332 (inline_analyze_function): Likewise.
17333 * ipa-inline-transform.c (inline_call): Likewise.
17334 * ipa-inline.c (report_inline_failed_reason): Likewise.
17335 (want_early_inline_function_p): Likewise.
17336 (edge_badness): Likewise.
17337 (update_edge_key): Likewise.
17338 (inline_small_functions): Likewise.
17339 * ipa-profile.c (ipa_profile): Likewise.
17340 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17341 (ipa_make_edge_direct_to_target): Likewise.
17342 (remove_described_reference): Likewise.
17343 (ipa_impossible_devirt_target): Likewise.
17344 (propagate_controlled_uses): Likewise.
17345 (ipa_print_node_params): Likewise.
17346 (ipcp_transform_function): Likewise.
17347 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17348 (propagate_pure_const): Likewise.
17349 * ipa-reference.c (generate_summary): Likewise.
17350 (read_write_all_from_decl): Likewise.
17351 (propagate): Likewise.
17352 (ipa_reference_read_optimization_summary): Likewise.
17353 * ipa-utils.c (ipa_merge_profiles): Likewise.
17354 * ipa.c (walk_polymorphic_call_targets): Likewise.
17355 (symbol_table::remove_unreachable_nodes): Likewise.
17356 (ipa_single_use): Likewise.
17357 * passes.c (execute_todo): Likewise.
17358 * predict.c (drop_profile): Likewise.
17359 * symtab.c (symtab_node::get_dump_name): New function.
17360 (symtab_node::dump_name): Likewise.
17361 (symtab_node::dump_asm_name): Likewise.
17362 (symtab_node::dump_references): Likewise.
17363 (symtab_node::dump_referring): Likewise.
17364 (symtab_node::dump_base): Likewise.
17365 (symtab_node::debug_symtab): Likewise.
17366 * tree-sra.c (convert_callers_for_node): Likewise.
17367 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17368 * value-prof.c (init_node_map): Likewise.
17370 2017-05-23 Martin Liska <mliska@suse.cz>
17372 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17373 and symtab_node::debug_symtab to symbol_table::debug.
17374 * cgraphunit.c (analyze_functions): Use the renamed function.
17375 (symbol_table::compile): Likewise.
17376 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17377 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17378 * passes.c (execute_todo): Likewise.
17379 * symtab.c (symbol_table::dump): New function.
17380 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17382 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17384 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17385 that nonconst implies exec.
17387 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17389 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17390 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17391 (inline_edge_summary_vec): Turn into ...
17392 (ipa_call_summaries): ... this one.
17393 (redirect_to_unreachable, edge_set_predicate,
17394 evaluate_properties_for_edge, inline_summary_alloc,
17395 reset_ipa_call_summary, reset_inline_summary,
17396 inline_summary_t::duplicate): Update.
17397 (inline_edge_duplication_hook): Turn to ...
17398 (ipa_call_summary_t::duplicate): ... this one.
17399 (inline_edge_removal_hook): Turn to ...
17400 (ipa_call_summary_t::remove): ... this one.
17401 (dump_inline_edge_summary): Turn to ...
17402 (dump_ipa_call_summary): ... this one.
17403 (estimate_function_body_sizes): Update.
17404 (inline_update_callee_summaries): Update.
17405 (remap_edge_change_prob): Update.
17406 (remap_edge_summaries): Update.
17407 (inline_merge_summary): Update.
17408 (do_estimate_edge_time): Update.
17409 (inline_generate_summary): Update.
17410 (inline_read_section): Update.
17411 (inline_read_summary): Update.
17412 (inline_free_summary): Update.
17413 * ipa-inline.c (can_inline_edge_p): Update.
17414 (compute_inlined_call_time): Update.
17415 (want_inline_small_function_p): Update.
17416 (edge_badness): Update.
17417 (early_inliner): Update.
17418 * ipa-inline.h (inline_edge_summary): Turn to ...
17419 (ipa_call_summary): ... this one.
17420 (ipa_call_summary_t): New class.
17421 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17422 (ipa_call_summaries): New.
17423 (inline_edge_summary): Remove.
17424 (estimate_edge_growth): Update.
17425 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17426 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17427 * ipa-split.c (execute_split_functions): Update.
17428 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17430 2017-05-23 Tom de Vries <tom@codesourcery.com>
17432 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17433 attributes): Document rdrand effective target.
17435 2017-05-23 Tom de Vries <tom@codesourcery.com>
17437 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17438 attributes): Sort alphabetically.
17440 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17442 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17444 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17447 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17448 V2DF/V2DI splat into two separate patterns, one that handles
17449 registers, and the other that only handles memory. Drop support
17450 for splatting from a GPR on ISA 2.07 and then splitting the
17451 splat into direct move and splat.
17452 (vsx_splat_<mode>_reg): Likewise.
17453 (vsx_splat_<mode>_mem): Likewise.
17455 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17457 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17459 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17461 PR middle-end/80809
17462 * omp-low.c (finish_taskreg_remap): New function.
17463 (finish_taskreg_scan): If unit size of ctx->record_type
17464 is non-constant, unshare the size expression and replace
17465 decls in it with possible outer var refs.
17467 PR middle-end/80809
17468 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17469 GOVD_SHARED rather than GOVD_PRIVATE with it.
17470 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17471 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17473 PR middle-end/80853
17474 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17475 as last argument to build_outer_var_ref for pointer bases of array
17476 section reductions.
17478 2017-05-19 Martin Sebor <msebor@redhat.com>
17480 * print-tree.c (print_node): Print DECL_READ_P flag.
17482 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17484 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17485 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17486 * cgraph.c: Likewise.
17487 * cgraphunit.c: Likewise.
17488 * gengtype.c: Likewise.
17489 * ipa-cp.c: Likewise.
17490 * ipa-devirt.c: Likewise.
17491 * ipa-icf.c: Likewise.
17492 * ipa-predicate.c: Likewise.
17493 * ipa-profile.c: Likewise.
17494 * ipa-prop.c: Likewise.
17495 * ipa-split.c: Likewise.
17497 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17498 edge_predicate_pool, dump_inline_hints,
17499 inline_summary::account_size_time, redirect_to_unreachable,
17500 edge_set_predicate, set_hint_predicate,
17501 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17502 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17503 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17504 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17505 ipa_call_summary_t::remove, initialize_growth_caches,
17506 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17507 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17508 mark_modified, unmodified_parm_1, unmodified_parm,
17509 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17510 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17511 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17512 will_be_nonconstant_predicate, record_modified_bb_info,
17513 get_minimal_bb, record_modified, param_change_prob,
17514 phi_result_unknown_predicate, predicate_for_phi_result,
17515 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17516 estimate_function_body_sizes, compute_inline_parameters,
17517 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17518 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17519 inline_update_callee_summaries, remap_edge_change_prob,
17520 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17521 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17522 inline_analyze_function, inline_summary_t::insert,
17523 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17524 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17525 inline_free_summary): Move to ipa-fnsummary.h
17526 (predicate_t): Remove.
17527 * ipa-fnsummary.c: New file.
17528 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17529 (enum inline_hints_vals, inline_hints, agg_position_info,
17530 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17531 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17532 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17533 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17534 inline_read_summary, inline_write_summary, inline_free_summary,
17535 inline_analyze_function, initialize_inline_failed,
17536 inline_merge_summary, inline_update_overall_summary,
17537 compute_inline_parameters): Move to ipa-fnsummary.h
17538 * ipa-fnsummary.h: New file.
17539 * ipa-inline-transform.h: Include ipa-inline.h.
17540 * ipa-inline.c: LIkewise.
17542 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17544 * ipa-inline.c (edge_badness): Use inlined_time instead of
17545 inline_summaries->get.
17547 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17549 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17551 2017-05-22 Nathan Sidwell <nathan@acm.org>
17553 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17554 (fdump-lang): Document 'raw' option.
17555 * dumpfile.h (TDI_tu): Delete.
17556 * dumpfile.c (dump_files): Remove translation-unit.
17557 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17559 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17561 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17562 command option from $(AWK) call.
17563 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17564 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17565 [FORMAT]: Remove handling of variable.
17566 * config/avr/t-multilib: Regenerate.
17568 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17570 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17572 (dump_inline_summary): Do not print self_time.
17573 (estimate_function_body_sizes): Do not set self_time.
17574 (compute_inline_parameters): Likewise.
17575 (inline_read_section, inline_write_summary): Do not stream self_time.
17576 * ipa-inline.h (inline_summary): Drop self_time.
17578 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17580 * ipa-inline-analysis.c (account_size_time): Rename to ...
17581 (inline_summary::account_size_time): ... this one.
17582 (reset_ipa_call_summary): Turn to ...
17583 (ipa_call_summary::reset): ... this one.
17584 (reset_inline_summary): Turn to ...
17585 (inline_summary::reset): ... this one.
17586 (inline_summary_t::remove): Update.
17587 (inline_summary_t::duplicate): Update.
17588 (ipa_call_summary_t::remove): Update.
17589 (dump_inline_summary): Update.
17590 (estimate_function_body_sizes): Update.
17591 (compute_inline_parameters): Update.
17592 (estimate_node_size_and_time): Update.
17593 (inline_merge_summary): Update.
17594 (inline_update_overall_summary): Update.
17595 (inline_read_section): Update.
17596 (inline_write_summary): Update.
17597 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17598 add account_size_time and reset member functions.
17599 (ipa_call_summary): Add reset function.
17600 * ipa-predicate.h (predicate::operator &): Constify.
17602 2017-05-22 Richard Biener <rguenther@suse.de>
17604 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17606 2017-05-19 Jason Merrill <jason@redhat.com>
17608 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17610 2017-05-19 Marek Polacek <polacek@redhat.com>
17613 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17614 TYPE_OVERFLOW_WRAPS checks.
17616 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17618 * tree-core.h (enum omp_clause_default_kind): Add
17619 "OMP_CLAUSE_DEFAULT_PRESENT".
17620 * tree-pretty-print.c (dump_omp_clause): Handle it.
17621 * gimplify.c (enum gimplify_omp_var_data): Add
17622 "GOVD_MAP_FORCE_PRESENT".
17623 (gimplify_adjust_omp_clauses_1): Map it to
17624 "GOMP_MAP_FORCE_PRESENT".
17625 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17627 * gimplify.c (oacc_default_clause): Clarify.
17629 2017-05-19 Nathan Sidwell <nathan@acm.org>
17631 LANG_HOOK_REGISTER_DUMPS
17632 * toplev.c (general_init): Call register dump lang hook.
17633 * doc/invoke.texi: Document -fdump-lang option family.
17634 * dumpfile.c (dump_files): Remove class dump here.
17635 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17636 * dumpfile.h (tree_dump_index): Remove TDI_class.
17637 * langhooks-def.h (lhd_register_dumps): Declare.
17638 (LANG_HOOKS_REGISTER_DUMPS): Define.
17639 (LANG_HOOKS_INITIALIZER): Add it.
17640 * langhooks.c (lhd_register_dumps): Define.
17641 * langhooks.h (struct lang_hooks): Add register_dumps.
17643 2017-05-19 Nathan Sidwell <nathan@acm.org>
17645 * context.h (context::set_passes): New.
17646 * context.c (context::context): Do not create pass manager.
17647 * toplev.c (general_init): Create pass manager here.
17649 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17651 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17652 use this splitter if two add or or instructions would also work for
17653 the constant we want to generate.
17655 2017-05-19 Richard Biener <rguenther@suse.de>
17658 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17659 predicate evaluation.
17661 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17663 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17665 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17668 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17670 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17671 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17672 (inline_edge_summary_vec): Turn into ...
17673 (ipa_call_summaries): ... this one.
17674 (redirect_to_unreachable, edge_set_predicate,
17675 evaluate_properties_for_edge, inline_summary_alloc,
17676 reset_ipa_call_summary, reset_inline_summary,
17677 inline_summary_t::duplicate): Update.
17678 (inline_edge_duplication_hook): Turn to ...
17679 (ipa_call_summary_t::duplicate): ... this one.
17680 (inline_edge_removal_hook): Turn to ...
17681 (ipa_call_summary_t::remove): ... this one.
17682 (dump_inline_edge_summary): Turn to ...
17683 (dump_ipa_call_summary): ... this one.
17684 (estimate_function_body_sizes): Update.
17685 (inline_update_callee_summaries): Update.
17686 (remap_edge_change_prob): Update.
17687 (remap_edge_summaries): Update.
17688 (inline_merge_summary): Update.
17689 (do_estimate_edge_time): Update.
17690 (inline_generate_summary): Update.
17691 (inline_read_section): Update.
17692 (inline_read_summary): Update.
17693 (inline_free_summary): Update.
17694 * ipa-inline.c (can_inline_edge_p): Update.
17695 (compute_inlined_call_time): Update.
17696 (want_inline_small_function_p): Update.
17697 (edge_badness): Update.
17698 (early_inliner): Update.
17699 * ipa-inline.h (inline_edge_summary): Turn to ...
17700 (ipa_call_summary): ... this one.
17701 (ipa_call_summary_t): New class.
17702 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17703 (ipa_call_summaries): New.
17704 (inline_edge_summary): Remove.
17705 (estimate_edge_growth): Update.
17706 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17707 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17708 * ipa-split.c (execute_split_functions): Update.
17709 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17711 2017-05-19 Richard Biener <rguenther@suse.de>
17713 PR middle-end/80764
17714 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17716 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17718 * config/rs6000/rs6000.c (struct machine_function): Add field
17719 fpr_is_wrapped_separately.
17720 (rs6000_get_separate_components): Use 64 components. Handle the
17721 new FPR components.
17722 (rs6000_components_for_bb): Handle the FPR components.
17723 (rs6000_emit_prologue_components): Handle the FPR components.
17724 (rs6000_emit_epilogue_components): Handle the FPR components.
17725 (rs6000_set_handled_components): Handle the FPR components.
17726 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17727 that are already separately shrink-wrapped.
17728 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17729 that are already separately shrink-wrapped.
17731 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17734 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17737 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17738 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17739 cases where the register allocator uses a move and an offsettable
17740 memory operation to/from a FPR register on ISA 2.06/2.07.
17741 (define_peephole2 for Altivec d-form store): Likewise.
17743 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17746 * config/i386/mmx.md (*mov<mode>_internal): Enable
17747 alternatives 11, 12, 13 and 14 also for 32bit targets.
17748 Remove alternatives 15, 16, 17 and 18.
17749 * config/i386/sse.md (vec_concatv2di): Change
17750 alternative (!x, *y) to (x, ?!*Yn).
17752 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17754 * dumpfile.h (enum dump_kind): Remove stray comma.
17756 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17758 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17759 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17760 predicate::num_conditions
17761 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17762 (CHANGED): turn into predicate::changed.
17763 (agg_position_info): Move to ipa-predicate.h
17764 (add_condition, predicate::add_clause, predicate::operator &=,
17765 predicate::or_with, predicate::evaluate, predicate::probability,
17766 dump_condition, dump_clause, predicate::dump,
17767 predicate::remap_after_duplication, predicate::remap_after_inlining,
17768 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17769 (evaluate_conditions_for_known_args): Update.
17770 (set_cond_stmt_execution_predicate): Update.
17771 * ipa-inline.h: Include ipa-predicate.h
17772 (condition, inline_param_summary, conditions, agg_position_info,
17773 predicate): Move to ipa-predicate.h
17774 * ipa-predicate.c: New file.
17775 * ipa-predicate.h: New file.
17777 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17779 * final.c (leaf_function_p): Check we are not in a sequence.
17781 2017-05-18 Martin Liska <mliska@suse.cz>
17783 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17784 * dumpfile.c (dump_register): Use new enum dump_kind.
17785 (get_dump_file_name): Likewise.
17786 (dump_enable_all): Likewise.
17787 (dump_switch_p_1): Likewise.
17788 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17789 * dumpfile.h (enum dump_kind): New enum type.
17790 (struct dump_file_info): Create constructor and
17791 format fields and comments.
17792 * passes.c (pass_manager::register_one_dump_file):
17794 * statistics.c (statistics_early_init): Likewise.
17795 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17796 TDF_TREE with TDF_SLIM.
17797 (gather_memory_references_ref): Likewise.
17799 2017-05-18 Martin Liska <mliska@suse.cz>
17801 * vec.h (struct vnull): Use it.
17803 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17805 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17806 (true_predicate, false_predicate, true_predicate_p,
17807 false_predicate_p): Remove.
17808 (single_cond_predicate, not_inlined_predicate): Turn to member function
17810 (add_condition): Update.
17811 (add_clause): Turn to...
17812 (predicate::add_clause): ... this one; update; allow passing NULL
17814 (and_predicates): Turn to ...
17815 (predicate::operator &=): ... this one.
17816 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17817 (or_predicates): Turn to ...
17818 (predicate::or_with): ... this one.
17819 (evaluate_predicate): Turn to ...
17820 (predicate::evaluate): ... this one.
17821 (predicate_probability): Turn to ...
17822 (predicate::probability): ... this one.
17823 (dump_condition): Update.
17824 (dump_predicate): Turn to ...
17825 (predicate::dump): ... this one.
17826 (account_size_time): Update.
17827 (edge_set_predicate): Update.
17828 (set_hint_predicate): UPdate.
17829 (evaluate_conditions_for_known_args): Update.
17830 (evaluate_properties_for_edge): Update.
17831 (remap_predicate_after_duplication): Turn to...
17832 (predicate::remap_after_duplication): ... this one.
17833 (remap_hint_predicate_after_duplication): Update.
17834 (inline_summary_t::duplicate): UPdate.
17835 (dump_inline_edge_summary): Update.
17836 (dump_inline_summary): Update.
17837 (set_cond_stmt_execution_predicate): Update.
17838 (set_switch_stmt_execution_predicate): Update.
17839 (compute_bb_predicates): Update.
17840 (will_be_nonconstant_expr_predicate): Update.
17841 (will_be_nonconstant_predicate): Update.
17842 (phi_result_unknown_predicate): Update.
17843 (predicate_for_phi_result): Update.
17844 (array_index_predicate): Update.
17845 (estimate_function_body_sizes): Update.
17846 (estimate_node_size_and_time): Update.
17847 (estimate_ipcp_clone_size_and_time): Update.
17848 (remap_predicate): Rename to ...
17849 (predicate::remap_after_inlining): ... this one.
17850 (remap_hint_predicate): Update.
17851 (inline_merge_summary): Update.
17852 (inline_update_overall_summary): Update.
17853 (estimate_size_after_inlining): Update.
17854 (read_predicate): Rename to ...
17855 (predicate::stream_in): ... this one.
17856 (read_inline_edge_summary): Update.
17857 (write_predicate): Rename to ...
17858 (predicate::stream_out): ... this one.
17859 (write_inline_edge_summary): Update.
17860 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17861 (clause_t): Turn to uint32_t
17862 (predicate): Turn to class; implement constructor and operators
17864 (size_time_entry): Update.
17865 (inline_summary): Update.
17866 (inline_edge_summary): Update.
17868 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17870 * fold-const.c (fold_binary_loc): Move transformation...
17871 * match.pd (C - X CMP X): ... here.
17873 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17875 * config/sparc/sparc.c (sparc_option_override): Set function
17876 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17877 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17879 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17881 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17883 2017-05-18 Marek Polacek <polacek@redhat.com>
17886 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17887 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17888 gimple_assign_load_p.
17890 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17892 PR middle-end/80692
17893 * real.c (do_compare): Give decimal_do_compare preference over
17894 comparing just the signs.
17896 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17898 * doc/md.texi (Canonicalization of Instructions): Describe the
17899 canonical form of instructions that inherently set a condition
17902 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17904 PR middle-end/80775
17905 * tree-cfg.c: Move deletion of unreachable case statements to after
17906 the merging of consecutive case labels.
17908 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17910 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17911 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17912 restoring of callee-saved registers.
17914 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17916 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17917 * config/visium/visium.c (single_set_and_flags): Likewise.
17918 * config/visium/visium.md (Substitutions): Likewise.
17920 2017-05-17 Martin Liska <mliska@suse.cz>
17922 * cfg.c: Introduce dump_flags_t type and
17923 use it instead of int type.
17925 * cfghooks.c: Likewise.
17926 * cfghooks.h (struct cfg_hooks): Likewise.
17927 * cfgrtl.c: Likewise.
17928 * cfgrtl.h: Likewise.
17929 * cgraph.c (cgraph_node::get_body): Likewise.
17930 * coretypes.h: Likewise.
17931 * domwalk.c: Likewise.
17932 * domwalk.h: Likewise.
17933 * dumpfile.c (struct dump_option_value_info): Likewise.
17934 (dump_enable_all): Likewise.
17935 (dump_switch_p_1): Likewise.
17936 (opt_info_switch_p): Likewise.
17937 * dumpfile.h (enum tree_dump_index): Likewise.
17938 (struct dump_file_info): Likewise.
17939 * genemit.c: Likewise.
17940 * generic-match-head.c: Likewise.
17941 * gengtype.c (open_base_files): Likewise.
17942 * gimple-pretty-print.c: Likewise.
17943 * gimple-pretty-print.h: Likewise.
17944 * graph.c (print_graph_cfg): Likewise.
17945 * graphite-scop-detection.c (dot_all_sese): Likewise.
17946 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17947 * loop-unroll.c (report_unroll): Likewise.
17948 * passes.c (pass_manager::register_one_dump_file): Likewise.
17949 * print-tree.c: Likewise.
17950 * statistics.c: Likewise.
17951 * tree-cfg.c: Likewise.
17952 * tree-cfg.h: Likewise.
17953 * tree-dfa.c: Likewise.
17954 * tree-dfa.h: Likewise.
17955 * tree-dump.c (dump_function): Likewise.
17956 * tree-dump.h (struct dump_info): Likewise.
17957 * tree-pretty-print.c: Likewise.
17958 * tree-pretty-print.h: Likewise.
17959 * tree-ssa-live.c: Likewise.
17960 * tree-ssa-live.h: Likewise.
17961 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17962 * tree-vect-loop.c: Likewise.
17963 * tree-vect-slp.c: Likewise.
17965 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17966 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17968 PR tree-optimization/80457
17969 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17970 of all arguments to a statement as scalar_to_vec operations.
17971 (vectorizable_call): Adjust call to vect_model_simple_cost for
17973 (vectorizable_conversion): Likewise.
17974 (vectorizable_assignment): Likewise.
17975 (vectorizable_shift): Likewise.
17976 (vectorizable_operation): Likewise.
17977 (vectorizable_comparison): Likewise.
17978 (vect_is_simple_cond): Record the def types for operands.
17979 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17980 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17981 for statement argument count.
17983 2017-05-16 Carl Love <cel@us.ibm.com>
17985 * config/rs6000/rs6000-c: Add support for built-in functions
17986 vector unsigned long long vec_bperm (vector unsigned long long,
17987 vector unsigned char)
17988 vector signed long long vec_mule (vector signed int,
17990 vector unsigned long long vec_mule (vector unsigned int,
17991 vector unsigned int)
17992 vector signed long long vec_mulo (vector signed int,
17994 vector unsigned long long vec_mulo (vector unsigned int,
17995 vector unsigned int)
17996 vector signed char vec_sldw (vector signed char,
17997 vector signed char,
17999 vector unsigned char vec_sldw (vector unsigned char,
18000 vector unsigned char,
18002 vector signed short vec_sldw (vector signed short,
18003 vector signed short,
18005 vector unsigned short vec_sldw (vector unsigned short,
18006 vector unsigned short,
18008 vector signed int vec_sldw (vector signed int,
18011 vector unsigned int vec_sldw (vector unsigned int,
18012 vector unsigned int,
18014 vector signed long long vec_sldw (vector signed long long,
18015 vector signed long long,
18017 vector unsigned long long vec_sldw (vector unsigned long long,
18018 vector unsigned long long,
18020 * config/rs6000/rs6000-c: Add support for built-in functions
18021 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18022 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18023 * doc/extend.texi: Update the built-in documentation for the
18024 new built-in functions.
18026 2017-05-16 Marek Polacek <polacek@redhat.com>
18030 * tree.c (save_expr): Don't fold the expression.
18032 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18034 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18035 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18036 and (?*y,m). Update insn attributes.
18038 2017-05-16 Martin Liska <mliska@suse.cz>
18040 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18041 flags argument of print_gimple_stmt, print_gimple_expr,
18042 print_generic_stmt and print_generic_expr.
18043 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18044 * coretypes.h: Likewise.
18045 * except.c (dump_eh_tree): Likewise.
18046 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18047 * gimple-pretty-print.h: Likewise.
18048 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18049 (backprop::push_to_worklist): Likewise.
18050 (backprop::pop_from_worklist): Likewise.
18051 (backprop::process_use): Likewise.
18052 (backprop::intersect_uses): Likewise.
18053 (note_replacement): Likewise.
18054 * gimple-ssa-store-merging.c
18055 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18056 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18057 (pass_store_merging::execute): Likewise.
18058 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18059 (ssa_base_cand_dump_callback): Likewise.
18060 (dump_incr_vec): Likewise.
18061 (replace_refs): Likewise.
18062 (replace_mult_candidate): Likewise.
18063 (create_add_on_incoming_edge): Likewise.
18064 (create_phi_basis): Likewise.
18065 (insert_initializers): Likewise.
18066 (all_phi_incrs_profitable): Likewise.
18067 (introduce_cast_before_cand): Likewise.
18068 (replace_one_candidate): Likewise.
18069 * gimplify.c (gimplify_expr): Likewise.
18070 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18071 (set_rename): Likewise.
18072 (rename_uses): Likewise.
18073 (copy_loop_phi_nodes): Likewise.
18074 (add_close_phis_to_merge_points): Likewise.
18075 (copy_loop_close_phi_args): Likewise.
18076 (copy_cond_phi_args): Likewise.
18077 (graphite_copy_stmts_from_block): Likewise.
18078 (translate_pending_phi_nodes): Likewise.
18079 * graphite-poly.c (print_pdr): Likewise.
18080 (dump_gbb_cases): Likewise.
18081 (dump_gbb_conditions): Likewise.
18082 (print_scop_params): Likewise.
18083 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18084 (build_cross_bb_scalars_use): Likewise.
18085 (gather_bbs::before_dom_children): Likewise.
18086 * hsa-dump.c (dump_hsa_immed): Likewise.
18087 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18088 (get_replacement_map): Likewise.
18089 * ipa-inline-analysis.c (dump_condition): Likewise.
18090 (estimate_function_body_sizes): Likewise.
18091 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18092 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18093 * ipa-prop.c (ipa_dump_param): Likewise.
18094 (ipa_print_node_jump_functions_for_edge): Likewise.
18095 (ipa_modify_call_arguments): Likewise.
18096 (ipa_modify_expr): Likewise.
18097 (ipa_dump_param_adjustments): Likewise.
18098 (ipa_dump_agg_replacement_values): Likewise.
18099 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18100 * ipa-pure-const.c (check_stmt): Likewise.
18101 (pass_nothrow::execute): Likewise.
18102 * ipa-split.c (execute_split_functions): Likewise.
18103 * omp-offload.c (dump_oacc_loop_part): Likewise.
18104 (dump_oacc_loop): Likewise.
18105 * trans-mem.c (tm_log_emit): Likewise.
18106 (tm_memopt_accumulate_memops): Likewise.
18107 (dump_tm_memopt_set): Likewise.
18108 (dump_tm_memopt_transform): Likewise.
18109 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18110 (print_loop): Likewise.
18111 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18112 (chkp_gather_checks_info): Likewise.
18113 (chkp_get_check_result): Likewise.
18114 (chkp_remove_check_if_pass): Likewise.
18115 (chkp_use_outer_bounds_if_possible): Likewise.
18116 (chkp_reduce_bounds_lifetime): Likewise.
18117 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18118 (chkp_mark_completed_bounds): Likewise.
18119 (chkp_register_incomplete_bounds): Likewise.
18120 (chkp_mark_invalid_bounds): Likewise.
18121 (chkp_maybe_copy_and_register_bounds): Likewise.
18122 (chkp_build_returned_bound): Likewise.
18123 (chkp_get_bound_for_parm): Likewise.
18124 (chkp_build_bndldx): Likewise.
18125 (chkp_get_bounds_by_definition): Likewise.
18126 (chkp_generate_extern_var_bounds): Likewise.
18127 (chkp_get_bounds_for_decl_addr): Likewise.
18128 * tree-chrec.c (chrec_apply): Likewise.
18129 * tree-data-ref.c (dump_data_reference): Likewise.
18130 (dump_subscript): Likewise.
18131 (dump_data_dependence_relation): Likewise.
18132 (analyze_overlapping_iterations): Likewise.
18133 * tree-inline.c (expand_call_inline): Likewise.
18134 (tree_function_versioning): Likewise.
18135 * tree-into-ssa.c (dump_defs_stack): Likewise.
18136 (dump_currdefs): Likewise.
18137 (dump_names_replaced_by): Likewise.
18138 (dump_update_ssa): Likewise.
18139 (update_ssa): Likewise.
18140 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18141 * tree-parloops.c (build_new_reduction): Likewise.
18142 (try_create_reduction_list): Likewise.
18143 (ref_conflicts_with_region): Likewise.
18144 (oacc_entry_exit_ok_1): Likewise.
18145 (oacc_entry_exit_single_gang): Likewise.
18146 * tree-pretty-print.h: Likewise.
18147 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18148 (get_scalar_evolution): Likewise.
18149 (add_to_evolution): Likewise.
18150 (get_loop_exit_condition): Likewise.
18151 (analyze_evolution_in_loop): Likewise.
18152 (analyze_initial_condition): Likewise.
18153 (analyze_scalar_evolution): Likewise.
18154 (instantiate_scev): Likewise.
18155 (number_of_latch_executions): Likewise.
18156 (gather_chrec_stats): Likewise.
18157 (final_value_replacement_loop): Likewise.
18158 (scev_const_prop): Likewise.
18159 * tree-sra.c (dump_access): Likewise.
18160 (disqualify_candidate): Likewise.
18161 (create_access): Likewise.
18162 (reject): Likewise.
18163 (maybe_add_sra_candidate): Likewise.
18164 (create_access_replacement): Likewise.
18165 (analyze_access_subtree): Likewise.
18166 (analyze_all_variable_accesses): Likewise.
18167 (sra_modify_assign): Likewise.
18168 (initialize_constant_pool_replacements): Likewise.
18169 (find_param_candidates): Likewise.
18170 (decide_one_param_reduction): Likewise.
18171 (replace_removed_params_ssa_names): Likewise.
18172 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18173 * tree-ssa-copy.c (dump_copy_of): Likewise.
18174 (copy_prop_visit_cond_stmt): Likewise.
18175 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18176 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18177 (record_equivalences_from_stmt): Likewise.
18178 * tree-ssa-dse.c (compute_trims): Likewise.
18179 (delete_dead_call): Likewise.
18180 (delete_dead_assignment): Likewise.
18181 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18182 (forward_propagate_into_cond): Likewise.
18183 (pass_forwprop::execute): Likewise.
18184 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18185 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18187 (move_computations_worker): Likewise.
18188 (execute_sm): Likewise.
18189 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18190 (remove_exits_and_undefined_stmts): Likewise.
18191 (remove_redundant_iv_tests): Likewise.
18192 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18193 (adjust_iv_update_pos): Likewise.
18194 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18195 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18196 (value_replacement): Likewise.
18197 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18198 * tree-ssa-pre.c (print_pre_expr): Likewise.
18199 (get_representative_for): Likewise.
18200 (create_expression_by_pieces): Likewise.
18201 (insert_into_preds_of_block): Likewise.
18202 (eliminate_insert): Likewise.
18203 (eliminate_dom_walker::before_dom_children): Likewise.
18204 (eliminate): Likewise.
18205 (remove_dead_inserted_code): Likewise.
18206 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18207 * tree-ssa-reassoc.c (get_rank): Likewise.
18208 (eliminate_duplicate_pair): Likewise.
18209 (eliminate_plus_minus_pair): Likewise.
18210 (eliminate_not_pairs): Likewise.
18211 (undistribute_ops_list): Likewise.
18212 (eliminate_redundant_comparison): Likewise.
18213 (update_range_test): Likewise.
18214 (optimize_range_tests_var_bound): Likewise.
18215 (optimize_vec_cond_expr): Likewise.
18216 (rewrite_expr_tree): Likewise.
18217 (rewrite_expr_tree_parallel): Likewise.
18218 (linearize_expr): Likewise.
18219 (break_up_subtract): Likewise.
18220 (linearize_expr_tree): Likewise.
18221 (attempt_builtin_powi): Likewise.
18222 (attempt_builtin_copysign): Likewise.
18223 (transform_stmt_to_copy): Likewise.
18224 (transform_stmt_to_multiply): Likewise.
18225 (dump_ops_vector): Likewise.
18226 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18227 (print_scc): Likewise.
18228 (set_ssa_val_to): Likewise.
18229 (visit_reference_op_store): Likewise.
18230 (visit_use): Likewise.
18231 (sccvn_dom_walker::before_dom_children): Likewise.
18232 (run_scc_vn): Likewise.
18233 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18235 (expr_hash_elt::print): Likewise.
18236 (const_and_copies::pop_to_marker): Likewise.
18237 (const_and_copies::record_const_or_copy_raw): Likewise.
18238 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18239 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18240 (dump_predicates): Likewise.
18241 (find_uninit_use): Likewise.
18242 (warn_uninitialized_phi): Likewise.
18243 (pass_late_warn_uninitialized::execute): Likewise.
18244 * tree-ssa.c (verify_vssa): Likewise.
18245 (verify_ssa): Likewise.
18246 (maybe_optimize_var): Likewise.
18247 * tree-vrp.c (dump_value_range): Likewise.
18248 (dump_all_value_ranges): Likewise.
18249 (dump_asserts_for): Likewise.
18250 (register_edge_assert_for_2): Likewise.
18251 (vrp_visit_cond_stmt): Likewise.
18252 (vrp_visit_switch_stmt): Likewise.
18253 (vrp_visit_stmt): Likewise.
18254 (vrp_visit_phi_node): Likewise.
18255 (simplify_cond_using_ranges_1): Likewise.
18256 (fold_predicate_in): Likewise.
18257 (evrp_dom_walker::before_dom_children): Likewise.
18258 (evrp_dom_walker::push_value_range): Likewise.
18259 (evrp_dom_walker::pop_value_range): Likewise.
18260 (execute_early_vrp): Likewise.
18262 2017-05-16 Richard Biener <rguenther@suse.de>
18264 * dwarf2out.c (loc_list_from_tree_1): Do not create
18265 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18267 2017-05-16 Richard Biener <rguenther@suse.de>
18269 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18271 (note_variable_value_in_expr): If we resolved the decl ref
18272 do not push to the stack.
18274 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18276 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18277 operations in fast-math mode.
18278 (vaddq_f16): Likewise.
18279 (vmul_f16): Likewise.
18280 (vmulq_f16): Likewise.
18281 (vsub_f16): Likewise.
18282 (vsubq_f16): Likewise.
18283 * config/arm/neon.md (add<mode>3): New.
18285 (fma:<VH:mode>3): New. Also remove outdated comment.
18288 2017-05-16 Martin Liska <mliska@suse.cz>
18292 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18293 (odr_types_equivalent_p): Likewise.
18295 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18297 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18299 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18302 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18303 non-interunit SSE move alternatives with '?'.
18304 (zero-extendsidi peephole2): New peephole to skip intermediate
18305 general register in SSE zero-extend sequence.
18307 2017-05-15 Jeff Law <law@redhat.com>
18309 * reorg.c (relax_delay_slots): Create a new variable to hold
18310 the temporary target rather than clobbering TARGET_LABEL.
18312 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18313 missing argument to extract_bit_field call.
18314 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18316 2017-05-15 Martin Liska <mliska@suse.cz>
18319 * gcc.c (process_command): Do not allow empty argument of -o option.
18321 2017-05-15 Renlin Li <renlin.li@arm.com>
18323 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18324 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18325 * config/aarch64/constraints.md (Usf): Add long call check.
18326 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18327 (call_value): Likewise.
18328 (sibcall): Likewise.
18329 (sibcall_value): Likewise.
18331 (call_value_insn): New.
18332 (sibcall_insn): Update rtx pattern.
18333 (sibcall_value_insn): Likewise.
18334 (call_internal): Remove.
18335 (call_value_internal): Likewise.
18336 (sibcall_internal): Likewise.
18337 (sibcall_value_internal): Likewise.
18338 (call_reg): Likewise.
18339 (call_symbol): Likewise.
18340 (call_value_reg): Likewise.
18341 (call_value_symbol): Likewise.
18343 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18346 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18348 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18350 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18351 compatible with CCGOCmode and with CCZmode.
18353 2017-05-14 Martin Sebor <msebor@redhat.com>
18355 PR middle-end/77671
18356 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18357 (gimple_fold_builtin_snprintf): Same.
18358 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18359 (gimple_fold_builtin_snprintf): Same.
18360 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18361 of character types.
18362 (is_call_safe): New function.
18363 (try_substitute_return_value): Call it.
18364 (try_simplify_call): New function.
18365 (pass_sprintf_length::handle_gimple_call): Call it.
18367 2017-05-14 Martin Sebor <msebor@redhat.com>
18369 PR middle-end/80669
18370 * builtins.c (expand_builtin_stpncpy): Simplify.
18372 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18374 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18375 * config/i386/i386.h
18376 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18377 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18378 (struct machine_function): Add new members call_ms2sysv,
18379 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18380 (struct machine_frame_state): New fields sp_realigned and
18381 sp_realigned_offset.
18382 * config/i386/i386.c
18383 (enum xlogue_stub): New enum.
18384 (enum xlogue_stub_sets): New enum.
18385 (class xlogue_layout): New class.
18386 (struct ix86_frame): New fields stack_realign_allocate_offset,
18387 stack_realign_offset and outlined_save_offset. Modify comments to
18388 detail stack layout when using out-of-line stubs.
18389 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18390 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18391 -mcall-ms2sysv-xlogues.
18392 (stub_managed_regs): New static variable.
18393 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18394 registers managed by out-of-line stub.
18395 (disable_call_ms2sysv_xlogues): New function.
18396 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18397 m->call_ms2sysv when appropriate and compute frame layout for
18399 (sp_valid_at, fp_valid_at): New inline functions.
18400 (choose_basereg): New function.
18401 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18403 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18404 Use align parameter of choose_baseaddr to generated aligned SSE movs
18406 (pro_epilogue_adjust_stack): Modify to track
18407 machine_frame_state::sp_realigned.
18408 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18409 (ix86_nsaved_sseregs): Likewise.
18410 (ix86_emit_save_regs): Likewise.
18411 (ix86_emit_save_regs_using_mov): Likewise.
18412 (ix86_emit_save_sse_regs_using_mov): Likewise.
18413 (get_scratch_register_on_entry): Likewise.
18414 (gen_frame_set): New function.
18415 (gen_frame_load): Likewise.
18416 (gen_frame_store): Likewise.
18417 (emit_outlined_ms2sysv_save): Likewise.
18418 (emit_outlined_ms2sysv_restore): Likewise.
18419 (ix86_expand_prologue): Modify stack re-alignment code and call
18420 emit_outlined_ms2sysv_save when appropriate.
18421 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18422 parameter rtx_insn *insn, which allows the function to be used to only
18423 generate the notes.
18424 (ix86_expand_epilogue): Modify validity checks of frame and stack
18425 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18426 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18427 * config/i386/predicates.md
18428 (save_multiple): New predicate.
18429 (restore_multiple): Likewise.
18430 * config/i386/sse.md
18431 (save_multiple<mode>): New pattern.
18432 (save_multiple_realign<mode>): Likewise.
18433 (restore_multiple<mode>): Likewise.
18434 (restore_multiple_and_return<mode>): Likewise.
18435 (restore_multiple_leave_return<mode>): Likewise.
18436 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18438 2017-05-14 Julia Koval <julia.koval@intel.com>
18440 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18441 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18442 (__builtin_ia32_xsetbv): New builtins.
18443 * config/i386/i386.c (ix86_expand_special_args_builtin):
18445 (ix86_expand_builtin): Special expand for new intrinsics.
18446 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18447 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18448 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18450 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18452 * cfganal.c (inverted_post_order_compute): Change argument type
18454 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18455 * df-core.c (rest_of_handle_df_initialize): Adjust.
18456 (rest_of_handle_df_finish): Likewise.
18457 (df_analyze_1): Likewise.
18458 (df_analyze): Likewise.
18459 (loop_inverted_post_order_compute): Change argument to be a vec *.
18460 (df_analyze_loop): Adjust.
18461 (df_get_n_blocks): Likewise.
18462 (df_get_postorder): Likewise.
18463 * df.h (struct df_d): Change field to be a vec.
18464 * lcm.c (compute_laterin): Adjust.
18465 (compute_available): Likewise.
18466 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18467 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18468 * tree-ssa-pre.c (compute_antic): Likewise.
18470 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18472 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18473 (depth_first_search::depth_first_search): Change structure init
18474 function to this constructor.
18475 (depth_first_search::add_bb): Rename function to this member.
18476 (depth_first_search::execute): Likewise.
18477 (flow_dfs_compute_reverse_finish): Adjust.
18479 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18481 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18482 (longest_simple_path): Likewise.
18483 * shrink-wrap.c (spread_components): Likewise.
18484 (disqualify_problematic_components): Likewise.
18485 (emit_common_heads_for_components): Likewise.
18486 (emit_common_tails_for_components): Likewise.
18487 (insert_prologue_epilogue_for_components): Likewise.
18489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18491 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18494 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18496 * df-core.c (df_set_blocks): Start using auto_bitmap.
18497 (df_compact_blocks): Likewise.
18498 * df-problems.c (df_rd_confluence_n): Likewise.
18499 * df-scan.c (df_insn_rescan_all): Likewise.
18500 (df_process_deferred_rescans): Likewise.
18501 (df_update_entry_block_defs): Likewise.
18502 (df_update_exit_block_uses): Likewise.
18503 (df_entry_block_bitmap_verify): Likewise.
18504 (df_exit_block_bitmap_verify): Likewise.
18505 (df_scan_verify): Likewise.
18506 * lra-constraints.c (lra_constraints): Likewise.
18507 (undo_optional_reloads): Likewise.
18508 (lra_undo_inheritance): Likewise.
18509 * lra-remat.c (calculate_gen_cands): Likewise.
18510 (do_remat): Likewise.
18511 * lra-spills.c (assign_spill_hard_regs): Likewise.
18512 (spill_pseudos): Likewise.
18513 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18514 (bitmap_set_subtract_values): Likewise.
18516 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18518 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18519 management with auto_bitmap.
18520 (fix_inter_tick): Likewise.
18521 (fix_recovery_deps): Likewise.
18522 * ira.c (add_store_equivs): Likewise.
18523 (find_moveable_pseudos): Likewise.
18524 (split_live_ranges_for_shrink_wrap): Likewise.
18525 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18526 (rtx_reuse_manager::seen_def_p): Likewise.
18527 (rtx_reuse_manager::set_seen_def): Likewise.
18528 * print-rtl.h (class rtx_reuse_manager): Likewise.
18530 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18532 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18534 (migrate_btr_def): Likewise.
18535 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18536 * df-core.c (loop_post_order_compute): Likewise.
18537 (loop_inverted_post_order_compute): Likewise.
18538 * hsa-common.h: Likewise.
18539 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18540 * init-regs.c (initialize_uninitialized_regs): Likewise.
18541 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18542 (inline_small_functions): Likewise.
18543 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18544 * ira.c (combine_and_move_insns): Likewise.
18545 (build_insn_chain): Likewise.
18546 * loop-invariant.c (find_invariants): Likewise.
18547 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18548 * predict.c (tree_predict_by_opcode): Likewise.
18549 (predict_paths_leading_to): Likewise.
18550 (predict_paths_leading_to_edge): Likewise.
18551 (estimate_loops_at_level): Likewise.
18552 (estimate_loops): Likewise.
18553 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18554 (spread_components): Likewise.
18555 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18556 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18557 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18558 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18559 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18560 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18561 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18562 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18563 (mark_threaded_blocks): Likewise.
18564 (thread_through_all_blocks): Likewise.
18565 * tree-ssa.c (verify_ssa): Likewise.
18566 (execute_update_addresses_taken): Likewise.
18567 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18569 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18571 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18573 (post_order_compute): Likewise.
18574 (inverted_post_order_compute): Likewise.
18575 (pre_and_rev_post_order_compute_fn): Likewise.
18577 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18579 * genrecog.c (int_set::int_set): Explicitly construct our
18580 auto_vec base class.
18581 * vec.h (auto_vec::auto_vec): New constructor.
18583 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18585 * bitmap.h (class auto_bitmap): New constructor taking
18586 bitmap_obstack * argument.
18588 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18590 * bitmap.h (class auto_bitmap): Change type of m_bits to
18591 bitmap_head, and adjust ctor / dtor and member operators.
18593 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18595 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18596 when returned register mode doesn't match original mode.
18598 2017-05-12 Jeff Law <law@redhat.com>
18599 Jakub Jelinek <jakub@redhat.com>
18601 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18602 we look for cc setter after the compare-elim changes.
18603 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18604 within the vector to match what compare-elim now expects.
18605 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18606 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18608 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18609 after the compare-elim changes.
18610 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18611 the vector to match what compare-elim now expects.
18612 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18613 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18614 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18615 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18616 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18618 * config/visium/visium.c (single_set_and_flags): Fix where
18619 we look for cc setter after the compare-elim changes.
18620 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18621 with the vector to match what compare-elim now expects.
18622 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18623 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18624 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18625 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18626 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18627 (neg<mode>2_insn_set_overflow): Likewise.
18629 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18631 PR middle-end/79794
18632 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18633 maybe_expand_insn call, set ops[0].target. If still set after call,
18634 set alt_rtl. Add extra arg to recursive calls.
18635 (extract_bit_field): Add alt_rtl argument. Pass to
18637 * expmed.h (extract_bit_field): Fix prototype.
18638 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18639 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18640 to extract_bit_field_calls.
18641 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18642 Pass alt_rtl to extract_bit_field calls.
18643 * calls.c (store_unaligned_arguments_into_psuedos)
18644 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18645 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18647 * optabs.h (struct expand_operand): Add target bitfield.
18649 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18651 * compare-elim.c (try_eliminate_compare): Canonicalize
18652 operation with embedded compare to
18653 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18654 (set (reg) (operation)].
18656 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18658 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18661 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18662 cost of adding a carry flag for ADC instruction.
18663 [case MINUS]: Ignore the cost of subtracting a carry flag
18664 for SBB instruction.
18666 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18668 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18670 * config/rs6000/bmiintrin.h: New file.
18671 * config/rs6000/bmi2intrin.h: New file.
18672 * config/rs6000/x86intrin.h: New file.
18674 2017-05-12 Jeff Law <law@redhat.com>
18676 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18679 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18681 PR middle-end/80707
18682 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18684 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18686 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18687 early expansion of vector divide builtins.
18688 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18689 builtins identified as having unsigned arguments.
18691 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18693 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18694 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18695 expansion of vector logical operations (and, andc, or, xor,
18698 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18700 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18701 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18703 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18705 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18706 early GIMPLE expansion of vector multiplies.
18708 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18710 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18711 TARGET_HAVE_MOVT conditional.
18712 (movt splitter): Likewise.
18714 2017-05-12 Richard Biener <rguenther@suse.de>
18716 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18717 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18718 Fold all stmts not inplace.
18720 2017-05-12 Richard Biener <rguenther@suse.de>
18722 PR tree-optimization/80713
18723 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18724 inserted_exprs bit for not removed stmts.
18726 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18728 PR middle-end/69921
18729 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18730 parallelized" attribute for parallelized OpenACC kernels.
18731 * omp-offload.c (execute_oacc_device_lower): Use it.
18733 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18734 Set "oacc kernels" attribute.
18735 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18736 parameter. Adjust all users.
18737 (oacc_fn_attrib_kernels_p): Remove function.
18738 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18739 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18740 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18741 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18742 assert "oacc kernels" attribute is set.
18744 2017-05-11 Carl Love <cel@us.ibm.com>
18746 * config/rs6000/rs6000-c: Add support for built-in functions
18747 vector unsigned char vec_popcnt (vector signed char)
18748 vector unsigned char vec_popcnt (vector unsigned char)
18749 vector unsigned short vec_popcnt (vector signed short)
18750 vector unsigned short vec_popcnt (vector unsigned short)
18751 vector unsigned int vec_popcnt (vector signed int)
18752 vector unsigned int vec_popcnt (vector unsigned int)
18753 vector unsigned long long vec_popcnt (vector signed long long)
18754 vector unsigned long long vec_popcnt (vector unsigned long long)
18755 vector signed long long vec_slo (vector signed long long,
18756 vector signed char)
18757 vector signed long long vec_slo (vector signed long long,
18758 vector unsigned char)
18759 vector unsigned long long vec_slo (vector unsigned long long,
18760 vector signed char)
18761 vector unsigned long long vec_slo (vector unsigned long long,
18762 vector unsigned char)
18763 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18764 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18765 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18766 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18767 * doc/extend.texi: Update the built-in documentation file for the
18768 new built-in functions.
18770 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18772 * attribs.h (sorted_attr_string): Move machine independent
18773 functions for target clone support from the i386 port to common
18774 code. Rename ix86_function_versions to common_function_versions.
18775 Rename make_name to make_unique_name.
18776 (common_function_versions): Likewise.
18777 (make_unique_name): Likewise.
18778 (make_dispatcher_decl): Likewise.
18779 (is_function_default_version): Likewise.
18780 * attribs.c (attr_strcmp): Likewise.
18781 (sorted_attr_string): Likewise.
18782 (common_function_versions): Likewise.
18783 (make_unique_name): Likewise.
18784 (make_dispatcher_decl): Likewise.
18785 (is_function_default_version): Likewise.
18786 * config/i386/i386.c (attr_strcmp): Likewise.
18787 (sorted_attr_string): Likewise.
18788 (ix86_function_versions): Likewise.
18789 (make_name): Likewise.
18790 (make_dispatcher_decl): Likewise.
18791 (is_function_default_version): Likewise.
18792 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18794 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18797 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18798 Account for direct move costs for vec_construct of integer
18801 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18804 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18805 (UNSPEC_STX_ATOMIC): Ditto.
18806 (loaddi_via_sse): New insn.
18807 (storedi_via_sse): Ditto.
18808 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18809 Update corresponding peephole2 patterns.
18810 (atomic_storedi_fpu): Ditto.
18812 2017-05-11 Julia Koval <julia.koval@intel.com>
18814 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18815 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18817 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18818 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18819 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18821 2017-05-11 Nathan Sidwell <nathan@acm.org>
18823 * graphite-poly.c: Include dumpfile.h.
18825 * dumpfle.h (dump_function): Declare here ...
18826 * tree-dump.h (dump_function): ... not here.
18827 * dumpfile.c: #include tree-cfg.h.
18828 (dump_function): Move here from ...
18829 * tree-dump.c (dump_function): ... here.
18830 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18831 * graphite-poly.c: Don't include tree-dump.h.
18832 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18833 * print-tree.c: Likewise.
18834 * stor-layout.c: Likewise.
18835 * tree-nested.c: Likewise.
18837 * dumpfile.c (dump_start): Use TDF_FLAGS.
18838 (dump_enable_all): Fix TDF_KIND check thinko.
18840 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18842 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18843 array entries to represent two legal parameterizations of the
18844 overloaded __builtin_cmpb function, as represented by the
18845 P6_OV_BUILTIN_CMPB constant.
18846 (altivec_resolve_overloaded_builtin): Add special case handling
18847 for the __builtin_cmpb function, as represented by the
18848 P6_OV_BUILTIN_CMPB constant.
18849 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18850 (BU_P6_64BIT_2): New macro.
18851 (BU_P6_OVERLOAD_2): New macro
18852 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18853 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18854 (CMPB): Add overload support to represent both 32-bit and 64-bit
18855 compare-bytes function.
18856 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18857 support for TARGET_CMPB.
18858 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18859 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18860 documentation of the __builtin_cmpb overloaded built-in function.
18862 2017-05-11 Richard Biener <rguenther@suse.de>
18864 PR tree-optimization/80705
18865 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18866 bases are not vectorizable.
18868 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18870 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18871 when counting register pressure.
18873 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18875 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18876 (iv_ca_more_deps): Renamed to ...
18877 (iv_ca_compare_deps): ... this.
18878 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18880 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18882 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18884 (determine_group_iv_costs): ... here.
18885 (find_inv_vars_cb): Record inv var if it's not recorded before.
18887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18889 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18890 (get_shiftadd_cost): Ditto.
18892 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18894 * tree-ssa-address.c: Include header file.
18895 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18897 (add_to_parts): Refactor.
18898 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18899 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18902 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18904 PR tree-optimization/53090
18905 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18907 (extract_cond_operands): Detect condition with IV on both sides
18908 and return COMP_IV_EXPR_2.
18909 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18910 (rewrite_use_compare): Simplify by removing call to function
18911 extract_cond_operands.
18913 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18915 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18916 (extract_cond_operands): Detect condition comparing against non-
18917 invariant bound and return appropriate enum value.
18918 (find_interesting_uses_cond): Update use of extract_cond_operands.
18919 Handle its return value accordingly.
18920 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18924 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18925 nonlinear iv_use computation in loop invariant sensitive way.
18927 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18929 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18930 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18932 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18934 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18935 (dump_cand): Support iv_cand.inv_exprs.
18936 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18938 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18941 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18943 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18945 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18946 as local function. Include necessary header files.
18947 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18949 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18951 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18953 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18955 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18956 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18957 RSHIFT_EXPR and BIT_NOT_EXPR.
18959 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18961 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18962 (adjust_setup_cost): New parameter supporting round up adjustment.
18963 (struct address_cost_data): Delete.
18964 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18965 (split_address_cost, ptr_difference_cost): Delete.
18966 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18967 (struct ainc_cost_data): New struct.
18968 (get_address_cost_ainc): New function.
18969 (get_address_cost, get_computation_cost): Reimplement.
18970 (determine_group_iv_cost_address): Record inv_expr for all uses of
18972 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18973 (iv_ca_has_deps): Reimplemented to ...
18974 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18976 (iv_ca_extend): Call iv_ca_more_deps.
18978 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18980 * tree-ssa-address.c (struct mem_address): Move to header file.
18981 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18982 * tree-ssa-address.h (struct mem_address): Move from C file.
18983 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18985 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18987 * tree-affine.h (aff_combination_type): New interface.
18988 (aff_combination_zero_p): Remove static.
18989 (aff_combination_const_p): New interface.
18990 (aff_combination_singleton_var_p): New interfaces.
18992 2017-05-11 Richard Biener <rguenther@suse.de>
18994 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18995 Skip unreachable blocks and destinations.
18996 (eliminate): Move stmt removal and fixup ...
18997 (fini_eliminate): ... here. Skip inserted exprs.
18998 (pass_pre::execute): Move fini_pre after fini_eliminate.
18999 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19000 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19001 PRE to get rid of dead code that has invalid SSA form and
19002 split critical edges again.
19004 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19006 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19008 2017-05-11 Richard Biener <rguenther@suse.de>
19010 * passes.c (execute_function_todo): Verify loops if they are
19011 said to be up-to-date.
19012 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19013 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19015 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19018 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19019 handle calling assemble_external ourself.
19022 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19023 modes with zero size. Enhance comment.
19025 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19027 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19028 built-ins for vec_xl and vec_xst with short and char pointer
19031 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19033 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19034 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19035 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19036 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19037 (_mm_maskz_min_round_ss): New intrinsics.
19038 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19039 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19040 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19041 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19042 (__builtin_ia32_minss_mask_round): New builtins.
19043 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19044 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19045 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19047 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19048 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19050 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19053 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19055 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19056 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19057 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19058 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19059 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19060 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19061 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19062 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19063 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19064 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19065 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19066 (__builtin_ia32_mulss_mask_round): New builtins.
19067 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19068 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19069 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19071 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19072 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19074 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19077 2017-05-10 Julia Koval <julia.koval@intel.com>
19079 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19080 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19081 (_mm256_setr_m128i): New intrinsics.
19083 2017-05-10 Julia Koval <julia.koval@intel.com>
19085 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19086 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19087 (_mm_maskz_rcp14_ss): New intrinsics.
19088 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19089 (__builtin_ia32_rcp14ss_mask): New builtins.
19090 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19092 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19094 PR tree-optimization/51513
19095 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19096 (assert_unreachable_fallthru_edge_p): Use it.
19097 (group_case_labels_stmt): Likewise.
19098 * tree-cfg.h: Prototype it.
19099 * stmt.c: Include cfghooks.h and tree-cfg.h.
19100 (emit_case_dispatch_table) <gap_label>: New local variable.
19101 Use it to fill dispatch table gaps.
19102 Test for default_label before updating probabilities.
19103 (expand_case) <default_label>: Remove unneeded initialization.
19104 Test for unreachable default case statement and remove its edge.
19105 Set default_label accordingly.
19106 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19108 2017-05-10 Carl Love <cel@us.ibm.com>
19110 * config/rs6000/rs6000-c: Add support for built-in functions
19111 vector signed char vec_neg (vector signed char)
19112 vector signed short int vec_neg (vector short int)
19113 vector signed int vec_neg (vector signed int)
19114 vector signed long long vec_neg (vector signed long long)
19115 vector float vec_neg (vector float)
19116 vector double vec_neg (vector double)
19117 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19119 * config/rs6000/altivec.h: Add define for vec_neg
19120 * doc/extend.texi: Update the built-in documentation for the
19121 new built-in functions.
19123 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19125 PR tree-optimization/77644
19126 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19128 2017-05-10 Nathan Sidwell <nathan@acm.org>
19130 * dumpfile.h (TDI_lang_all): New.
19131 (TDF_KIND): New. Renumber others
19132 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19134 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19136 (get_dump_file_name): Adjust suffix generation.
19137 (dump_enable_all): Use TDF_KIND.
19138 * doc/invoke.texi (-fdump-lang-all): Document.
19140 * dumpfile.h: Tabify.
19142 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19145 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19146 Move member access before delete.
19148 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19150 * tree-inline.c (expand_call_inline): Split block at stmt
19153 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19156 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19157 are now unused after splitting mov{sf,sd}_hardfloat.
19158 (f32_lr2): Likewise.
19159 (f32_lm): Likewise.
19160 (f32_lm2): Likewise.
19161 (f32_li): Likewise.
19162 (f32_li2): Likewise.
19163 (f32_lv): Likewise.
19164 (f32_sr): Likewise.
19165 (f32_sr2): Likewise.
19166 (f32_sm): Likewise.
19167 (f32_sm2): Likewise.
19168 (f32_si): Likewise.
19169 (f32_si2): Likewise.
19170 (f32_sv): Likewise.
19171 (f32_dm): Likewise.
19172 (f32_vsx): Likewise.
19173 (f32_av): Likewise.
19174 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19175 For movsf, order stores so the VSX stores occur before the GPR
19176 store which encourages the register allocator to use a traditional
19177 FPR instead of a GPR. For movsd, order the stores so that the GPR
19178 store comes before the VSX stores to allow the power6 to work.
19179 This is due to the power6 not having a 32-bit integer store
19180 instruction from a FPR.
19181 (movsf_hardfloat): Likewise.
19182 (movsd_hardfloat): Likewise.
19184 2017-05-09 Martin Sebor <msebor@redhat.com>
19186 PR translation/80280
19187 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19190 PR translation/80280
19191 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19192 data member added in r247778.
19193 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19195 2017-05-09 Nathan Sidwell <nathan@acm.org>
19197 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19199 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19202 2017-05-09 Marek Polacek <polacek@redhat.com>
19204 * doc/invoke.texi: Fix typo.
19206 2017-05-09 Richard Biener <rguenther@suse.de>
19208 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19209 (vrp_val_is_min): Likewise.
19210 (set_value_range_to_value): Likewise.
19211 (set_value_range_to_nonnegative): Likewise.
19212 (gimple_assign_nonzero_p): Likewise.
19213 (gimple_stmt_nonzero_p): Likewise.
19214 (vrp_int_const_binop): Likewise. Remove unreachable case.
19215 (adjust_range_with_scev): Adjust comments.
19216 (compare_range_with_value): Likewise.
19217 (extract_range_from_phi_node): Likewise.
19218 (test_for_singularity): Likewise.
19220 2017-05-09 Richard Biener <rguenther@suse.de>
19222 * tree-vrp.c (get_single_symbol): Add assert that we don't
19223 get overflowed constants as invariant part.
19224 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19225 checks. Use wi::cmp instead of recursing for integer constants.
19226 (compare_values): Just ignore whether we assumed undefined
19227 overflow instead of failing the compare.
19228 (extract_range_for_var_from_comparison_expr): Add comment before the
19229 TREE_NO_WARNING sets.
19230 (test_for_singularity): Likewise.
19231 (extract_range_from_comparison): Do not disable optimization
19232 when we assumed undefined overflow.
19233 (extract_range_basic): Remove init of unused var.
19235 2017-05-09 Richard Biener <rguenther@suse.de>
19237 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19238 (extract_range_from_multiplicative_op_1): Adjust.
19239 (extract_range_from_binary_expr_1): Use int_const_binop.
19241 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19244 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19245 rs6000_store_data_bypass_p in seven define_bypass directives and
19246 in several comments.
19247 * config/rs6000/rs6000-protos.h: Add prototype for
19248 rs6000_store_data_bypass_p function.
19249 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19250 function implements slightly different (rs6000-specific) semantics
19251 than store_data_bypass_p, returning false rather than aborting
19252 with assertion error when arguments do not satisfy the
19253 requirements of store data bypass.
19254 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19255 rs6000_store_data_bypass_p.
19257 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19259 * config/xtensa/xtensa-protos.h
19260 (xtensa_initial_elimination_offset): New declaration.
19261 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19262 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19263 macro definition, add case for FRAME_POINTER_REGNUM when
19264 FRAME_GROWS_DOWNWARD.
19265 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19266 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19267 xtensa_initial_elimination_offset.
19269 2017-05-08 Nathan Sidwell <nathan@acm.org>
19271 * doc/invoke.texi: Alphabetize -fdump options.
19273 2017-05-08 Martin Sebor <msebor@redhat.com>
19275 PR translation/80280
19276 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19278 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19280 * target.def (compute_frame_layout): New optional target hook.
19281 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19282 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19283 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19285 * reload1.c (verify_initial_elim_offsets): Likewise.
19286 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19287 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19288 (arm_get_frame_offsets): Split up into this ...
19289 (arm_compute_frame_layout): ... and this function.
19291 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19293 * config/aarch64/constraints.md (Usa): New constraint.
19294 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19296 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19298 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19299 with_multilib_list after it has been checked.
19301 2017-05-08 Richard Biener <rguenther@suse.de>
19303 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19304 (bitmap_set_subtract_values): Likewise.
19306 2017-05-08 Richard Biener <rguenther@suse.de>
19308 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19309 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19311 (gimple_stmt_nonzero_warnv_p): Rename to ...
19312 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19314 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19315 (extract_range_basic): Adjust, do not disable propagation on
19316 strict overflow sensitive simplification.
19317 (vrp_visit_cond_stmt): Likewise.
19319 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19321 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19322 body size unconditionally.
19324 2017-05-07 Jeff Law <law@redhat.com>
19327 2017-05-06 Jeff Law <law@redhat.com>
19328 PR tree-optimization/78496
19329 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19332 PR tree-optimization/78496
19333 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19334 (simplify_stmt_using_ranges): Call it.
19335 (vrp_dom_walker::before_dom_children): Extract equivalences
19336 from an ASSERT_EXPR with an equality comparison against a
19339 2017-05-06 Jeff Law <law@redhat.com>
19341 PR tree-optimization/78496
19342 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19345 PR tree-optimization/78496
19346 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19347 (simplify_stmt_using_ranges): Call it.
19348 (vrp_dom_walker::before_dom_children): Extract equivalences
19349 from an ASSERT_EXPR with an equality comparison against a
19352 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19354 * lra-constraints.c (lra_copy_reg_equiv): New function.
19355 (split_reg): Use it to copy equivalence information from the
19356 original register to the spill register.
19358 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19360 PR rtl-optimization/75964
19361 * simplify-rtx.c (simplify_const_relational_operation): Remove
19362 invalid handling of comparisons of integer ABS.
19364 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19366 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19367 initialize to zero.
19368 (init_regs): Remove declaration.
19369 (function_arg_advance_32): Initialize error_p as boolean variable.
19371 2017-05-05 Nathan Sidwell <nathan@acm.org>
19373 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19374 lines. Use for (;;).
19376 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19378 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19379 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19380 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19381 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19382 VF=2 that require versioning.
19384 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19386 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19389 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19391 * diagnostic.h (diagnostic_override_option_index): Convert from
19392 macro to inline function.
19394 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19396 * diagnostic.c (last_module_changed_p): New function.
19397 (set_last_module): New function.
19398 (diagnostic_report_current_module): Convert macro usage to
19399 the above functions.
19400 * diagnostic.h (diagnostic_context::last_module): Strengthen
19401 from const line_map * to const line_map_ordinary *.
19402 (diagnostic_last_module_changed): Delete macro.
19403 (diagnostic_set_last_module): Delete macro.
19405 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19407 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19408 with diagnostic_report_diagnostic.
19409 (diagnostic_n_impl_richloc): Likewise.
19410 * diagnostic.h (report_diagnostic): Delete macro.
19411 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19412 with diagnostic_report_diagnostic.
19413 * substring-locations.c (format_warning_va): Likewise.
19415 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19417 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19418 save/restor of format_spec. Move option-printing code to...
19419 (print_option_information): ...this new function, and
19420 reimplement by simply printing to the pretty_printer,
19421 rather than appending to the format string.
19423 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19425 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19426 handling logic into...
19427 (update_effective_level_from_pragmas): ...this new function.
19429 2017-05-04 Andrew Waterman <andrew@sifive.com>
19431 * config/riscv/riscv.opt (mstrict-align): New option.
19432 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19433 (SLOW_UNALIGNED_ACCESS): Define.
19434 (riscv_slow_unaligned_access): Declare.
19435 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19437 (riscv_slow_unaligned_access): New variable.
19438 (rocket_tune_info): Set slow_unaligned_access to true.
19439 (optimize_size_tune_info): Set slow_unaligned_access to false.
19440 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19441 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19442 (riscv_option_override): Set riscv_slow_unaligned_access.
19443 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19445 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19447 * config/riscv/riscv.md: Unify indentation.
19449 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19454 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19456 (extendsi<mode>2): Add support for doing sign extension via
19457 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19458 don't have ISA 3.0 instructions.
19459 (extendsi<mode>2 splitter): Likewise.
19460 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19461 generate the normal insns since SImode can now go in vector
19462 registers. Disallow the special UNSPECs needed for previous
19463 machines to hide SImode being used. Add new insns
19464 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19465 (fix_trunc<mode>si2_stfiwx): Likewise.
19466 (fix_trunc<mode>si2_internal): Likewise.
19467 (fixuns_trunc<mode>si2): Likewise.
19468 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19469 (fctiw<u>z_<mode>_smallint): Likewise.
19470 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19471 of floating point to 32-bit integer from doing a direct move to
19472 the GPR registers to do a store.
19473 (fctiwz_<mode>): Break long line.
19475 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19477 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19478 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19479 (addr_list, addr_offset_valid_p): New.
19480 (split_address_groups): Check offset validity with above function.
19481 (gt-tree-ssa-loop-ivopts.h): Include header file.
19483 2017-05-05 Nathan Sidwell <nathan@acm.org>
19485 * config.gcc (arm*-*-*): Add missing 'fi'.
19487 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19489 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19490 included in -fopt-info does not matter.
19491 * doc/optinfo.texi (-fopt-info): Fix description of default
19492 behavour. Explicitly say order of options included in -fopt-info
19495 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19497 * config.gcc: Allow combinations of aprofile and rmprofile values for
19498 --with-multilib-list.
19499 * config/arm/t-multilib: New file.
19500 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19501 variables. Remove setting of ISA and floating-point ABI in
19502 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19503 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19504 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19505 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19507 * config/arm/t-rmprofile: Likewise except for the matches changes.
19508 * doc/install.texi (--with-multilib-list): Document the combination of
19509 aprofile and rmprofile values and warn about pitfalls in doing that.
19511 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19513 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19514 (movdi_aarch64): Likewise.
19516 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19518 PR tree-optimization/80632
19519 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19521 (build_arrays): Initialize it for virtual phis.
19522 (fix_phi_nodes): Use it for virtual phis.
19524 PR tree-optimization/80558
19525 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19526 [x, y] op z into [x op, y op z] for op & or | if conditions
19529 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19530 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19533 * config/arm/arm.md (use_literal_pool): Remove.
19534 (64-bit immediate split): No longer takes cost into consideration
19535 if arm_disable_literal_pool is enabled.
19536 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19537 used when arm_disable_literal_pool is enabled.
19538 (arm_max_const_double_inline_cost): Remove use of
19539 arm_disable_literal_pool.
19540 (push_minipool_fix): Add assert.
19541 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19542 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19543 (no_literal_pool_sf_immediate): New.
19545 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19547 PR tree-optimization/80613
19548 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19549 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19551 2017-05-05 Richard Biener <rguenther@suse.de>
19553 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19555 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19557 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19558 of this flag from insn conditions due to removal from r247495.
19560 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19562 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19564 (arm_early_store_addr_dep_ptr): Likewise.
19565 * config/arm/aarch-common-protos.h
19566 (arm_early_load_addr_dep_ptr): Add prototype.
19567 (arm_early_store_addr_dep_ptr): Likewise.
19568 * config/arm/cortex-a53.md: Add new bypasses.
19570 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19572 * tree.c (next_type_uid): Change type to unsigned.
19573 (type_hash_canon): Decrement back next_type_uid if
19574 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19575 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19578 2017-05-04 Martin Sebor <msebor@redhat.com>
19580 * builtins.c: Fix a trivial typo in a comment.
19582 PR middle-end/79234
19583 * builtins.c (check_sizes): Adjust to handle reading past the end.
19584 Avoid printing excessive upper bound of ranges. Use %E to print
19585 tree nodes instead of converting them to %wu.
19586 (expand_builtin_memchr): New function.
19587 (compute_dest_size): Rename...
19588 (compute_objsize): ...to this.
19589 (expand_builtin_memcpy): Adjust.
19590 (expand_builtin_mempcpy): Adjust.
19591 (expand_builtin_strcat): Adjust.
19592 (expand_builtin_strcpy): Adjust.
19593 (check_strncat_sizes): Adjust.
19594 (expand_builtin_strncat): Adjust.
19595 (expand_builtin_strncpy): Adjust and simplify.
19596 (expand_builtin_memset): Adjust.
19597 (expand_builtin_bzero): Adjust.
19598 (expand_builtin_memcmp): Adjust.
19599 (expand_builtin): Handle memcmp.
19600 (maybe_emit_chk_warning): Check strncat just once.
19602 2017-05-04 Martin Sebor <msebor@redhat.com>
19604 PR preprocessor/79214
19605 PR middle-end/79222
19606 PR middle-end/79223
19607 * builtins.c (check_sizes): Add inlining context and issue
19608 warnings even when -Wno-system-headers is set.
19609 (check_strncat_sizes): Same.
19610 (expand_builtin_strncat): Same.
19611 (expand_builtin_memmove): New function.
19612 (expand_builtin_stpncpy): Same.
19613 (expand_builtin): Handle memmove and stpncpy.
19615 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19617 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19618 which is not used any more.
19620 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19622 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19624 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19626 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19627 (cortexa53_tunings): Likewise.
19628 (cortexa57_tunings): Likewise.
19629 (cortexa72_tunings): Likewise.
19630 (cortexa73_tunings): Likewise.
19632 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19634 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19635 Set loop alignment to 8.
19637 2017-05-04 Martin Sebor <msebor@redhat.com>
19639 PR translation/80280
19640 * builtins.c (expand_builtin_object_size): Add missing quoting to
19641 %D and like directives.
19642 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19643 (hsa_type_for_tree_type): Same.
19644 (verify_function_arguments): Same.
19645 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19646 * varasm.c (get_section): Same.
19649 2017-05-04 Martin Sebor <msebor@redhat.com>
19651 PR translation/80280
19652 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19654 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19656 * config/aarch64/aarch64.c (generic_addrcost_table):
19657 Change HI/TI mode setting.
19659 2017-05-04 Martin Jambor <mjambor@suse.cz>
19661 PR tree-optimization/80622
19662 * tree-sra.c (comes_initialized_p): New function.
19663 (build_accesses_from_assign): Only set write lazily when
19664 comes_initialized_p is false.
19665 (analyze_access_subtree): Use comes_initialized_p.
19666 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19667 instead of testing for PARM_DECL.
19669 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19671 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19672 constraint on operand 0 to allow more general addressing modes.
19673 Adjust output template.
19674 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19676 * config/aarch64/aarch64-protos.h
19677 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19678 * config/aarch64/constraints.md (Dp): New address constraint.
19679 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19682 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19684 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19685 update use of estimate_ipcp_clone_size_and_time.
19686 (estimate_local_effects): Update use of
19687 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19688 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19689 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19690 Return nonspecialized time.
19692 2017-05-04 Richard Biener <rguenther@suse.de>
19694 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19695 for the last VUSE which def dominates the PHI. Directly call
19697 (get_continuation_for_phi_1): Remove.
19699 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19701 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19702 to explain the use of truncating division. Cap the number of
19703 iterations to the maximum given by nb_iterations_upper_bound,
19706 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19708 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19709 * configure: Regenerate.
19710 * config.in: Regenerate.
19711 * config/i386/driver-mingw32.c: new file.
19712 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19713 * config.host: Link driver-mingw32.o on MinGW host.
19714 * doc/install.texi: Document new --enable-mingw-wildcard configure
19717 2017-05-04 Marek Polacek <polacek@redhat.com>
19719 PR tree-optimization/80612
19720 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19722 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19723 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19725 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19726 (movt splitter): Likewise.
19727 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19728 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19729 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19730 block for Thumb-1 with MOVT.
19731 (thumb2_legitimate_address_p): Move code block ...
19732 (can_avoid_literal_pool_for_label_p): ... into this new function.
19733 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19735 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19736 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19737 "M-profile targets with the MOVT instruction".
19739 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19741 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19742 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19743 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19745 2017-05-04 Martin Liska <mliska@suse.cz>
19747 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19748 variable cond_code.
19750 2017-05-04 Richard Biener <rguenther@suse.de>
19752 * tree.c (array_at_struct_end_p): Handle arrays at struct
19753 end with flexarrays more conservatively. Refactor and treat
19754 arrays of arrays or aggregates more strict. Fix
19755 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19756 * tree.c (array_at_struct_end_p): Adjust prototype.
19757 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19758 * gimple-fold.c (get_range_strlen): Likewise.
19759 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19761 2017-05-04 Richard Biener <rguenther@suse.de>
19763 PR tree-optimization/31130
19764 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19766 (supports_overflow_infinity): Likewise.
19767 (is_negative_overflow_infinity): Likewise.
19768 (is_positive_overflow_infinity): Likewise.
19769 (is_overflow_infinity): Likewise.
19770 (stmt_overflow_infinity): Likewise.
19771 (overflow_infinity_range_p): Likewise.
19772 (usable_range_p): Remove as always returning true.
19773 (make_overflow_infinity): Remove.
19774 (negative_overflow_infinity): Likewise.
19775 (positive_overflow_infinity): Likewise.
19776 (avoid_overflow_infinity): Likewise.
19777 (set_value_range): Adjust accordingly.
19778 (set_value_range_to_nonnegative): Likewise, remove now unused
19779 overflow_infinity arg.
19780 (vrp_operand_equal_p): Adjust.
19781 (update_value_range): Likewise.
19782 (range_int_cst_singleton_p): Likewise.
19783 (operand_less_p): Likewise.
19784 (compare_values_warnv): Likewise.
19785 (extract_range_for_var_from_comparison_expr): Likewise.
19786 (vrp_int_const_binop): Likewise.
19787 (zero_nonzero_bits_from_vr): Likewise.
19788 (extract_range_from_multiplicative_op_1): Likewise.
19789 (extract_range_from_binary_expr_1): Likewise.
19790 (extract_range_from_unary_expr): Likewise.
19791 (extract_range_from_comparison): Likewise.
19792 (extract_range_basic): Likewise.
19793 (adjust_range_with_scev): Likewise.
19794 (compare_ranges): Likewise.
19795 (compare_range_with_value): Likewise.
19796 (dump_value_range): Likewise.
19797 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19799 (simplify_cond_using_ranges): Adjust.
19801 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19803 * brig-builtins.def: Added a builtin for class_f64.
19804 * builtin-types.def: Added a builtin type needed by class_f64.
19806 2017-05-03 Jason Merrill <jason@redhat.com>
19808 * timevar.def: Add TV_CONSTEXPR.
19810 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19812 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19814 2017-05-03 Martin Jambor <mjambor@suse.cz>
19816 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19817 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19818 * ipa-cp.c (ipcp_propagate_stage): Do not call
19819 ipa_update_after_lto_read.
19820 * ipa-inline.c (ipa_inline): Likewise.
19822 2017-05-03 Martin Jambor <mjambor@suse.cz>
19824 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19825 tag. Added a default constructor and a destructor.
19826 (ipa_edge_args_sum_t): New class;
19827 (ipa_edge_args_sum): Declare.
19828 (ipa_edge_args_vector): Remove declaration.
19829 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19830 (ipa_free_edge_args_substructures): Remove declaration.
19831 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19832 (ipa_edge_args_info_available_for_edge_p): Likewise.
19833 * ipa-prop.c (ipa_edge_args_vector): Removed.
19834 (edge_removal_hook_holder): Likewise.
19835 (edge_duplication_hook_holder): Likewise.
19836 (ipa_edge_args_sum): New variable.
19837 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19838 ipa_edge_args_vector.
19839 (ipa_free_edge_args_substructures): Likewise.
19840 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19841 ipa_edge_args_vector.
19842 (ipa_edge_removal_hook): Turned into method
19843 ipa_edge_args_sum_t::remove.
19844 (ipa_edge_duplication_hook): Turned into method
19845 ipa_edge_args_sum_t::duplicate.
19846 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19847 registering edge hooks.
19848 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19849 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19850 ipa_edge_args_sum instead of ipa_edge_args_vector.
19851 * ipa-profile.c (ipa_profile): Likewise.
19853 2017-05-03 Martin Jambor <mjambor@suse.cz>
19855 * symbol-summary.h (function_summary): New method exists.
19856 (function_summary::symtab_removal): Deallocate through release.
19857 (call_summary): New class.
19858 (gt_ggc_mx): New overload.
19859 (gt_pch_nx): Likewise.
19860 (gt_pch_nx): Likewise.
19862 2017-05-03 Jeff Law <law@redhat.com>
19864 PR tree-optimization/78496
19865 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19866 from simplify_cond_using_ranges. Split off code to walk
19867 backwards through casts into ...
19868 (simplify_cond_using_ranges_2): New function.
19869 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19870 (execute_vrp): After identifying jump threads, call
19871 simplify_cond_using_ranges_2.
19873 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19876 * ipa-inline.h (inline_summary): Add ctor.
19877 (create_ggc): Do not use ggc_cleared_alloc.
19879 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19880 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19882 * gcc.c (handle_braces): Support escaping in switch matching
19884 * doc/invoke.texi (Spec Files): Document it.
19885 Remove superfluous @code markup in items.
19887 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19889 * diagnostic-show-locus.c (struct column_range): New struct.
19890 (get_affected_columns): New function.
19891 (get_printed_columns): New function.
19892 (struct correction): New struct.
19893 (correction::ensure_capacity): New function.
19894 (correction::ensure_terminated): New function.
19895 (struct line_corrections): New struct.
19896 (line_corrections::~line_corrections): New dtor.
19897 (line_corrections::add_hint): New function.
19898 (layout::print_trailing_fixits): Reimplement in terms of the new
19900 (selftest::test_overlapped_fixit_printing): New function.
19901 (selftest::diagnostic_show_locus_c_tests): Call it.
19903 2017-05-03 Nathan Sidwell <nathan@acm.org>
19905 Canonicalize canonical type hashing
19906 * tree.h (type_hash_canon_hash): Declare.
19907 * tree.c (type_hash_list, attribute_hash_list): Move into
19908 type_hash_canon_hash.
19909 (build_type_attribute_qual_variant): Break out hash code calc into
19910 type_hash_canon_hash.
19911 (type_hash_canon_hash): New. Generic type hash computation.
19912 (build_range_type_1, build_array_type_1, build_function_type,
19913 build_method_type_directly, build_offset_type, build_complex_type,
19914 make_vector_type): Call it.
19916 2017-05-03 Richard Biener <rguenther@suse.de>
19918 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19919 When all DRs have unknown misaligned do not always peel
19920 when there is a store but apply the same costing model as if
19921 there were only loads.
19923 2017-05-03 Richard Biener <rguenther@suse.de>
19926 PR tree-optimization/80492
19927 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19928 compare_base_decls returning dont-know properly.
19930 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19932 * config/arm/iterators.md (CCSI): New mode iterator.
19933 (arch): New mode attribute.
19934 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19935 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19936 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19937 code iterator for success result mode.
19938 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19939 the corresponding new insn generators.
19941 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19944 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19945 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19947 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19949 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19950 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19951 (DDR_A): Wrap DDR argument in brackets.
19952 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19953 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19954 (DDR_REVERSED_P): Likewise.
19956 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19958 PR tree-optimization/79472
19959 * tree-switch-conversion.c (struct switch_conv_info): Add
19960 contiguous_range and default_case_nonstandard fields.
19961 (collect_switch_conv_info): Compute contiguous_range and
19962 default_case_nonstandard fields, don't clear final_bb if
19963 contiguous_range and only the default case doesn't have the required
19965 (check_all_empty_except_final): Set default_case_nonstandard instead
19966 of failing if contiguous_range and the default case doesn't have empty
19968 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19969 and only the default case doesn't have the required constants. Skip
19971 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19972 if default_case_nonstandard.
19973 (build_constructors): Build constant 1 just once. Assert that default
19974 values aren't inserted in between cases if contiguous_range. Skip
19976 (build_arrays): Skip virtual phis.
19977 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19978 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19979 Handle virtual phis.
19980 (gen_inbound_check): Handle default_case_nonstandard case.
19981 (process_switch): Adjust check_final_bb caller. Call
19982 gather_default_values with the first non-default case instead of
19983 default case if default_case_nonstandard.
19985 2017-05-02 Nathan Sidwell <nathan@acm.org>
19987 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19988 check. Fix formatting.
19990 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19992 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19993 errors when comparing specialized and unspecialized times.
19995 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19997 * diagnostic-show-locus.c
19998 (layout::should_print_annotation_line_p): Make private.
19999 (layout::print_annotation_line): Make private.
20000 (layout::annotation_line_showed_range_p): Make private.
20001 (layout::show_ruler): Make private.
20002 (layout::print_source_line): Make private. Pass in line and
20003 line_width, rather than calling location_get_source_line. Drop
20005 (layout::print_leading_fixits): New method.
20006 (layout::print_any_fixits): Rename to...
20007 (layout::print_trailing_fixits): ...this, and make private.
20008 Don't print newline fixits.
20009 (diagnostic_show_locus): Move logic for printing one row into...
20010 (layout::print_line): ...this new function. Move the
20011 location_get_source_line call and error-handling from
20012 print_source_line to here. Call print_leading_fixits, and rename
20013 print_any_fixits to print_trailing_fixits.
20014 (selftest::test_fixit_insert_containing_newline): Update now that
20015 newlines are partially supported.
20016 (selftest::test_fixit_insert_containing_newline_2): New test.
20017 (selftest::test_fixit_replace_containing_newline): Update comments.
20018 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20019 * edit-context.c (class added_line): New class.
20020 (class edited_line): Describe newline handling in comment.
20021 (edited_line::actually_edited_p): New method.
20022 (edited_line::print_content): Delete redundant decl.
20023 (edited_line::m_predecessors): New field.
20024 (edited_file::print_content): Call edited_line::print_content.
20025 (edited_file::print_diff): Update to support newlines.
20026 (edited_file::print_diff_hunk): Likewise.
20027 (edited_file::print_run_of_changed_lines): New function.
20028 (edited_file::print_diff_line): Convert to...
20029 (print_diff_line): ...this.
20030 (edited_file::get_effective_line_count): New function.
20031 (edited_line::edited_line): Initialize new field m_predecessors.
20032 (edited_line::~edited_line): Clean up m_predecessors.
20033 (edited_line::apply_fixit): Handle newlines.
20034 (edited_line::get_effective_line_count): New function.
20035 (edited_line::print_content): New function.
20036 (edited_line::print_diff_lines): New function.
20037 (selftest::test_applying_fixits_insert_containing_newline): New
20039 (selftest::test_applying_fixits_replace_containing_newline): New
20041 (selftest::insert_line): New function.
20042 (selftest::test_applying_fixits_multiple_lines): Add example of
20044 (selftest::edit_context_c_tests): Call the new tests.
20046 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20048 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20049 parameter cand. Update dump information.
20050 (get_computation_cost): Update uses.
20052 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20054 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20055 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20056 (get_computation_at, rewrite_use_address): Update use of
20057 get_computation_aff.
20059 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20061 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20062 (get_computation): Delete.
20063 (get_computation_cost): Implement like get_computation_cost_at.
20064 Use get_computation_at.
20065 (get_computation_cost_at): Delete.
20066 (rewrite_use_nonlinear_expr): Use get_computation_at.
20067 (rewrite_use_compare, remove_unused_ivs): Ditto.
20069 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20071 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20073 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20075 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20076 (ivopts_global_cost_for_size): Rename parameter and update uses.
20077 (iv_ca_recount_cost): Update uses.
20078 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20079 candidates seperately in n_invs and n_cands.
20080 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20082 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20084 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20085 (find_inv_vars_cb): New.
20086 (find_depends): Renamed to ...
20087 (find_inv_vars): ... this.
20088 (add_candidate_1, force_var_cost): Call find_inv_vars.
20089 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20091 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20093 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20094 inv_vars. Add inv_exprs.
20095 (struct iv_cand): Rename depends_on to inv_vars.
20096 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20097 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20098 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20099 (dump_cand): Dump inv_vars.
20100 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20101 (record_invariant, find_depends, add_candidate_1): Ditto.
20102 (set_group_iv_cost, force_var_cost): Ditto.
20103 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20104 (get_computation_cost_at, get_computation_cost): Ditto.
20105 (determine_group_iv_cost_generic): Ditto.
20106 (determine_group_iv_cost_address): Ditto.
20107 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20108 (determine_group_iv_costs): Ditto.
20109 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20110 (iv_ca_set_remove_invariants): Renamed to ...
20111 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20112 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20113 (iv_ca_set_add_invariants): Renamed to ...
20114 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20115 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20116 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20117 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20118 (create_new_ivs): Remove useless dump.
20120 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20122 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20124 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20125 (iv_ca_set_no_cp, create_new_iv): Ditto.
20127 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20129 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20131 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20133 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20134 function tree_check2.
20136 2017-05-02 Martin Liska <mliska@suse.cz>
20138 * doc/gcov.texi: Add missing preposition.
20139 * gcov.c (function_info::function_info): Properly fill up
20140 all member variables.
20142 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20144 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20146 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20148 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20150 2017-05-02 Martin Liska <mliska@suse.cz>
20153 * lto-streamer-in.c (lto_read_tree_1): Remove
20154 LTO_STREAMER_DEBUG.
20155 * lto-streamer.c (struct tree_hash_entry): Likewise.
20156 (struct tree_entry_hasher): Likewise.
20157 (tree_entry_hasher::hash): Likewise.
20158 (tree_entry_hasher::equal): Likewise.
20159 (lto_streamer_init): Likewise.
20160 (lto_orig_address_map): Likewise.
20161 (lto_orig_address_get): Likewise.
20162 (lto_orig_address_remove): Likewise.
20163 * lto-streamer.h: Likewise.
20164 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20165 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20167 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20169 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20170 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20171 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20172 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20173 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20174 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20175 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20176 (mm_maskz_sub_ss): New intrinsics.
20177 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20178 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20179 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20180 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20181 (__builtin_ia32_subss_mask_round): New builtins.
20182 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20183 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20184 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20186 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20187 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20189 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20192 2017-05-02 Martin Jambor <mjambor@suse.cz>
20194 PR tree-optimization/78687
20195 * tree-sra.c (access): New field parent.
20196 (process_subtree_disqualification): New function.
20197 (disqualify_candidate): Call it.
20198 (build_accesses_from_assign): Reset write flag if creating an
20200 (build_access_subtree): Fill in parent field and also prpagate
20201 down grp_write flag.
20202 (create_artificial_child_access): New parameter set_grp_write, set
20203 grp_write to its value.
20204 (propagate_subaccesses_across_link): Also propagate grp_write flag
20206 (propagate_all_subaccesses): Push the closest parent back to work
20207 queue if add_access_to_work_queue returned true.
20209 2017-05-02 Richard Biener <rguenther@suse.de>
20211 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20212 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20213 -fstrict-overflow documentation.
20214 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20215 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20216 flag_strict_overflow.
20217 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20218 * lto-opts.c (lto_write_options): Do not stream it.
20219 * lto-wrapper.c (merge_and_complain): Do not handle it.
20220 * opts.c (default_options_table): Do not set -fstrict-overflow.
20221 (finish_options): Likewise do not clear it when sanitizing.
20222 * simplify-rtx.c (simplify_const_relational_operation): Do not
20223 test flag_strict_overflow.
20225 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20227 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20228 using enabled attribute.
20229 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20230 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20231 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20232 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20233 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20234 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20235 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20236 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20237 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20238 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20240 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20242 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20244 2017-05-02 Richard Biener <rguenther@suse.de>
20246 PR tree-optimization/80591
20248 2017-04-10 Richard Biener <rguenther@suse.de>
20250 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20253 2017-05-02 Richard Biener <rguenther@suse.de>
20255 PR tree-optimization/80549
20256 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20257 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20258 headers if they do not have a preheader.
20260 2017-05-02 Martin Liska <mliska@suse.cz>
20263 * common.opt: Fix typo.
20264 * doc/invoke.texi: Likewise.
20266 2017-05-01 Jan Beulich <jbeulich@suse.com>
20268 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20269 swapping, add (x,x,m,x,n) alternative.
20271 2017-05-01 Nathan Sidwell <nathan@acm.org>
20273 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20274 unnecessary unadjusted_alignment check.
20276 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20279 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20281 * gimplify.c (gimplify_cilk_detach): New function.
20282 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20283 * tree-core.h: Document EXPR_CILK_SPAWN.
20284 * tree.h (EXPR_CILK_SPAWN): Define.
20286 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20288 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20289 to use new fixit_hint representation, using the "replace" logic.
20290 (get_line_span_for_fixit_hint): Likewise.
20291 (layout::print_any_fixits): Likewise.
20292 (selftest::test_one_liner_many_fixits): Rename to...
20293 (selftest::test_one_liner_many_fixits_1): ...this, and update
20294 comment and expected output to reflect that the multiple fix-it
20295 hints are now consolidated into one insertion.
20296 (selftest::test_one_liner_many_fixits_2): New test.
20297 (selftest::test_diagnostic_show_locus_one_liner): Update for
20299 (selftest::test_fixit_consolidation): Update for fix-it API
20301 * diagnostic.c (print_parseable_fixits): Likewise.
20302 * edit-context.c (edited_line::m_line_events): Convert from
20303 auto_vec <line_event *> to auto_vec <line_event>.
20304 (class line_event): Convert from abstract base class to a concrete
20305 class, taking over the role of replace_event.
20306 (class insert_event): Delete.
20307 (class replace_event): Rename to class line_event. Convert to
20309 (edit_context::add_fixits): Reimplement.
20310 (edit_context::apply_insert): Delete.
20311 (edit_context::apply_replace): Rename to...
20312 (edit_context::apply_fixit): ...this. Convert to half-open range.
20313 (edited_file::apply_insert): Delete.
20314 (edited_file::apply_replace): Rename to...
20315 (edited_file::apply_fixit): ...this.
20316 (edited_line::~edited_line): Drop deletion of events.
20317 (edited_line::apply_insert): Delete.
20318 (edited_line::apply_replace): Rename to...
20319 (edited_line::apply_fixit): ...this. Convert to half-open range.
20320 Update for change to type of m_line_events.
20321 * edit-context.h (edit_context::apply_insert): Delete.
20322 (edit_context::apply_replace): Rename to...
20323 (edit_context::apply_fixit): ...this.
20325 2017-05-01 Martin Sebor <msebor@redhat.com>
20327 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20330 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20333 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20334 __get_cpuid_max returns 0.
20335 (__get_cpuid_count): Ditto.
20337 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20339 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20340 replacement expression is another instance of one of its arguments.
20342 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20345 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20346 check for stack push/pop autoinc.
20347 * config/i386/i386.c (ix86_agi_dependent): Return false
20348 if the only reason why modified_in_p returned true is that
20349 addr is SP based and set_insn is a push or pop.
20351 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20353 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20356 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20359 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20360 (account_size_time): Use two predicates - exec_pred and
20362 (evaluate_conditions_for_known_args): Compute both clause and
20364 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20365 (inline_summary_t::duplicate): Update.
20366 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20368 (compute_inline_parameters): Likewise.
20369 (estimate_edge_size_and_time): Update caluclation of time.
20370 (estimate_node_size_and_time): Compute both time and nonspecialized
20372 (estimate_ipcp_clone_size_and_time): Update.
20373 (inline_merge_summary): Update.
20374 (do_estimate_edge_time): Update.
20375 (do_estimate_edge_size): Update.
20376 (do_estimate_edge_hints): Update.
20377 (inline_read_section, inline_write_summary): Stream both new predicates.
20378 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20380 (compute_inlined_call_time): Cleanup.
20381 (big_speedup_p): Update.
20382 (edge_badness): Update.
20383 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20384 (size_time_entry): Replace predicate by exec_predicate and
20385 nonconst_predicate.
20386 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20387 (estimate_edge_time): Return also nonspec_time.
20388 (reset_edge_growth_cache): Update.
20390 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20392 PR rtl-optimization/80491
20393 * ifcvt.c (noce_process_if_block): When looking for x setter
20394 with missing else_bb, don't check only the insn right before
20395 cond_earliest, but look for the last insn that x is modified in
20396 within the same bb.
20398 PR rtl-optimization/80491
20399 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20401 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20403 PR tree-optimization/80487
20404 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20406 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20408 PR tree-optimization/79697
20409 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20410 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20411 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20413 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20414 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20416 2017-04-28 Martin Sebor <msebor@redhat.com>
20418 PR tree-optimization/80523
20419 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20420 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20422 (maybe_warn, format_directive, parse_directive): Use new functions.
20423 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20425 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20427 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20429 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20431 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20432 target_header_dir): Set correctly.
20433 * configure: Regenerated.
20434 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20435 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20436 instead of SYSTEM_HEADER_DIR.
20438 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20440 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20441 (estimate_local_effects): Likewise.
20442 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20443 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20444 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20445 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20446 do_estimate_edge_time, estimate_edge_time): Likewise.
20447 * ipa-inline-analysis.c (estimate_node_size_and_time,
20448 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20449 (estimate_time_after_inlining): Remove.
20451 2017-04-28 Martin Liska <mliska@suse.cz>
20453 * doc/gcov.texi: Enhance documentation of gcov.
20455 2017-04-28 Martin Liska <mliska@suse.cz>
20457 * doc/gcov.texi: Sort options in alphabetic order.
20458 * doc/gcov-dump.texi: Likewise.
20459 * doc/gcov-tool.texi: Likewise.
20460 * gcov.c (print_usage): Likewise.
20461 * gcov-dump.c (print_usage): Likewise.
20462 * gcov-tool.c (print_merge_usage_message): Likewise.
20463 (print_rewrite_usage_message): Likewise.
20464 (print_overlap_usage_message): Likewise.
20466 2017-04-28 Martin Liska <mliska@suse.cz>
20468 PR gcov-profile/53915
20469 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20471 2017-04-28 Martin Liska <mliska@suse.cz>
20473 PR gcov-profile/79891
20474 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20475 is marked by compiler as living on a line.
20476 (get_cycles_count): Remove usage of the union.
20477 (output_intermediate_file): Likewise.
20478 (find_source): Fix GNU coding style.
20479 (accumulate_line_counts): Remove old non-all block mode.
20480 (output_lines): Remove usage of the union.
20481 * profile.c (output_location): Include all BBs, even if
20482 belonging to a same line (and file) as a previous BB.
20484 2017-04-28 Martin Liska <mliska@suse.cz>
20486 * gcov.c (process_args): Handle new argument 'w'.
20487 (read_graph_file): Assign ID to BBs.
20488 (output_branch_count): Display BB # if verbose flag is set.
20489 (output_lines): Likewise for arcs.
20490 (print_usage): Add '--verbose' option help.
20491 * doc/gcov.texi: Document --verbose (-w) option.
20493 2017-04-28 Martin Liska <mliska@suse.cz>
20495 * gcov.c (struct block_location_info): New struct.
20496 (process_file): Fill up the new structure.
20497 (read_graph_file): Replace usage of encoding by the newly added
20499 (add_line_counts): Likewise.
20500 (accumulate_line_counts): Remove usage of the union.
20501 (function_info::function_info): New function.
20502 (function_info::~function_info): Likewise.
20503 (process_file): Call delete instead of release_function.
20504 (release_function): Release the function.
20505 (release_structures): Call delete instead of release_function.
20506 (solve_flow_graph): Replace usage of num_blocks.
20507 (find_exception_blocks): Likewise.
20508 (output_lines): Fix GNU coding style.
20510 2017-04-28 Martin Liska <mliska@suse.cz>
20513 * coverage.c (coverage_remove_note_file): New function.
20514 * coverage.h: Declare the function.
20515 * toplev.c (finalize): Clean if an error has been seen.
20517 2017-04-28 Martin Liska <mliska@suse.cz>
20519 PR gcov-profile/80031
20520 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20521 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20522 * gcov.c (read_graph_file): Read just number of blocks.
20523 * profile.c (branch_prob): Do not stream 0 flags per a basic
20526 2017-04-28 Martin Liska <mliska@suse.cz>
20528 * gcov-dump.c (tag_*): Add new argument to declarations.
20529 (dump_gcov_file): Likewise.
20530 (tag_blocks): Add and use new argument depth.
20531 (tag_arcs): Likewise.
20532 (tag_lines): Likewise.
20533 (tag_counters): Likewise.
20534 (tag_summary): Likewise.
20535 (dump_working_sets): Use depth to do a proper indentation.
20537 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20540 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20541 * symtab.c (symtab_node::debug_symtab): Move definition here.
20543 2017-04-28 Richard Biener <rguenther@suse.de>
20545 * lto-streamer.h (LTO_major_version): Bump to 7.
20547 2017-04-28 Richard Biener <rguenther@suse.de>
20549 * tree-vrp.c (assert_info): New struct.
20550 (add_assert_info): New helper.
20551 (register_edge_assert_for_2): Refactor to add asserts to a vector
20553 (register_edge_assert_for_1): Likewise.
20554 (register_edge_assert_for): Likewise.
20555 (finish_register_edge_assert_for): New helper actually registering
20556 asserts where live on edge.
20557 (find_conditional_asserts): Adjust.
20558 (find_switch_asserts): Likewise.
20559 (evrp_dom_walker::try_find_new_range): Generalize.
20560 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20562 2017-04-27 Marek Polacek <polacek@redhat.com>
20565 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20566 arg10 and arg11 to itype.
20568 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20570 * doc/extend.texi (Object Size Checking): Improve grammar.
20572 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20575 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20576 that the logic for permitting reciprocal estimates matches that
20579 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20582 * tree.c (type_cache_hasher::equal): Only compare
20583 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20584 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20585 non-aggregate element types.
20586 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20587 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20589 2017-04-27 Richard Biener <rguenther@suse.de>
20591 PR middle-end/80533
20592 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20593 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20594 keeping a reference to a trailing array.
20596 2017-04-27 Richard Biener <rguenther@suse.de>
20598 PR middle-end/80539
20599 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20600 being in loop-closed SSA form conservatively.
20601 (chrec_fold_multiply_poly_poly): Likewise.
20603 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20605 PR middle-end/79665
20606 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20607 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20609 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20612 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20613 (aarch64_function_arg_alignment): Return unsigned int again, but still
20614 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20615 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20616 Don't emit -Wpsabi note.
20617 (aarch64_function_arg_boundary): Likewise.
20618 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20621 2017-04-26 Nathan Sidwell <nathan@acm.org>
20623 * tree.h (crc32_unsigned_n): Declare.
20624 (crc32_unsigned, crc32_unsigned): Make inline.
20625 * tree.c (crc32_unsigned_bits): Replace with ...
20626 (crc32_unsigned_n): ... this.
20627 (crc32_unsigned, crc32_byte): Remove.
20628 (crc32_string): Remove unnecessary braces.
20630 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20632 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20633 * ipa-inline-analysis.c (MAX_TIME): Remove.
20634 (account_size_time): Use sreal for time.
20635 (dump_inline_summary): Update.
20636 (estimate_function_body_sizes): Update.
20637 (estimate_edge_size_and_time): Update.
20638 (estimate_calls_size_and_time): Update.
20639 (estimate_node_size_and_time): Update.
20640 (inline_merge_summary): Update.
20641 (inline_update_overall_summary): Update.
20642 (estimate_time_after_inlining): Update.
20643 (inline_read_section): Update.
20644 (inline_write_summary): Update.
20645 * ipa-inline.c (compute_uninlined_call_time): Update.
20646 (compute_inlined_call_time): Update.
20647 (recursive_inlining): Update.
20648 (inline_small_functions): Update.
20649 (dump_overall_stats): Update.
20650 * ipa-inline.h: Include sreal.h.
20651 (size_time_entry): Turn time to sreal.
20652 (inline_summary): Turn self_time nad time to sreal.
20654 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20656 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20658 (sreal::stream_out, sreal::stream_in): New.
20659 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20661 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20663 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20666 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20669 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20671 (dimode_scalar_chain::compute_convert_gain): Ditto.
20672 (dimode_scalar_chain::make_vector_copies): Ditto.
20673 (dimode_scalar_chain::convert_reg): Ditto.
20674 (dimode_scalar_chain::convert_insn): Ditto.
20675 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20676 (VI248_AVX512BW_1): New mode iterator.
20677 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20678 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20681 2017-04-25 Martin Sebor <msebor@redhat.com>
20683 PR tree-optimization/80497
20684 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20685 constants are representable in HOST_WIDE_INT.
20686 (parse_directive): Ditto.
20688 2017-04-25 Martin Sebor <msebor@redhat.com>
20691 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20692 (new_zero_array): Adjust signature.
20693 (dom_info::dom_init): Used unsigned rather that size_t.
20694 (dom_info::dom_info): Same.
20696 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20697 Jakub Jelinek <jakub@redhat.com>
20700 * config/arm/arm.c: Include gimple.h.
20701 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20702 returns negative, increment ncrn only if it returned positive.
20703 (arm_needs_doubleword_align): Return int instead of bool,
20704 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20705 members, but if there is any such non-FIELD_DECL
20706 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20707 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20708 returns negative, increment nregs only if it returned positive.
20709 (arm_setup_incoming_varargs): Likewise.
20710 (arm_function_arg_boundary): Emit -Wpsabi note if
20711 arm_needs_doubleword_align returns negative, return
20712 DOUBLEWORD_ALIGNMENT only if it returned positive.
20714 2017-04-25 Marek Polacek <polacek@redhat.com>
20717 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20718 first argument to type.
20720 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20724 type checks to test for compatibility instead of equality.
20726 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20727 Jakub Jelinek <jakub@redhat.com>
20730 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20732 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20733 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20734 the alignment computation, but return their maximum in warn_alignment.
20735 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20736 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20738 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20739 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20742 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20744 * config/arc/simdext.md (dmpyh): Fix typo.
20746 2017-04-25 Richard Biener <rguenther@suse.de>
20748 PR tree-optimization/80492
20749 * alias.c (compare_base_decls): Handle registers with asm
20750 specification conservatively.
20751 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20752 compare_base_decls returning dont-know properly.
20754 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20756 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20757 (legitimate_offset_address_p): New function.
20758 (arc_legitimate_address_p): Use above function.
20760 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20762 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20764 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20766 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20767 ACCH registers whenever they are available.
20769 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20771 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20772 double regs fix when not used.
20774 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20776 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20778 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20779 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20781 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20783 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20784 when emitting short ADD instructions.
20786 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20788 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20790 (cmpsi_cc_c_insn): Likewise.
20791 (cbranchsi4_scratch): Compute proper instruction length using
20792 compact_hreg_operand.
20793 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20795 2017-04-25 Richard Biener <rguenther@suse.de>
20797 PR middle-end/80509
20798 * passes.c (pass_manager::pass_manager): Initialize
20799 m_name_to_pass_map.
20801 2017-04-25 Richard Biener <rguenther@suse.de>
20803 PR tree-optimization/79201
20804 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20806 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20809 * config/s390/vector.md: Split MEM->GPR vector moves for
20810 non-s_operand addresses.
20812 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20815 * config/s390/predicates.md (reload_const_wide_int_operand): New
20817 * config/s390/s390.md ("movti"): Remove d/P alternative.
20818 ("movti_bigconst"): New pattern definition.
20820 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20823 * s390-protos.h (s390_expand_cs_hqi): Removed.
20824 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20825 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20826 modes as well as CCZ1mode and CCZmode.
20827 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20828 signature of s390_emit_compare_and_swap.
20829 (s390_expand_cs_hqi): Likewise, make static.
20830 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20831 compare-and-swap, in some cases.
20832 (s390_expand_cs): Wrapper function.
20833 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20835 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20836 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20837 patterns for small and large integers. Forbid symref memory operands.
20838 Move expander to s390.c. Require cc register.
20839 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20840 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20841 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20842 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20843 symref memory operands. Remove CC mode and call s390_match_ccmode
20845 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20847 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20849 * config/s390/s390.md (define_peephole2): New peephole to help
20850 combining the load-and-test pattern with volatile memory.
20852 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20854 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20855 with CCZmode for TARGET_Z196.
20857 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20859 PR rtl-optimization/80501
20860 * combine.c (make_compound_operation_int): Set subreg_code to SET
20861 even for AND with mask of the sign bit of mode.
20863 PR rtl-optimization/80500
20864 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20865 sum's initial value.
20867 2017-04-25 Julian Brown <julian@codesourcery.com>
20868 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20870 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20872 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20874 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20876 2017-04-25 Julian Brown <julian@codesourcery.com>
20877 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20879 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20880 (thunderx2t99_sha): New Reservation.
20882 2017-04-25 Julian Brown <julian@codesourcery.com>
20883 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20885 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20886 type for 1-element load.
20888 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20890 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20892 2017-04-24 Martin Jambor <mjambor@suse.cz>
20894 PR tree-optimization/80293
20895 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20896 char arrays not totally scalarizable if it is false.
20897 (analyze_all_variable_accesses): Pass correct value in the new
20898 parameter. Add a statistics counter.
20900 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20902 PR middle-end/79931
20903 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20905 2017-04-24 Richard Biener <rguenther@suse.de>
20907 PR tree-optimization/80494
20908 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20909 out for complex types.
20911 2017-04-24 Richard Biener <rguenther@suse.de>
20913 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20914 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20915 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20916 (DFS): Adjust and never fail.
20917 (sccvn_dom_walker::fail): Remove.
20918 (sccvn_dom_walker::before_dom_children): Adjust.
20919 (run_scc_vn): Likewise and never fail.
20920 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20921 (pass_fre::execute): Likewise.
20923 2017-04-24 Richard Biener <rguenther@suse.de>
20925 PR tree-optimization/79725
20926 * tree-ssa-sink.c (statement_sink_location): Return whether
20927 failure reason was zero uses. Move that check later.
20928 (sink_code_in_bb): Deal with zero uses by removing the stmt
20931 2017-04-24 Richard Biener <rguenther@suse.de>
20934 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20935 pointer-based references.
20937 2017-04-24 Richard Biener <rguenther@suse.de>
20940 * pass_manager.h (pass_manager::operator new): Remove.
20941 (pass_manager::operator delete): Likewise.
20942 * passes.c (pass_manager::operator new): Remove.
20943 (pass_manager::operator delete): Likewise.
20944 (pass_manager::pass_manager): Zero individual pass members.
20946 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20949 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20950 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20951 Check "XEXP (src, 1)" operand here.
20952 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20953 Check "XEXP (src, 1)" operand here.
20954 (dimode_scalar_chain::make_vector_copies): Detect count register
20955 of a shift instruction. Zero extend count register from QImode
20956 to DImode to satisfy vector shift pattern count operand predicate.
20957 Substitute vector shift count operand with a DImode copy.
20958 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20961 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20963 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20964 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20965 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20966 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20967 (UNSPEC_NOREX_MEM): Remove definition.
20969 2017-04-21 Richard Biener <rguenther@suse.de>
20971 PR tree-optimization/79547
20972 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20973 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20974 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20975 without any constraints.
20977 2017-04-21 Richard Biener <rguenther@suse.de>
20979 PR tree-optimization/78847
20980 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20982 2017-04-21 Richard Biener <rguenther@suse.de>
20984 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20985 (build_distinct_type_copy): Likewise.
20986 (build_variant_type_copy): Likewise.
20987 * tree.c (build_qualified_type): Pass down mem-stat info.
20988 (build_distinct_type_copy): Likewise.
20989 (build_variant_type_copy): Likewise.
20991 2017-04-21 Richard Biener <rguenther@suse.de>
20993 PR tree-optimization/80237
20994 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20996 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20997 for a simplified result.
20999 2016-04-21 Richard Biener <rguenther@suse.de>
21001 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21002 sth as strict as a simple_iv but a chrec without symbols and an
21003 operand defined in the loop we are peeling (and not some subloop).
21004 (propagate_constants_for_unrolling): Propagate all constants.
21006 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21009 * config/i386/i386.c (print_reg): Remove assert for disalowed
21010 regno values, call output_operand_lossage instead.
21012 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21015 * config/i386/constraints.md (Yc): New register constraint.
21016 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21017 Use Yc constraint for alternative 2 of operand 0. Remove
21018 preferred_for_speed attribute.
21020 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21022 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21023 lastprivate clauses in SIMT case.
21025 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21027 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21029 2017-04-20 Richard Biener <rguenther@suse.de>
21031 PR tree-optimization/57796
21032 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21033 as N scalar stores.
21034 (vect_model_load_cost): Cost gathers as N scalar loads.
21036 2017-04-20 Richard Biener <rguenther@suse.de>
21038 * ggc-page.c (ggc_allocated_p): Rename to ...
21039 (safe_lookup_page_table_entry): ... this and return the lookup
21041 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21043 2017-04-20 Richard Biener <rguenther@suse.de>
21045 PR tree-optimization/80453
21046 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21047 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21048 from the conditions.
21049 (vn_phi_eq): Pass them down.
21050 (vn_phi_lookup): Record them.
21051 (vn_phi_insert): Likewise.
21053 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21055 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21056 uninitialized variable warning to avoid buffer overrun.
21058 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21061 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21062 is suppressed for '{ 0 }' in C.
21064 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21066 * BASE-VER: Set to 8.0.0.
21068 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21070 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21071 priority .init_array and .fini_array section with SECTION_NOTYPE
21074 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21076 PR middle-end/80423
21077 * tree.h (build_array_type): Add typeless_storage default argument.
21078 * tree.c (type_cache_hasher::equal): Also compare
21079 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21080 (build_array_type): Add typeless_storage argument, set
21081 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21083 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21084 (build_array_type): Likewise. Add typeless_storage argument.
21086 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21087 Jakub Jelinek <jakub@redhat.com>
21089 PR tree-optimization/80426
21090 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21091 operation on symbolic operands, also compute the overflow for the
21092 invariant part when the operation degenerates into a negation.
21094 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21097 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21098 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21101 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21103 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21106 * config/avr/avr.c (tree.h): Include it.
21107 (cgraph.h): Include it.
21108 (avr_encode_section_info): Don't warn for uninitialized progmem
21109 variable if it's just an alias.
21111 2017-04-19 Richard Biener <rguenther@suse.de>
21114 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21115 when needed by AutoPGO.
21117 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21120 * doc/lto.texi: Remove an extra 'that'.
21122 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21124 PR rtl-optimization/80429
21125 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21126 are only used in debug insns.
21128 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21129 Vladimir Makarov <vmakarov@redhat.com>
21131 * config/sparc/predicates.md (input_operand): Add comment. Return
21132 true for any memory operand when LRA is in progress.
21133 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21135 2017-04-18 Jeff Law <law@redhat.com>
21138 * mips.md ({return,simple_return}_internal): Do not overwrite
21141 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21143 PR tree-optimization/80443
21144 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21145 instead of adding 1, subtract -1 and similarly instead of subtracting
21148 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21150 PR rtl-optimization/80357
21151 * haifa-sched.c (tmp_bitmap): New variable.
21152 (model_recompute): Handle duplicate use records.
21153 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21154 (free_global_sched_pressure_data): Free it.
21156 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21159 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21160 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21161 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21162 instead of SYSTEM_HEADER_DIR.
21164 2017-04-18 Jeff Law <law@redhat.com>
21166 PR middle-end/80422
21167 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21168 predecessors after walking up the insn chain.
21170 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21173 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21174 sizetype type into debug info.
21176 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21179 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21180 unneeded test for TARGET_UPPER_REGS_SF.
21181 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21183 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21186 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21187 instead of gsi_after_labels.
21189 2017-04-18 Jeff Law <law@redhat.com>
21191 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21195 2017-04-13 Jeff Law <law@redhat.com>
21196 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21197 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21199 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21202 * config/avr/avr.c (intl.h): Include it.
21203 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21205 2017-04-18 Martin Liska <mliska@suse.cz>
21207 PR gcov-profile/78783
21208 * gcov-tool.c (gcov_output_files): Validate that destination
21209 file is either removed by the tool or by a user.
21211 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21212 Guy Benyei <guybe@mellanox.com>
21214 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21215 block, and do not negate it, the stored id is already negative.
21217 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21219 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21221 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21224 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21225 masks of options that should be turned off if the VSX vector
21226 options are turned off.
21227 (OTHER_P8_VECTOR_MASKS): Likewise.
21228 (OTHER_VSX_VECTOR_MASKS): Likewise.
21229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21230 rs6000_disable_incompatible_switches to validate no type switches
21232 (rs6000_incompatible_switch): New function to disallow turning on
21233 other vector options if -mno-vsx, -mno-power8-vector, or
21234 -mno-power9-vector are specified.
21236 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21238 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21240 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21242 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21243 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21244 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21245 (ARG_POINTER_CFA_OFFSET): Likewise.
21247 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21249 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21250 conditions to take advantage of various optimizations.
21252 2017-04-13 Jeff Law <law@redhat.com>
21254 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21255 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21256 (zero_extendsidi2_dext): Likewise.
21258 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21261 * fold-const.c (fold_ternary_loc): Revert
21262 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21265 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21267 PR rtl-optimization/80343
21268 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21269 new scratch pseudo.
21271 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21274 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21275 to ubsan_encode_value.
21277 2017-04-13 Jeff Law <law@redhat.com>
21279 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21280 appearing in DEBUG_INSNs.
21282 2017-04-13 Martin Liska <mliska@suse.cz>
21284 PR gcov-profile/80413
21285 * gcov-io.c (gcov_write_string): Copy to buffer just when
21286 allocated size is greater than zero.
21288 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21291 * dwarf2out.c (decls_for_scope): Ignore declarations of
21292 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21294 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21297 * ipa-visibility.c (non_local_p): Fix typos.
21298 (localize_node): When localizing symbol in same comdat group,
21299 dissolve the group only when we know external symbols are going
21301 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21303 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21305 PR tree-optimization/79390
21306 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21307 order does not result in usable sequence, retry with reversed operand
21313 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21314 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21315 op0 instead of fold_convert_loc (loc, type, arg0).
21317 2017-04-12 Jeff Law <law@redhat.com>
21319 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21320 has a delay slot in the generated code.
21322 * config/cris/cris.md (cris_preferred_reload_class): Return
21323 GENNONACR_REGS rather than GENERAL_REGS.
21325 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21328 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21329 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21330 signedness of the result type.
21332 2017-04-12 Richard Biener <rguenther@suse.de>
21333 Jeff Law <law@redhat.com>
21335 PR tree-optimization/80359
21336 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21337 trim stores to TARGET_MEM_REFs.
21339 2017-04-12 Richard Biener <rguenther@suse.de>
21341 PR tree-optimization/79390
21342 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21343 threading case even more.
21345 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21348 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21349 for quad_address_p for TImode, instead of just not indexed_address.
21351 2017-04-12 Richard Biener <rguenther@suse.de>
21352 Bernd Edlinger <bernd.edlinger@hotmail.de>
21354 PR middle-end/79671
21355 * alias.c (component_uses_parent_alias_set_from): Handle
21356 TYPE_TYPELESS_STORAGE.
21357 (get_alias_set): Likewise.
21358 * tree-core.h (tree_type_common): Add typeless_storage flag.
21359 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21360 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21361 for types containing members with TYPE_TYPELESS_STORAGE.
21362 (place_field): Likewise.
21363 (layout_type): Likewise for ARRAY_TYPE.
21364 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21365 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21366 TYPE_TYPELESS_STORAGE.
21367 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21369 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21372 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21373 first argument to type.
21375 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21379 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21380 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21381 (rs6000_expand_binop_builtin): Likewise.
21382 (rs6000_expand_ternop_builtin): Likewise; also add missing
21383 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21384 vshasigma built-ins.
21385 * doc/extend.texi: Document that vec_xxpermdi's third argument
21386 must be a constant.
21388 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21390 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21391 Use shift_const cost parameter when calculating gain of STV shifts.
21393 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21395 PR rtl-optimization/70478
21396 * lra-constraints.c (process_alt_operands): Check memory for
21397 disfavoring memory insn operand.
21399 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21401 PR middle-end/80100
21402 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21403 left shift in unsigned HOST_WIDE_INT type.
21405 PR rtl-optimization/80385
21406 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21407 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21410 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21411 if they have any depend clauses.
21413 2017-04-11 Martin Liska <mliska@suse.cz>
21416 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21417 * ipa-split.c (split_function): Create a local comdat symbol
21418 if caller is in a comdat group.
21420 2017-04-11 Martin Liska <mliska@suse.cz>
21423 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21426 2017-04-11 Martin Sebor <msebor@redhat.com>
21428 PR middle-end/80364
21429 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21430 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21432 (directive::set_width, directive::set_precision, format_character):
21434 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21437 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21440 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21441 conflict, set target->arch_name instead of target->cpu_name.
21443 2017-04-11 Richard Biener <rguenther@suse.de>
21445 PR tree-optimization/80374
21446 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21447 build_zero_cst, remove fold_convertible_p check again.
21449 2017-04-11 Martin Liska <mliska@suse.cz>
21452 * ubsan.c (instrument_object_size): Do not instrument register
21455 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21458 * config/i386/i386-builtin-types.def
21459 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21460 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21461 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21462 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21463 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21464 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21465 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21466 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21467 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21468 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21469 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21470 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21471 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21472 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21473 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21474 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21475 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21476 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21477 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21478 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21479 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21480 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21481 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21482 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21483 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21484 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21485 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21486 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21487 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21488 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21489 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21490 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21491 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21492 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21493 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21494 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21495 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21496 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21497 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21498 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21499 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21500 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21501 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21502 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21503 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21504 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21506 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21507 flag to second_arg_count, handle 4 argument function type _COUNT
21508 aliases, handle second_arg_count on second argument rather than last.
21510 2017-04-10 Jeff Law <law@redhat.com>
21512 PR tree-optimization/80374
21513 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21514 record anything if we can not convert integer_zero_node to the
21517 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21520 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21521 Enhance special handling given to the TARGET_P9_MINMAX option in
21522 relation to certain other options.
21524 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21526 PR tree-optimization/80153
21527 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21528 remove POINTER_PLUS_EXPR's base part directly, rather than through
21531 2017-04-10 Richard Biener <rguenther@suse.de>
21532 Bin Cheng <bin.cheng@arm.com>
21534 PR tree-optimization/80153
21535 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21536 the first element of pointer type aff_tree. Build result expr in
21538 (add_elt_to_tree): Convert to type unconditionally. Remove other
21539 fold_convert calls.
21540 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21541 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21543 2017-04-10 Richard Biener <rguenther@suse.de>
21545 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21548 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21550 PR rtl-optimization/70478
21551 * lra-constraints.c (curr_small_class_check): New.
21552 (update_and_check_small_class_inputs): New.
21553 (process_alt_operands): Update curr_small_class_check. Disfavor
21554 alternative insn memory operands. Check available regs for small
21557 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21560 * config/mips/mips.opt (-mvirt): Update description.
21561 * doc/invoke.texi (-mvirt): Likewise.
21563 2017-04-10 Richard Biener <rguenther@suse.de>
21565 PR middle-end/80362
21566 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21567 looking for NEGATE_EXPR in -A / -B to A / B folding.
21569 2017-04-10 Martin Liska <mliska@suse.cz>
21571 PR gcov-profile/80224
21572 * gcov.c (print_usage): Fix usage string.
21573 (get_gcov_intermediate_filename): Remove.
21574 (output_gcov_file): Use both for normal and intermediate format.
21575 (generate_results): Do not initialize special file for
21576 intermediate format.
21578 2017-04-10 Richard Biener <rguenther@suse.de>
21580 PR tree-optimization/80304
21581 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21584 2017-04-10 Nathan Sidwell <nathan@acm.org>
21587 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21588 (rs6000_init_builtins): Use it.
21590 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21592 * config/arm/arm.md (<mrc>): Add mode to SET source.
21593 (<mrrc>): Likewise.
21595 2017-04-10 Richard Biener <rguenther@suse.de>
21597 PR middle-end/80344
21598 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21600 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21603 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21604 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21605 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21606 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21607 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21608 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21609 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21610 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21611 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21612 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21613 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21614 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21615 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21616 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21617 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21618 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21619 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21620 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21621 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21622 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21623 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21624 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21625 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21627 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21629 PR rtl-optimization/70478
21630 * lra-constraints.c: Reverse the last patch.
21632 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21634 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21635 Add comment for WCHAR_T.
21637 2017-04-08 Martin Liska <mliska@suse.cz>
21640 2017-04-07 Martin Liska <mliska@suse.cz>
21643 * ipa-split.c (split_function): Add function part to a same comdat
21646 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21649 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21651 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21653 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21654 (vec_pairing): ...to this.
21655 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21656 (rs6000_sched_init): Adjust for name change.
21657 (struct rs6000_sched_context): Likewise.
21658 (rs6000_init_sched_context): Likewise.
21659 (rs6000_set_sched_context): Likewise.
21661 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21667 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21669 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21670 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21671 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21673 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21675 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21677 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21679 PR rtl-optimization/70703
21680 * ira-color.c (update_conflict_hard_regno_costs): Use
21681 int64_t instead of HOST_WIDE_INT.
21683 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21685 PR rtl-optimization/70478
21686 * lra-constraints.c (process_alt_operands): Disfavor alternative
21687 insn memory operands.
21689 2017-04-07 Jeff Law <law@redhat.com>
21691 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21692 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21694 2017-04-07 Martin Liska <mliska@suse.cz>
21697 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21698 Show error message instead of an ICE.
21700 2017-04-07 Martin Liska <mliska@suse.cz>
21703 * ipa-split.c (split_function): Add function part to a same comdat
21706 2017-04-07 Richard Biener <rguenther@suse.de>
21708 PR middle-end/80341
21709 * tree.c (get_unwidened): Also handle ! for_type case for
21711 * convert.c (do_narrow): Split out from ...
21712 (convert_to_integer_1): ... here. Do not pass final truncation
21713 type to get_unwidened for TRUNC_DIV_EXPR.
21715 2017-04-07 Richard Biener <rguenther@suse.de>
21717 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21719 (aff_combination_const): Adjust.
21720 (aff_combination_scale): Likewise.
21721 (aff_combination_add_elt): Likewise.
21722 (aff_combination_add_cst): Likewise.
21723 (aff_combination_convert): Likewise.
21724 (add_elt_to_tree): Likewise. Remove unused argument.
21725 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21727 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21729 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21731 * config/arm/arm.c (arm_default_short_enums): Use
21732 ARM_DEFAULT_SHORT_ENUMS.
21733 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21735 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21738 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21739 members with redundant out-of-class redeclaration.
21741 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21744 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21745 * config/i386/i386.md (*zero_extendsidi2):
21746 Add (?*x,*x) and (?*v,*v) alternatives.
21748 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21751 * config/i386/i386.c (ix86_expand_builtin)
21752 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21753 mode from insn data. Convert operands to insn operand mode.
21754 Copy operands that don't satisfy insn predicate to a register.
21756 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21758 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21761 2017-04-06 Richard Biener <rguenther@suse.de>
21763 PR tree-optimization/80334
21764 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21765 preserve alignment of accesses.
21767 2017-04-06 Richard Biener <rguenther@suse.de>
21769 PR tree-optimization/80262
21770 * tree-sra.c (build_ref_for_offset): Preserve address-space
21772 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21773 Drop useless address-space information on MEM_REF offsets.
21775 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21777 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21779 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21781 PR rtl-optimization/70703
21782 * ira-color.c (update_conflict_hard_regno_costs): Use
21783 HOST_WIDE_INT instead of long.
21785 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21788 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21789 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21791 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21792 for x86_64 target. Handle -m3dnowa option.
21794 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21796 PR rtl-optimization/70703
21797 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21798 (update_conflict_hard_regno_costs): Use long instead of unsigned
21799 arithmetic for cost calculation.
21801 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21802 Bernd Edlinger <bernd.edlinger@hotmail.de>
21805 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21808 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21811 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21812 ptr_mode with Pmode throughout.
21813 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21814 into probe_stack_range and use DImode.
21816 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21819 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21820 call_eh_return is true.
21822 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21824 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21825 Initialize last_match_fntype_index.
21827 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21830 * tree-nvr.c: Include internal-fn.h.
21831 (pass_return_slot::execute): Ignore internal calls without
21834 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21835 Richard Biener <rguenther@suse.de>
21838 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21839 captures used multiple times, except for the last use.
21840 * generic-match-head.c: Include gimplify.h.
21842 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21844 PR tree-optimization/79390
21845 * target.h (struct noce_if_info): Declare.
21846 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21847 * target.def (noce_conversion_profitable_p): New target hook.
21848 * ifcvt.h (struct noce_if_info): New type, moved from ...
21849 * ifcvt.c (struct noce_if_info): ... here.
21850 (noce_conversion_profitable_p): Renamed to ...
21851 (default_noce_conversion_profitable_p): ... this. No longer
21853 (noce_try_store_flag_constants, noce_try_addcc,
21854 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21855 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21856 instead of noce_conversion_profitable_p.
21857 * config/i386/i386.c: Include ifcvt.h.
21858 (ix86_option_override_internal): Don't override
21859 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21860 (ix86_noce_conversion_profitable_p): New function.
21861 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21862 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21863 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21864 * doc/tm.texi: Regenerated.
21866 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21868 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21871 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21874 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21875 instructions for small multiply cores.
21877 2017-04-04 Jeff Law <law@redhat.com>
21879 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21881 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21882 that are not set by the loop over the elements.
21884 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21887 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21888 int mode, convert_modes it to mode as unsigned, otherwise use
21889 lowpart_subreg to mode rather than SImode.
21890 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21891 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21892 Use DImode instead of SImode for the shift count operand.
21893 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21896 2017-04-04 Richard Biener <rguenther@suse.de>
21898 PR middle-end/80281
21899 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21900 arithmetic done for the negate or the plus. Simplify.
21901 (A - (-B) -> A + B): Likewise.
21902 * fold-const.c (split_tree): Make sure to not negate pointers.
21904 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21906 PR rtl-optimization/60818
21907 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21908 a compare of comparisons with the thing compared if this results
21909 in a different machine mode.
21911 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21913 * alias.c (base_alias_check): Fix typo in comment.
21914 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21915 * cgraphunit.c (symbol_table::compile): Likewise.
21916 * collect2.c (maybe_run_lto_and_relink): Likewise.
21917 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21918 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21919 * config/avr/avr.c (avr_map_op_t): Likewise.
21920 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21921 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21922 * config/epiphany/epiphany.md (movcc): Likewise.
21923 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21924 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21926 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21927 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21928 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21929 * config/sh/sh.c (sh_rtx_costs): Likewise.
21930 * fold-const.c (fold_truth_andor): Likewise.
21931 * genautomata.c (collapse_flag): Likewise.
21932 * gengtype.h (struct type::u::s): Likewise.
21933 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21934 * input.c (FORMAT_AMOUNT): Likewise.
21935 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21936 (known_aggs_to_agg_replacement_list): Likewise.
21937 * ipa-inline-analysis.c: Likewise.
21938 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21939 * ipa-polymorphic-call.c
21940 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21941 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21942 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21944 * modulo-sched.c (apply_reg_moves): Likewise.
21945 * omp-expand.c (build_omp_regions_1): Likewise.
21946 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21947 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21948 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21949 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21950 * value-prof.c: Likewise.
21951 * var-tracking.c (val_reset): Likewise.
21953 2017-04-03 Richard Biener <rguenther@suse.de>
21955 PR tree-optimization/80275
21956 * fold-const.c (split_address_to_core_and_offset): Handle
21959 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21961 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21962 descriptors is at least equal to that of functions.
21964 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21966 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21968 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21971 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21972 (mov<IMOD4:mode>): New expander.
21973 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21975 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21977 PR rtl-optimization/79405
21978 * fwprop.c (propagations_left): New variable.
21979 (forward_propagate_into): Decrement it.
21980 (fwprop_init): Initialize it.
21981 (fw_prop): If the variable has reached zero, stop propagating.
21982 (fwprop_addr): Ditto.
21984 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21987 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21988 a FUNCTION_DECL, pass it as decl instead of origin to
21991 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21993 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21996 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21999 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22000 TARGET_VSX_SMALL_INTEGER.
22002 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22004 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22005 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22007 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22009 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22010 extraction from odd-numbered MSA register.
22012 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22014 PR middle-end/80173
22015 * expmed.c (store_bit_field_1): Don't attempt to create
22016 a word subreg out of hard registers wider than word if they
22017 have HARD_REGNO_NREGS of 1 for their mode.
22019 PR middle-end/80163
22020 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22021 conversions to integer types wider than word and pointer.
22024 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22025 (rtx_equal_for_cselib_p): Pass 0 to it.
22026 * cselib.c (cselib_hasher::equal): Likewise.
22027 (rtx_equal_for_cselib_1): Add depth argument. If depth
22028 is 128, don't look up VALUE locs and punt. Increment
22029 depth in recursive calls when walking VALUE locs.
22031 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22033 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22034 (make_gcov_file_name): Use the canonical path name for generating
22036 (read_line): Fix handling of files with ascii null bytes.
22038 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22040 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22041 to initialise a vector register instead
22042 of using a const_int.
22044 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22046 PR translation/80189
22047 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22048 diagnostic messages.
22050 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22053 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22054 (dfp_diex_<mode>): Update mode of operand 1.
22055 * doc/extend.texi (dxex, dxexq): Document change to return type.
22056 (diex, diexq): Document change to argument type.
22058 2017-03-30 Martin Jambor <mjambor@suse.cz>
22061 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22062 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22063 it reflects the signature changes performed at the callee side.
22064 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22065 to cgraph_build_function_type_skip_args.
22066 (build_function_decl_skip_args): Adjust call to the above function.
22068 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22071 * config/i386/sse.md
22072 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22073 register as dest whenever it is a MEM not rtx_equal_p to the
22074 corresponding dup operand, and when forcing into reg move the
22075 reg into the memory afterwards.
22076 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22077 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22078 for the force_reg mode.
22079 (avx512vl_vextractf128<mode>): Use register as dest either
22080 always when a MEM, or when it is a MEM not rtx_equal_p to the
22081 corresponding dup operand, or even not when it is a CONST_VECTOR
22082 depending on the mode and lo vs. hi.
22083 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22085 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22086 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22087 Likewise. Require that operands[2] is even.
22088 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22089 Remove extraneous parens. Require that operands[2] is a multiple
22091 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22092 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22093 disallow memory then.
22095 2017-03-30 Richard Biener <rguenther@suse.de>
22097 PR tree-optimization/77498
22098 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22099 to non-constants over backedges.
22101 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22103 PR rtl-optimization/80233
22104 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22105 as last_combined_insn. Do not test for BARRIER_P separately.
22107 2017-03-29 Andreas Schwab <schwab@suse.de>
22110 * calls.c (prepare_call_address): Convert funexp to Pmode before
22111 copying to temp reg.
22113 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22115 PR tree-optimization/80158
22116 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22117 Handle possible future case of more than one alternate
22119 (replace_rhs_if_not_dup): Likewise.
22120 (replace_one_candidate): Likewise.
22122 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22124 PR rtl-optimization/80193
22125 * ira.c (ira): Do not check allocation for LRA.
22127 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22129 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22130 (nvptx_output_simt_exit): Declare.
22131 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22132 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22133 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22134 (nvptx_declare_function_name): ...here. Emit declaration of local
22135 memory space buffer for omp_simt_enter insn.
22136 (nvptx_output_unisimt_switch): New.
22137 (nvptx_output_softstack_switch): New.
22138 (nvptx_output_simt_enter): New.
22139 (nvptx_output_simt_exit): New.
22140 * config/nvptx/nvptx.h (struct machine_function): New fields
22141 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22142 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22143 (UNSPECV_SIMT_EXIT): Ditto.
22144 (omp_simt_enter_insn): New insn.
22145 (omp_simt_enter): New expansion.
22146 (omp_simt_exit): New insn.
22147 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22149 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22150 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22151 (expand_GOMP_SIMT_EXIT): New.
22152 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22153 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22154 (GOMP_SIMT_EXIT): Ditto.
22155 * target-insns.def (omp_simt_enter): New insn.
22156 (omp_simt_exit): Ditto.
22157 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22159 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22160 (lower_rec_input_clauses): Likewise.
22161 (lower_lastprivate_clauses): Handle SIMT privatization.
22163 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22164 (ompdevlow_adjust_simt_enter): New.
22165 (find_simtpriv_var_op): New.
22166 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22167 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22169 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22170 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22171 (copy_decl_for_dup_finish): Ditto.
22173 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22175 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22178 * config/i386/i386.c (ix86_option_override_internal): Always
22179 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22181 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22183 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22185 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22187 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22188 mark new edge's irreducible flag accordign to it.
22189 (vect_do_peeling): Check loop preheader edge's irreducible flag
22190 and pass it to function slpeel_add_loop_guard.
22192 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22194 PR tree-optimization/80218
22195 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22196 Update block frequencies and counts.
22198 2017-03-28 Richard Biener <rguenther@suse.de>
22200 PR tree-optimization/78644
22201 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22202 of a simplification result we may not use it at all.
22204 2017-03-28 Richard Biener <rguenther@suse.de>
22207 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22208 without arguments, generate default definition of a SSA name.
22210 2017-03-28 Richard Biener <rguenther@suse.de>
22212 PR middle-end/80222
22213 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22214 TYPE_REF_CAN_ALIAS_ALL references.
22215 * fold-const.c (fold_indirect_ref_1): Likewise.
22217 2017-03-28 Martin Liska <mliska@suse.cz>
22220 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22221 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22223 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22224 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22226 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22227 (EXTRA_SPECS): Define.
22228 (SUBTARGET_EXTRA_SPECS): Likewise.
22229 (SUBTARGET_CPP_SPEC): Likewise.
22230 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22231 SUBTARGET_EXTRA_SPECS.
22232 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22234 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22236 * config/arc/simdext.md (vst64_insn): Update pattern.
22237 (vld32wh_insn): Likewise.
22238 (vld32wl_insn): Likewise.
22239 (vld64_insn): Likewise.
22240 (vld32_insn): Likewise.
22242 2017-03-28 Marek Polacek <polacek@redhat.com>
22245 * fold-const.c (fold_comparison): Use protected_set_expr_location
22246 instead of SET_EXPR_LOCATION.
22248 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22250 * tree.c (add_expr): Avoid name lookup warning.
22252 2017-03-27 Jeff Law <law@redhat.com>
22254 PR tree-optimization/80216
22255 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22256 function name. Limit recursion depth.
22257 (record_temporary_equivalences): Corresponding changes.
22259 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22261 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22264 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22267 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22269 * cfgcleanup.c (reg_note_cfa_p): New array.
22270 (insns_have_identical_cfa_notes): New function.
22271 (old_insns_match_p): Don't cross-jump in between /f
22272 and non-/f instructions. If both i1 and i2 are frame related,
22273 verify all CFA notes, their order and content.
22275 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22278 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22279 HImode and SImode with zero extend to DImode to one insn.
22280 (bswap<mode>2_extenddi): Likewise.
22281 (bswapsi2_extenddi): Likewise.
22282 (bswaphi2_extendsi): Likewise.
22283 (bswaphi2): Combine bswap HImode and SImode into one insn.
22284 Separate memory insns from swapping register.
22285 (bswapsi2): Likewise.
22286 (bswap<mode>2): Likewise.
22287 (bswaphi2_internal): Delete, no longer used.
22288 (bswapsi2_internal): Likewise.
22289 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22290 store, and gpr<-gpr swap insns.
22291 (bswap<mode>2_store): Likewise.
22292 (bswaphi2_reg): Register only splitter, combine with the splitter.
22293 (bswaphi2 splitter): Likewise.
22294 (bswapsi2_reg): Likewise.
22295 (bswapsi2 splitter): Likewise.
22296 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22297 the insns into load, store, and register/register insns.
22298 (bswapdi2_ldbrx): Likewise.
22299 (bswapdi2_load): Likewise.
22300 (bswapdi2_store): Likewise.
22301 (bswapdi2_reg): Likewise.
22303 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22305 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22306 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22308 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22311 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22313 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22314 special handling for target option conflicts between dform
22315 options (-mpower9-dform, -mpower9-dform-vector,
22316 -mpower9-dform-scalar) and -mno-direct-move.
22318 2017-03-27 Richard Biener <rguenther@suse.de>
22320 PR tree-optimization/80181
22321 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22323 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22325 * config/arc/predicates.md (move_double_src_operand): Replace the
22326 call to move_double_src_operand with a call to address_operand.
22328 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22330 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22331 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22332 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22334 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22336 * config/arc/predicates.md (long_immediate_loadstore_operand):
22337 Consider scaled addresses cases.
22339 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22341 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22342 restored when in interrupt.
22343 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22344 doesn't have delay slot.
22346 2017-03-27 Richard Biener <rguenther@suse.de>
22349 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22350 inlined thunk clones.
22352 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22355 * asan.c (instrument_derefs): Copy over last operand from
22356 original COMPONENT_REF to the new COMPONENT_REF with
22357 DECL_BIT_FIELD_REPRESENTATIVE.
22358 * ubsan.c (instrument_object_size): Likewise.
22360 2017-03-27 Richard Biener <rguenther@suse.de>
22362 PR tree-optimization/80170
22363 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22364 sure DR/SCEV didnt fold in constants we do not see when looking
22365 at the reference base alignment.
22367 2017-03-27 Richard Biener <rguenther@suse.de>
22369 PR middle-end/80171
22370 * gimple-fold.c (fold_ctor_reference): Properly guard against
22371 NULL return value from canonicalize_constructor_val.
22373 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22376 * config/i386/i386.c (ix86_expand_builtin)
22377 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22378 flags reg setting and flags reg using instructions.
22379 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22380 clobbering instructions to zero extend op2.
22382 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22384 * doc/install.texi (Configuration) <--with-aix-soname>:
22385 Update link to AIX ld.
22387 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22389 PR rtl-optimization/80160
22390 PR rtl-optimization/80159
22391 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22392 reg_alternate_class into account.
22394 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22397 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22398 to consider in curr_insn_transform.
22400 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22402 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22403 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22404 and emit_mode_inner.
22406 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22408 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22409 argument to the overloaded builtin variants. Use the new flag to
22410 deprecate certain builtin variants.
22411 * config/s390/s390-builtin-types.def: Add new builtin types.
22412 * config/s390/s390-builtins.h: Support new flags field for
22413 overloaded builtins.
22414 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22415 (s390_macro_to_expand): Enable vector float data type.
22416 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22417 builtins by incrementing the __VEC__ version number.
22418 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22420 (s390_resolve_overloaded_builtin): Emit error messages depending
22421 on the builtin flags.
22422 * config/s390/s390.c (s390_expand_builtin): Support additional
22423 flags argument. Change error message to match the messages
22424 emitted in s390-c.c.
22425 * config/s390/s390.md: New UNSPEC_* constants.
22426 (op_type): Add new instruction types.
22427 * config/s390/vecintrin.h: Add new builtins and test data class
22429 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22430 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22431 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22432 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22433 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22434 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22436 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22437 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22438 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22439 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22441 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22442 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22443 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22444 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22446 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22447 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22448 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22449 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22450 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22451 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22452 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22454 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22455 ("vec_scatter_element<V_HW_4:mode>_DI")
22456 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22457 ("vec_fpint<mode>", "vflls")
22458 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22459 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22460 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22461 ("*vec_cmphe<mode>_cc"): ... these.
22463 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22464 mode constant instead of magic value.
22466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22468 * config/s390/s390.c (s390_expand_vec_compare): Support other
22469 vector floating point modes than just V2DF.
22470 (s390_expand_vcond): Likewise.
22471 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22472 (s390_cannot_change_mode_class): Prevent mode changes between TF
22473 and V1TF in vector registers.
22474 * config/s390/s390.md (DF, SF): New mode attributes.
22475 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22476 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22477 SFmode support for VRs.
22478 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22480 (VFT, VF_HW): New mode iterators.
22481 (vw, sdx): New mode attributes.
22482 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22483 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22484 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22485 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22486 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22487 also the new vector floating point modes. Renaming to ...
22489 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22490 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22491 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22492 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22493 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22494 ("vec_unordered<mode>"): ... these.
22496 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22497 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22498 ("*vec_extendv2df"): New insn definitions.
22500 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22502 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22503 ("mulditi3_2", "*muldi3_sign"): New patterns.
22504 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22505 rename the pattern definition.
22507 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22509 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22511 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22513 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22515 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22516 instruction if possible.
22517 * config/s390/vector.md (vec_halfnumelts): New mode
22519 ("*vec_vllezlf<mode>"): New pattern.
22521 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22523 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22524 ("popcountv4si2", "popcountv2di2"): Rename to ...
22525 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22526 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22528 ("popcount<mode>2_vxe"): New pattern.
22530 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22532 * common/config/s390/s390-common.c (processor_flags_table): Add
22534 * config.gcc: Add arch12.
22535 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22536 Default to arch12 for unknown CPU model numbers.
22537 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22538 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22539 PROCESSOR_max sanity check.
22540 * config/s390/s390-opts.h (enum processor_type): Add
22542 * config/s390/s390.c (processor_table): Add arch12.
22543 (s390_expand_builtin): Add check for B_VXE flag.
22544 (s390_issue_rate): Add PROCESSOR_ARCH12.
22545 (s390_get_sched_attrmask): Likewise.
22546 (s390_get_unit_mask): Likewise.
22547 (s390_sched_score): Enable z13 scheduling for arch12.
22548 (s390_sched_reorder): Likewise.
22549 (s390_sched_variable_issue): Likewise.
22550 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22552 (s390_tune_attr): Use z13 scheduling also for arch12.
22553 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22554 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22555 (TARGET_VXE_P): New macros.
22556 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22557 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22558 * config/s390/s390.opt: Add arch12 as processor_type.
22560 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22562 * config/s390/s390.md
22563 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22564 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22565 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22567 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22568 Rename expanders to ...
22570 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22571 ("fixuns_truncdddi2_emu"): ... these.
22573 ("fixuns_trunc<mode>si2_emu"): New expander.
22575 ("*fixuns_truncdfdi2_z13"): Rename to ...
22576 ("*fixuns_truncdfdi2_vx"): ... this.
22578 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22580 * config/s390/2964.md: Remove the single element vector compare
22581 instructions which are no longer used.
22582 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22584 (s390_canonicalize_comparison): Remove handling of DFmode
22586 (s390_expand_vec_compare_scalar): Remove function.
22587 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22588 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22590 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22592 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22594 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22596 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22597 will anyway by matched by mov<mode>_64dfp.
22599 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22601 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22602 vlef/vstef. Add missing operand to vleif.
22604 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22606 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22607 pair for all vector types with 64 bit elements.
22608 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22609 * config/s390/vector.md (V_HW_64): ... here.
22610 (V_128_NOSINGLE): New mode iterator.
22611 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22612 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22613 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22614 ("*vec_load_pairv2di"): Change to ...
22615 ("*vec_load_pair<mode>"): ... this one.
22617 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22619 * config/s390/constraints.md: Add comments.
22620 (jKK): Reject element sizes > 8 bytes.
22621 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22623 * config/s390/s390.md: Add the s_operand checks formerly in
22624 s390_split_ok_p to various splitters where they are still
22626 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22627 for 128 bit vectors. Plus two splitters.
22629 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22631 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22634 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22637 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22638 error if the boundary argument is not constant.
22640 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22642 PR rtl-optimization/80112
22643 * loop-doloop.c (doloop_condition_get): Don't check condition
22644 if cmp isn't SET with IF_THEN_ELSE src.
22646 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22648 PR tree-optimization/80158
22649 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22650 replacing a candidate statement, also replace it for the
22651 candidate's alternate interpretation.
22652 (replace_rhs_if_not_dup): Likewise.
22653 (replace_one_candidate): Likewise.
22655 2017-03-24 Richard Biener <rguenther@suse.de>
22657 PR tree-optimization/80167
22658 * graphite-isl-ast-to-gimple.c
22659 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22661 (translate_isl_ast_to_gimple::get_rename): Likewise.
22663 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22665 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22666 handling of certain combinations of target options, including the
22667 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22668 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22670 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22673 * config/arm/arm.md (*load_multiple): Add reload_completed to
22674 matching condition.
22676 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22677 Richard Biener <rguenth@suse.de>
22679 PR tree-optimization/79908
22680 PR tree-optimization/80136
22681 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22682 been cast away, gimplify_and_add suffices.
22684 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22686 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22688 2017-03-23 Richard Biener <rguenther@suse.de>
22690 PR tree-optimization/80032
22691 * gimplify.c (gimple_push_cleanup): Forced unconditional
22692 cleanups still have to go to the conditional_cleanups
22695 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22697 PR tree-optimization/80072
22698 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22700 (next_operand_entry_id): Change type to unsigned int.
22701 (sort_by_operand_rank): Make sure to return the right return value
22702 even if unsigned fields are bigger than INT_MAX.
22703 (struct oecount): Change cnt and id type to unsigned int.
22704 (oecount_hasher::equal): Formatting fix.
22705 (oecount_cmp): Make sure to return the right return value
22706 even if unsigned fields are bigger than INT_MAX.
22707 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22710 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22711 TREE_READONLY on result if writing it more than once.
22714 * doc/invoke.texi (-fsanitize=thread): Document that with
22715 -fnon-call-exceptions atomics are not able to throw
22719 * tsan.c: Include tree-eh.h.
22720 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22721 maybe_clean_or_replace_eh_stmt where needed.
22722 (instrument_memory_accesses): Add cfg_changed argument.
22723 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22724 if it returned true.
22725 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22727 PR rtl-optimization/63191
22728 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22729 wrapper function, moved the whole old content into ...
22730 (ix86_delegitimize_address_1): ... this. New inline function.
22731 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22732 true as last argument instead of ix86_delegitimize_address.
22734 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22736 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22737 cortexa57_branch_cost.
22739 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22741 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22743 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22746 * doc/md.texi (Constraints): Document wA constraint.
22747 * config/rs6000/constraints.md (wA): New.
22748 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22749 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22750 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22751 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22753 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22756 * gimplify.c (is_oacc_declared): New function.
22757 (oacc_default_clause): Use it to set default flags for acc declared
22758 variables inside parallel regions.
22759 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22760 declared variables.
22761 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22762 declare attribute to any decl as necessary.
22764 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22767 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22768 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22769 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22770 (arm_arch_lpae): This.
22771 * config/arm/arm.c (arm_arch7ve): Rename into ...
22772 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22773 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22776 2017-03-22 Martin Liska <mliska@suse.cz>
22779 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22780 error message instead of an ICE.
22782 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22784 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22786 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22788 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22790 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22793 2017-03-21 Martin Sebor <msebor@redhat.com>
22795 * doc/extend.texi: Use "cannot" instead of "can't."
22796 * doc/hostconfig.texi: Same.
22797 * doc/install.texi: Same.
22798 * doc/invoke.texi: Same.
22799 * doc/loop.texi: Same.
22800 * doc/md.texi: Same.
22801 * doc/objc.texi: Same.
22802 * doc/rtl.texi: Same.
22803 * doc/tm.texi: Same.
22804 * doc/tm.texi.in: Same.
22805 * doc/trouble.texi: Same.
22807 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22810 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22811 (collect_checksum_attributes): Set it.
22812 (die_checksum_ordered): Use it.
22814 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22816 PR tree-optimization/79908
22817 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22818 change: For a VA_ARG whose LHS has been cast away, use
22819 force_gimple_operand to construct the side effects.
22821 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22823 PR translation/80001
22824 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22825 more amenable to translation.
22826 (oacc_loop_auto_partitions): Likewise.
22828 2017-03-21 Marek Polacek <polacek@redhat.com>
22829 Martin Sebor <msebor@redhat.com>
22831 PR tree-optimization/80109
22832 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22833 on INTEGRAL_TYPE_P.
22835 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22836 Segher Boessenkool <segher@kernel.crashing.org>
22839 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22840 check reg_used_between_p between insn and one of succ or succ2
22841 depending on if succ is artificial insn not inserted into insn
22844 2017-03-21 Martin Liska <mliska@suse.cz>
22846 PR gcov-profile/80081
22847 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22848 * doc/gcc.texi: Include gcov-dump stuff.
22849 * doc/gcov-dump.texi: New file.
22851 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22853 PR rtl-optimization/79150
22854 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22855 conditional jump, if the jump is the last insn of the loop.
22857 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22858 Richard Biener <rguenth@suse.de>
22860 PR tree-optimization/79908
22861 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22862 been cast away, use force_gimple_operand to construct the side
22865 2017-03-21 Martin Liska <mliska@suse.cz>
22867 PR libfortran/79956
22868 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22871 2017-03-21 Brad Spengler <spender@grsecurity.net>
22874 * plugin.c (htab_hash_plugin): New function.
22875 (add_new_plugin): Use it and adjust.
22876 (parse_plugin_arg_opt): Adjust.
22877 (init_one_plugin): Likewise.
22879 2017-03-21 Richard Biener <rguenther@suse.de>
22881 PR tree-optimization/80032
22882 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22883 if set force the cleanup to happen unconditionally.
22884 (gimplify_target_expr): Push inserted clobbers with force_uncond
22885 to avoid them being removed by control-dependent DCE.
22887 2017-03-21 Richard Biener <rguenther@suse.de>
22889 PR tree-optimization/80122
22890 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22891 va_arg_pack_len when the inlined call stmt requires pack
22893 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22895 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22898 * tsan.c (instrument_builtin_call): If the memory model argument
22899 is not a constant, assume it is valid.
22902 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22905 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22907 PR rtl-optimization/79910
22908 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22909 if its dest is used by an insn before I2 (other than the combined
22910 insns themselves, which are properly handled already).
22912 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22915 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22917 * combine.c (record_used_regs): New static function.
22918 (try_combine): Handle situations where there is an additional
22919 instruction between I2 and I3 which needs to have a LOG_LINK
22923 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22925 * combine.c (try_combine): Delete redundant i1 test. Call
22926 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22928 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22931 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22932 alternatives 13/14.
22934 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22936 PR tree-optimization/80054
22937 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22938 the optimization if a PHI or any of its arguments is not dominated
22939 by the candidate's basis. Use gphi* rather than gimple* as
22941 (replace_profitable_candidates): Clean up a gimple* variable that
22942 should be a gphi* variable.
22944 2017-03-20 Martin Sebor <msebor@redhat.com>
22947 * doc/extend.texi (attribute constructor): Document present limitation.
22949 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22952 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22953 __POWER9_VECTOR__ #ifdef control, change template definition to
22954 use Power9-specific built-in function.
22955 (vec_any_eq): Likewise.
22956 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22957 to control outcomes from this test.
22958 (vector_ae_<mode>p): For VEC_F modes, likewise.
22960 2017-03-20 Ian Lance Taylor <iant@google.com>
22962 * config/i386/i386.c (ix86_function_regparm): Save an extra
22963 register for -fsplit-stack with DECL_STATIC_CHAIN.
22965 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22968 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22969 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22971 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22973 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22975 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22978 2017-03-20 Marek Polacek <polacek@redhat.com>
22981 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22983 2017-03-20 Richard Biener <rguenther@suse.de>
22985 PR tree-optimization/80113
22986 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22987 allocate extra SSA name for PHI def.
22988 (add_close_phis_to_outer_loops): Likewise.
22989 (add_close_phis_to_merge_points): Likewise.
22990 (copy_loop_close_phi_args): Likewise.
22991 (copy_cond_phi_nodes): Likewise.
22993 2017-03-20 Martin Liska <mliska@suse.cz>
22995 PR middle-end/79753
22996 * tree-chkp.c (chkp_build_returned_bound): Do not build
22997 returned bounds for a LHS that's not a BOUNDED_P type.
22999 2017-03-20 Martin Liska <mliska@suse.cz>
23003 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23004 COMPLEX_CST and VECTOR_CST.
23006 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23009 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23010 target operand. A new splitter adds the clobber statement in case
23011 the target operand is dead anyway.
23013 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23015 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23016 to age-old versions of binutils and glibc.
23018 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23020 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23022 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23024 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23026 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23028 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23029 requirement for binutils 2.13.
23031 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23033 * combine.c (try_combine): Delete redundant i1 test. Call
23034 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23036 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23038 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23039 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23041 <riscv64-*-elf>: Re-arrange section
23042 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23043 <riscv32-*-linux>: Likewise.
23044 <riscv64-*-elf>: Likewise
23045 <riscv64-*-linux>: Likewise.
23047 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23050 * aarch64.opt(verbose-cost-dump): Fix typo.
23052 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23055 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23056 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23058 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23060 * reload.c (find_reloads): When reloading a nonoffsettable address,
23061 use RELOAD_OTHER for it and its address reloads.
23063 PR rtl-optimization/79910
23064 * combine.c (record_used_regs): New static function.
23065 (try_combine): Handle situations where there is an additional
23066 instruction between I2 and I3 which needs to have a LOG_LINK
23069 2017-03-17 Jeff Law <law@redhat.com>
23071 PR tree-optimization/71437
23072 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23073 conditional in the hash table first.
23074 (vrp_dom_walker::before_dom_children): Extract condition from
23075 ASSERT_EXPR. Record condition, its inverion and any implied
23076 conditions as well.
23078 2017-03-17 Marek Polacek <polacek@redhat.com>
23079 Markus Trippelsdorf <markus@trippelsdorf.de>
23081 PR tree-optimization/80079
23082 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23085 2017-03-17 Richard Biener <rguenther@suse.de>
23087 PR middle-end/80075
23088 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23089 Properly verify the LHS before the RHS possibly claims to be
23091 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23094 2017-03-17 Martin Jambor <mjambor@suse.cz>
23096 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23097 (List of -O2 options): Likewise.
23098 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23101 2017-03-17 Tom de Vries <tom@codesourcery.com>
23103 * gcov-dump.c (print_usage): Print bug_report_url.
23105 2017-03-17 Richard Biener <rguenther@suse.de>
23107 PR middle-end/80050
23108 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23109 (parser::peek): Likewise.
23111 2017-03-17 Richard Biener <rguenther@suse.de>
23113 PR tree-optimization/80048
23114 * sese.c (free_sese_info): Properly release rename_map and
23115 copied_bb_map elements.
23117 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23119 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23120 Add linked-list forward and backlinks. Insert on
23121 construction, remove on destruction.
23122 (class pass_store_merging): Add m_stores_head field.
23123 (pass_store_merging::terminate_and_process_all_chains):
23124 Iterate over m_stores_head list.
23125 (pass_store_merging::terminate_all_aliasing_chains):
23127 (pass_store_merging::execute): Check for debug stmts first.
23128 Push new chains onto the m_stores_head stack.
23130 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23133 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23134 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23135 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23137 2017-03-16 Jeff Law <law@redhat.com>
23139 PR tree-optimization/71437
23140 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23141 member function. Implementation moved into after_dom_children
23142 member function and into the threader's thread_outgoing_edges
23144 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23145 some code into new thread_outgoing_edges.
23146 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23147 definition. Simplify marker handling (do it here). Assume we always
23148 have the available expression and the const/copies tables.
23149 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23151 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23152 * tree-vrp.c (equiv_stack): No longer file scoped.
23153 (vrp_dom_walker): New class.
23154 (vrp_dom_walker::before_dom_children): New member function.
23155 (vrp_dom_walker::after_dom_children): Likewise.
23156 (identify_jump_threads): Setup domwalker. Use it rather than
23157 walking edges in a random order by hand. Simplify setup/finalization.
23158 (finalize_jump_threads): Remove.
23159 (vrp_finalize): Do not call identify_jump_threads here.
23160 (execute_vrp): Do it here instead and call thread_through_all_blocks
23163 PR tree-optimization/71437
23164 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23166 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23168 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23169 (dom_opt_dom_walker::thread_across_edge): Remove
23170 handle_dominating_asserts argument. All callers changed.
23171 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23172 changes. Remove calls to lhs_of_dominating_assert. Other
23173 uses of handle_dominating_asserts turn into unconditional code
23174 (simplify_control_stmt_condition_1): Likewise.
23175 (simplify_control_stmt_condition): Likewise.
23176 (thread_through_normal_block, thread_across_edge): Likewise.
23177 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23178 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23179 object if it is not an SSA_NAME.
23180 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23181 before calling into the VRP specific simplifiers.
23182 (identify_jump_threads): Remove handle_dominating_asserts
23185 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23188 * tree-diagnostic.c (default_tree_printer): No longer static.
23189 * tree-diagnostic.h (default_tree_printer): New prototype.
23191 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23193 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23194 Change ins into fmov.
23196 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23198 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23199 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23200 Use h_con constraint for operand 1.
23201 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23202 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23204 2017-03-15 Jeff Law <law@redhat.com>
23206 PR tree-optimization/71437
23207 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23208 (record_temporary_equivalences): Use it.
23210 PR tree-optimization/71437
23211 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23212 tree-ssa-scopedtables.
23213 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23214 (record_conditions, record_cond, vuse_eq): Likewise.
23215 (record_edge_info): Adjust to API tweak of record_conditions.
23216 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23217 (record_temporary_equivalences, optimize_stmt): Likewise.
23218 (eliminate_redundant_computations): Likewise.
23219 (record_equivalences_from_stmt): Likewise.
23220 * tree-ssa-scopedtables.c: Include options.h and params.h.
23221 (vuse_eq): New function, moved from tree-ssa-dom.c
23222 (build_and_record_new_cond): Likewise.
23223 (record_conditions): Likewise. Accept vector of conditions rather
23224 than edge_equivalence structure for first argument.
23225 for the first argument.
23226 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23227 from tree-ssa-dom.c.
23228 (avail_exprs_stack::record_cond): Likewise.
23229 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23230 from tree-ssa-dom.c.
23231 (avail_exprs_stack): Add new member functions lookup_avail_expr
23233 (record_conditions): Declare.
23235 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23238 * lra-constraints.c (process_alt_operands): Increase reject for
23239 reloading an input/output operand.
23241 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23244 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23245 insns to convert from signed/unsigned char/short to IEEE 128-bit
23247 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23249 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23252 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23253 subreg of inner mode for values already in registers.
23255 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23257 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23258 iteration reg is used after the loop.
23260 2017-03-14 Martin Sebor <msebor@redhat.com>
23262 PR tree-optimization/79800
23263 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23264 precision in negative-positive range.
23265 (format_floating): Call non-const overload with adjusted precision.
23267 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23270 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23273 2017-03-14 Martin Sebor <msebor@redhat.com>
23275 PR middle-end/80020
23276 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23277 * builtins.def (aligned_alloc): Use it.
23280 * Makefile.in (GTFILES): Add calls.c.
23281 * calls.c: Include "gt-calls.h".
23283 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23285 PR rtl-optimization/79728
23286 * regs.h (struct target_regs): New field
23287 x_contains_allocatable_regs_of_mode.
23288 (contains_allocatable_regs_of_mode): New macro.
23289 * reginfo.c (init_reg_sets_1): Initialize it, and change
23290 contains_reg_of_mode so it includes global regs as well.
23291 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23292 rather than contains_regs_of_mode.
23294 2017-03-14 Martin Liska <mliska@suse.cz>
23296 * doc/invoke.texi: Document options that can't be combined with
23297 -fcheck-pointer-bounds.
23299 2017-03-14 Martin Liska <mliska@suse.cz>
23301 PR middle-end/79831
23302 * doc/invoke.texi (-Wchkp): Document the option.
23304 2017-03-14 Martin Liska <mliska@suse.cz>
23306 * Makefile.in: Install gcov-dump.
23308 2017-03-14 Martin Liska <mliska@suse.cz>
23310 * multiple_target.c (expand_target_clones): Bail out for
23311 an invalid attribute.
23313 2017-03-14 Richard Biener <rguenther@suse.de>
23315 * alias.c (struct alias_set_entry): Pack properly.
23316 * cfgloop.h (struct loop): Likewise.
23317 * cse.c (struct set): Likewise.
23318 * ipa-utils.c (struct searchc_env): Likewise.
23319 * loop-invariant.c (struct invariant): Likewise.
23320 * lra-remat.c (struct cand): Likewise.
23321 * recog.c (struct change_t): Likewise.
23322 * rtl.h (struct address_info): Likewise.
23323 * symbol-summary.h (function_summary): Likewise.
23324 * tree-loop-distribution.c (struct partition): Likewise.
23325 * tree-object-size.c (struct object_size_info): Likewise.
23326 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23327 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23328 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23329 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23330 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23331 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23332 (struct _stmt_vec_info): Likewise.
23334 2017-03-14 Martin Liska <mliska@suse.cz>
23337 * multiple_target.c (create_dispatcher_calls): Check that
23338 a target can create a function dispatcher.
23340 2017-03-14 Martin Liska <mliska@suse.cz>
23343 * multiple_target.c (expand_target_clones): Drop local.local
23344 flag for default implementation.
23346 2017-03-14 Richard Biener <rguenther@suse.de>
23348 PR tree-optimization/80030
23349 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23351 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23353 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23354 gcc_fallthrough() instead of __attribute__((fallthrough));
23356 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23358 * doc/gcc.texi: Remove "up" link to (DIR).
23359 * doc/gccint.texi: Ditto.
23361 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23363 * doc/install.texi (Specific) <avr>: Remove reference to
23366 2017-03-13 Jeff Law <law@redhat.com>
23368 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23369 attribute rather than comments.
23371 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23372 match_scratch operand is highest.
23374 2017-03-13 Martin Liska <mliska@suse.cz>
23376 PR middle-end/78339
23377 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23378 is a CHKP clone, use original declaration.
23380 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23382 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23383 (arc_conditional_register_usage): Use a different allocation order
23384 when optimizing for size.
23385 * common/config/arc/arc-common.c (arc_option_optimization_table):
23386 Section anchors default on when optimizing for size.
23388 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23390 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23392 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23394 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23395 * config/arc/arc.md (cpu_facility): Add cd variant.
23396 (*movqi_insn): Add code density variant.
23397 (*movhi_insn): Likewise.
23398 (*movqi_insn): Likewise.
23399 (*addsi3_mixed): Likewise.
23400 (subsi3_insn): Likewise.
23402 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23404 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23406 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23408 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23409 expressions with MINUS and UNARY ops.
23411 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23414 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23416 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23417 between vec_select and vector argument.
23418 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23419 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23420 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23421 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23422 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23423 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23425 2017-03-13 Richard Biener <rguenther@suse.de>
23428 * params.def (vect-max-peeling-for-alignment): Fix typo.
23430 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23432 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23433 issue that only occurred with binutils below 2.18.
23435 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23437 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23438 refer to binutils 2.11/2.12 minimum.
23440 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23442 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23443 ftp.kernel.org and simplify binutils requirement.
23445 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23447 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23449 (Optimize Options): Ditto. Also remove redundancy.
23451 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23453 PR translation/79848
23454 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23456 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23457 to G_ to avoid double translation.
23459 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23461 PR translation/79923
23462 * auto-profile.c (get_combined_location): Convert leading
23463 character of diagnostics to lower case and remove trailing period.
23464 (read_profile): Likewise for various diagnostics.
23465 * config/arm/arm.c (arm_option_override): Remove trailing period
23466 from various diagnostics.
23467 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23468 (msp430_expand_delay_cycles): Likewise.
23470 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23473 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23474 full command-line argument, rather than just "str".
23475 (aarch64_validate_march): Likewise.
23476 (aarch64_validate_mtune): Likewise.
23478 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23480 PR rtl-optimization/78911
23481 * lra-assigns.c (must_not_spill_p): New function.
23482 (spill_for): Use it.
23484 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23486 PR tree-optimization/79981
23487 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23488 ATOMIC_COMPARE_EXCHANGE ifn result.
23489 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23490 IFN_ATOMIC_COMPARE_EXCHANGE.
23492 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23495 * opts.c (parse_sanitizer_options): Add missing question mark to
23496 "did you mean" message.
23498 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23500 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23502 (VMULEUH_UNS): Likewise.
23503 (VMULOUB_UNS): Likewise.
23504 (VMULOUH_UNS): Likewise.
23505 * config/rs6000/rs6000.c (builtin_function_type): Remove
23506 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23508 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23511 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23512 x with result of extra_parsing_for_operand_code_0.
23513 (function_reader::extra_parsing_for_operand_code_0): Convert
23514 return type from void to rtx, returning x. When reading
23515 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23516 larger size containing struct block_symbol.
23518 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23520 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23521 -mfloat128-hardware without -m64.
23523 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23526 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23527 entries to the case statement that marks unsigned arguments to
23528 overloaded functions.
23530 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23532 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23533 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23535 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23538 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23539 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23541 2017-03-10 Martin Liska <mliska@suse.cz>
23545 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23546 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23549 2017-03-10 Olivier Hainque <hainque@adacore.com>
23551 * tree-switch-conversion (array_value_type): Start by resetting
23552 candidate type to it's main variant.
23554 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23556 PR rtl-optimization/79909
23557 * combine.c (try_combine): Use simplify_replace_rtx on individual
23558 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23559 of the whole CALL_INSN_FUNCTION_USAGE.
23561 PR tree-optimization/79972
23562 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23563 get_range_info on SSA_NAMEs. Formatting fixes.
23565 2017-03-10 Richard Biener <rguenther@suse.de>
23566 Jakub Jelinek <jakub@redhat.com>
23568 PR tree-optimization/77975
23569 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23570 edge to be constant.
23571 (get_val_for): For constant x return it. Formatting fix.
23572 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23573 would use the same bases as the current one.
23575 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23577 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23578 instead of vec_select for V1TImode.
23579 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23581 (VSX_LE_128): Add V1TI to this mode iterator.
23582 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23583 (*vsx_le_perm_store_<mode>): Likewise.
23584 (pre-reload splitter for VSX stores): Likewise.
23585 (post-reload splitter for VSX stores): Likewise.
23586 (*vsx_xxpermdi2_le_<mode>): Likewise.
23587 (*vsx_lxvd2x2_le_<mode>): Likewise.
23588 (*vsx_stxvd2x2_le_<mode>): Likewise.
23590 2017-03-09 Michael Eager <eager@eagercon.com>
23592 Correct failures with --enable-checking=yes,rtl.
23594 * config/microblaze/microblaze.c (microblaze_expand_shift):
23595 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23596 test for const0_rtx.
23597 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23598 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23600 2017-03-09 Richard Biener <rguenther@suse.de>
23602 PR tree-optimization/79977
23603 * graphite-scop-detection.c (scop_detection::merge_sese):
23604 Handle the case of extra exits to blocks dominating the entry.
23606 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23608 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23611 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23613 PR rtl-optimization/79949
23614 * lra-constraints.c (process_alt_operands): Check memory when
23615 trying to predict a cycle. Print about the overall increase.
23617 2017-03-09 Richard Biener <rguenther@suse.de>
23619 PR middle-end/79971
23620 * gimple-expr.c (useless_type_conversion_p): Preserve
23621 TYPE_SATURATING for fixed-point types.
23623 2017-03-09 Richard Biener <rguenther@suse.de>
23626 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23627 alignment of BLKmode params.
23629 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23632 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23633 (VALL_NO_V2Q): Likewise.
23635 * config/aarch64/aarch64-simd.md
23636 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23638 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23639 VALL_NO_V2Q mode iterator.
23640 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23642 2017-03-09 Martin Liska <mliska@suse.cz>
23644 PR tree-optimization/79631
23645 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23646 tree_int_cst_sign_bit just for INTEGER constants.
23648 2017-03-09 Martin Liska <mliska@suse.cz>
23652 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23655 2017-03-09 Marek Polacek <polacek@redhat.com>
23658 * tree.c (inchash::add_expr): Handle TREE_VEC.
23660 2017-03-09 Martin Liska <mliska@suse.cz>
23663 (chkp_narrow_size_and_offset): New function.
23664 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23665 (void chkp_parse_bit_field_ref): New function.
23666 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23667 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23669 2017-03-09 Martin Liska <mliska@suse.cz>
23672 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23673 (chkp_find_bounds_1): Remove gcc_unreachable.
23675 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23678 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23679 BUILT_IN_SYNC*, determine the access type from the size suffix and
23680 always build a MEM_REF with that type. Handle forgotten
23681 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23684 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23685 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23686 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23687 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23688 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23689 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23690 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23691 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23692 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23693 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23694 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23695 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23696 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23697 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23698 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23699 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23700 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23701 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23702 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23703 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23704 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23705 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23706 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23707 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23708 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23709 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23710 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23711 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23712 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23713 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23714 definitions outside of __OPTIMIZE__ guarded section.
23717 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23718 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23719 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23720 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23723 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23725 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23726 ("vfenez<mode>"): Add missing constraints.
23728 2017-03-08 Martin Sebor <msebor@redhat.com>
23731 * config/nds32/nds32.c (nds32_option_override):
23732 Fix misspelled diagnostic.
23734 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23737 * gimplify.c (gimplify_omp_for): Replace index var in outer
23738 taskloop statement with an artificial variable and add
23739 OMP_CLAUSE_PRIVATE clause for it.
23741 2017-03-08 Richard Biener <rguenther@suse.de>
23743 PR tree-optimization/79955
23744 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23745 for accesses that are completely outside of the variable.
23747 2017-03-08 Andrew Haley <aph@redhat.com>
23749 PR tree-optimization/79943
23750 * tree-ssa-loop-split.c (compute_new_first_bound): When
23751 calculating the new upper bound, (END-BEG) should be added, not
23754 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23756 * config/avr/avr.md (setmemhi): Make sure match_dup
23757 operand number comes before match_scratch.
23759 2017-03-08 Richard Biener <rguenther@suse.de>
23761 PR tree-optimization/79920
23762 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23763 with ncopies == 1 to ...
23764 (vect_transform_slp_perm_load): ... here. Properly compute
23765 all element loads by iterating VF times over the group. Do
23766 not handle ncopies (computed in a broken way) in
23767 vect_create_mask_and_perm.
23769 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23772 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23773 is a uniform vector, use uniform_vector_p return value instead of
23774 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23776 2017-03-07 Marek Polacek <polacek@redhat.com>
23778 PR middle-end/79809
23779 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23780 (alloca_call_type): Likewise.
23782 2017-03-07 Martin Liska <mliska@suse.cz>
23784 * gcov.c (process_args): Put comment to correct location.
23786 2017-03-07 Martin Liska <mliska@suse.cz>
23788 PR middle-end/68270
23789 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23790 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23791 (chkp_narrow_bounds_for_field): Likewise.
23792 (chkp_parse_array_and_component_ref): Pass one more argument to
23795 2017-03-07 Richard Biener <rguenther@suse.de>
23797 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23800 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23802 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23803 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23805 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23808 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23809 to end of description.
23810 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23812 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23814 PR rtl-optimization/79901
23815 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23817 (*avx512f_<code><mode>3<mask_name>): ... this.
23818 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23819 iterator instead of VI8_AVX2_AVX512BW.
23821 PR rtl-optimization/79901
23822 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23823 min/max expander, expand it using expand_vec_cond_expr.
23826 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23829 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23832 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23833 to void * for PCH reasons.
23834 * dwarf2out.c (output_loc_operands, output_die): Cast
23835 v.val_vec.array to unsigned char *.
23837 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23840 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23843 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23845 PR rtl-optimization/79571
23846 * lra-constraints.c (process_alt_operands): Calculate static
23847 reject and subtract it from overall when only addresses will be
23850 2017-03-06 Julia Koval <julia.koval@intel.com>
23853 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23854 incoming stack boundary to 128 for 64-bit targets.
23856 2017-03-06 Richard Biener <rguenther@suse.de>
23858 PR tree-optimization/79894
23859 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23860 to NULL after folding it.
23862 2017-03-06 Richard Biener <rguenther@suse.de>
23864 PR tree-optimization/79824
23865 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23866 check disabling peeling for gaps.
23868 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23870 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23871 attributes): Document gettimeofday.
23873 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23875 * config/s390/s390.c (s390_option_override_internal): Set
23876 PARAM_MIN_VECT_LOOP_BOUND
23878 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23880 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23881 * config/s390/s390.md: Likewise.
23883 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23886 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23887 (<avx2_avx512>_perm<mode>): Rename to ...
23888 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23890 (<avx512>_perm<mode>_mask): Rename to ...
23891 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23893 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23894 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23895 instead of VI8F_256_512.
23896 (avx512f_perm<mode>): New define_expand.
23897 (avx512f_perm<mode>_mask): Likewise.
23898 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23899 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23901 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23903 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23904 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23906 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23908 2017-03-06 Martin Liska <mliska@suse.cz>
23911 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23912 when having a SSA NAME w/o VAR_DECL assigned to it.
23914 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23916 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23917 msa_dpsub_<su>_d): Fix MODE for vec_select.
23919 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23921 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23923 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23925 2017-03-06 Richard Biener <rguenther@suse.de>
23927 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23928 * plugin.c (register_plugin_info): Likewise.
23929 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23931 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23933 * config/i386/sse.md (sse_storehps, sse_storelps,
23934 avx_<castmode><avxsizesuffix>_<castmode>,
23935 avx512f_<castmode><avxsizesuffix>_<castmode>,
23936 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23937 in condition that at least one operand is not a MEM.
23939 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23941 PR middle-end/79805
23942 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23943 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23945 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23946 gimple_call_nothrow_p flag based on whether original builtin can throw.
23947 If it can, emit following stmts on the fallthrough edge.
23948 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23949 don't create new bb if inserting just debug stmts on the edge, try to
23950 insert them on the fallthru bb or just reset debug stmts.
23952 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23955 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23956 restore recog_data (including the operand rtxes inside it) around
23957 the call to get_insn_template.
23959 2017-03-03 Martin Sebor <msebor@redhat.com>
23961 PR tree-optimization/79699
23962 * context.c (context::~context): Free MPFR caches to avoid
23963 a memory leak on program exit.
23965 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23967 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23968 Use wide_int::ulow () instead of .elt (0).
23970 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23972 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23973 (*pushxf): Limit oF constraint to 32bit targets and add oC
23974 constraint for 64bit targets.
23975 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23976 (*pushdf): Change rmF constraint to rmC.
23978 2017-03-03 Martin Liska <mliska@suse.cz>
23980 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23981 Remove unused variable.
23983 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23986 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23987 is a memory operand, increase num_memory.
23988 (ix86_expand_args_builtin): Likewise.
23990 2017-03-03 Jan Hubicka <jh@suse.cz>
23993 * ipa-devirt.c (maybe_record_node): Properly handle
23994 __cxa_pure_virtual visibility.
23996 2017-03-03 Martin Liska <mliska@suse.cz>
23998 PR tree-optimization/79803
23999 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24001 (pass_loop_prefetch::execute): Disabled optimization if an
24002 assumption about L1 cache size is not met.
24004 2017-03-03 Martin Liska <mliska@suse.cz>
24006 PR rtl-optimization/79574
24007 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24008 (hash_scan_set): Likewise.
24009 (dump_hash_table): Likewise.
24010 (hoist_code): Likewise.
24012 2017-03-03 Richard Biener <rguenther@suse.de>
24014 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24015 in place of uhigh ().
24016 (fixed_convert_from_real): Likewise.
24018 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24021 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24023 2017-03-03 Richard Biener <rguenther@suse.de>
24025 PR middle-end/79818
24026 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24027 TYPE_OVERFLOW_UNDEFINED check.
24029 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24031 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24033 (vector_ae_<mode>_p): Likewise.
24034 (vector_nez_<mode>_p): Likewise.
24035 (vector_ne_v2di_p): Likewise.
24036 (vector_ae_v2di_p): Likewise.
24037 (vector_ne_<mode>_p): Likewise.
24038 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24040 (vsx_tsqrt<mode>2_fe): Likewise.
24042 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24045 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24047 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24049 PR rtl-optimization/79780
24050 * cprop.c (one_cprop_pass): When second and further conditional trap
24051 in a single basic block is turned into an unconditional trap, turn it
24052 into a deleted note to avoid RTL verification failures.
24054 2017-03-02 Richard Biener <rguenther@suse.de>
24056 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24058 2017-03-02 Richard Biener <rguenther@suse.de>
24060 PR tree-optimization/79345
24062 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24063 param and abort the walk, returning -1 if it is hit.
24064 (walk_aliased_vdefs): Take a limit param and pass it on.
24065 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24066 defaulting to 0 and return a signed int.
24067 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24068 (check_defs): New helper.
24069 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24070 about uninitialized memory.
24071 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24072 bogus uninitialized warning.
24073 (fixed_convert_from_real): Likewise.
24075 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24077 PR tree-optimization/66768
24078 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24079 iv_use if base object can't be determined.
24081 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24083 PR tree-optimization/79345
24084 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24085 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24086 (get_pattern_stats): Initialize it.
24087 * genemit.c (gen_expand): Verify match_scratch numbers come after
24088 match_operand/match_dup numbers.
24089 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24090 match_scratch numbers.
24091 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24093 * config/s390/s390.md (trunctdsd2): Likewise.
24095 2017-03-02 Richard Biener <rguenther@suse.de>
24097 * wide-int.h (wide_int_storage::operator=): Implement in terms
24100 2017-03-02 Richard Biener <rguenther@suse.de>
24102 PR tree-optimization/79777
24103 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24104 the to insert expression to sth existing.
24106 2017-03-01 Martin Sebor <msebor@redhat.com>
24108 PR middle-end/79692
24109 * gimple-ssa-sprintf.c
24110 (directive::known_width_and_precision): New function.
24111 (format_integer): Use it.
24112 (get_mpfr_format_length): Consider the full range of precision
24113 when computing %g output with the # flag. Set the likely byte
24114 count to 3 rather than 1 when precision is indeterminate.
24115 (format_floating): Correct the lower bound of precision.
24117 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24119 * doc/invoke.texi: Document default code model for 64-bit Linux.
24121 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24124 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24125 udiv rather than div since input pattern is unsigned.
24127 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24129 * config/i386/i386.c (print_reg): Warn for values of
24130 unsupported size in integer register.
24132 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24135 * config/rs6000/predicates.md (current_file_function_operand): Do
24136 not allow self calls to be local if the function is replaceable.
24138 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24141 * config/rs6000/altivec.h (vec_ctz and others): Change the
24142 preprocessor macro that controls conditional compilation from
24143 _ARCH_PWR9 to __POWER9_VECTOR__.
24144 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24145 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24146 control (instead of _ARCH_PWR9 control) so that template
24147 definition uses power9-specific function.
24148 (vec_any_eq): Likewise.
24149 (vec_all_ne): Change macro definition to use a power9-specific
24150 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24151 _ARCH_PWR9 control).
24152 (vec_any_eq) Likewise.
24153 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24154 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24155 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24156 support for xvcmpnedp instruction.
24157 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24158 macro expansion so that Power9 implementation of vec_all_ne does
24159 not use the AltiVec predicate framework.
24160 (VCMPNEH_P): Likewise.
24161 (VCMPNEW_P): Likewise.
24162 (VCMPNED_P): Likewise.
24163 (VCMPNEFP_P): Likewise.
24164 (VCMPNEDP_P): Likewise.
24165 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24166 implementation of vec_any_eq to not use AltiVec predicate
24168 (VCMPAEH_P): Likewise.
24169 (VCMPAEW_P): Likewise.
24170 (VCMPAED_P): Likewise.
24171 (VCMPAEFP_P): Likewise.
24172 (VCMPAEDP_P): Likewise.
24173 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24174 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24175 not use the AltiVec predicate framework.
24176 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24177 of vec_any_eq to not use AltiVec predicate framework.
24178 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24179 support for predefined __POWER9_VECTOR__ macro to indicate that
24180 Power9 instruction selection is enabled.
24181 (altivec_overloaded_builtins): Remove extraneous
24182 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24183 function argument types RS6000_BTI_bool_V16QI and
24184 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24185 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24186 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24187 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24188 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24189 Power9 for implementations of vec_cmpne. Change the signature for
24190 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24191 (representing vec_all_ne) to remove the previously described first
24192 argument of type RS6000_BTI_INTSI, as this was an artifact of
24193 reliance on the AltiVec predicate framework, which is no longer
24194 used in the implementation of these functions. Add
24195 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24196 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24197 since, unlike the AltiVec predicate framework implementation, we
24198 do not share function descriptors between vec_alle and vec_anyeq.
24199 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24200 set of modes that receive special treatment even when
24201 TARGET_P9_VECTOR is true. The special treatment emits code that
24202 does not depend on Power9 instructions.
24203 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24204 define_expand to not rely on AltiVec predicate framework.
24205 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24207 (vector_ne_v2di_p): Change this define_expand to not rely on
24208 AltiVec predicate framework.
24209 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24211 (vector_ne_<mode>_p): Change this define_expand to not rely on
24212 AltiVec predicate framework.
24213 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24215 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24216 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24217 define_insn pattern.
24218 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24219 define_insn pattern because the xvcmpne<VSs>. instruction is not
24221 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24222 instruction is not supported.
24224 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24226 * config/nvptx/nvptx.c: Include intl.h.
24228 2017-03-01 Martin Jambor <mjambor@suse.cz>
24231 * ipa-prop.h (ipa_bits): Removed field known.
24232 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24233 to pointers. Adjusted their comments to warn about their sharing.
24234 (ipcp_transformation_summary): Change bits to a vector of pointers.
24235 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24236 (ipa_get_ipa_bits_for_value): Declare.
24237 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24238 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24239 (ipa_bits_hash_table): Likewise.
24240 (ipa_vr_ggc_hash_traits): Likewise.
24241 (ipa_vr_hash_table): Likewise.
24242 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24243 being pointers and vr_known being removed.
24244 (ipa_set_jf_unknown): Likewise.
24245 (ipa_get_ipa_bits_for_value): New function.
24246 (ipa_set_jfunc_bits): Likewise.
24247 (ipa_get_value_range): New overloaded functions.
24248 (ipa_set_jfunc_vr): Likewise.
24249 (ipa_compute_jump_functions_for_edge): Use the above functions to
24250 construct bits and vr parts of jump functions.
24251 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24252 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24254 (ipcp_grow_transformations_if_necessary): Also allocate
24255 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24257 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24258 them. Fix too long lines.
24259 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24260 vr_known being removed.
24261 (ipa_read_jump_function): Use new setter functions to construct bits
24262 and vr parts of jump functions or set them to NULL.
24263 (write_ipcp_transformation_info): Adjust for bits being pointers.
24264 (read_ipcp_transformation_info): Likewise.
24265 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24267 Include gt-ipa-prop.h.
24268 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24270 (ipcp_store_bits_results): Likewise.
24271 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24272 Do not write to existing jump functions but use a temporary instead.
24274 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24277 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24278 attempt to use its first operand as BIT_FIELD_REF base.
24280 2017-03-01 Richard Biener <rguenther@suse.de>
24282 PR middle-end/79721
24283 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24284 interpolating formula in wrapping arithmetic.
24285 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24287 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24289 PR tree-optimization/79734
24290 * tree-vect-generic.c (expand_vector_condition): Optimize
24291 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24292 Handle VEC_COND_EXPR where comparison has different inner width from
24293 type's inner width.
24295 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24297 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24298 markup, and similar issues. Remove @opindex entries for things
24299 that aren't options. Add missing -mmpy-option entries.
24301 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24303 PR tree-optimization/79737
24304 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24305 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24306 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24307 instead of byte_size. Formatting fix.
24308 (shift_bytes_in_array_right): Formatting fix.
24310 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24313 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24314 condition on optimize for the leaf function test.
24316 2017-02-28 Martin Liska <mliska@suse.cz>
24319 * read-rtl-function.c (function_reader::handle_unknown_directive):
24320 Bail out when one uses -flto.
24322 2017-02-28 Martin Liska <mliska@suse.cz>
24324 * common.opt: Replace space with tabular for options of <number>
24326 * config/i386/i386.opt: Show <number> value for
24327 -mlarge-data-threshold.
24328 * opts.c (print_filtered_help): Do not display number in hexadecimal
24331 2017-02-28 Martin Liska <mliska@suse.cz>
24333 * common.opt: Fix --help=option -Q for options which are of
24336 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24338 * config/i386/i386.c (print_reg): Error out for values
24339 of 8-bit size in invalid integer register.
24341 2017-02-28 Martin Sebor <msebor@redhat.com>
24343 PR tree-optimization/79691
24344 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24346 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24349 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24350 gcc_unreachable with output_operand_lossage.
24352 2017-02-28 Richard Biener <rguenther@suse.de>
24354 PR tree-optimization/79740
24355 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24357 (visit_nary_op): Insert the nary into the hashtable if we
24358 pattern-matched sth.
24359 * tree-ssa-pre.c (eliminate_insert): Robustify.
24361 2017-02-28 Richard Biener <rguenther@suse.de>
24363 PR middle-end/79731
24364 * fold-const.c (decode_field_reference): Reject out-of-bound
24367 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24369 * config/i386/i386.c: Include intl.h.
24370 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24371 instead of just cond ? "..." : "...".
24372 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24373 * coverage.c (read_counts_file): Likewise.
24374 * omp-offload.c: Include intl.h.
24375 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24376 of just cond ? "..." : "...".
24377 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24378 of just cond ? "..." : "...".
24380 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24383 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24385 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24386 'tune for' CPU name.
24387 * config/arm/arm-cpu-data.h: Regenerated.
24389 2017-02-28 Richard Biener <rguenther@suse.de>
24391 PR tree-optimization/79732
24392 * tree-inline.c (expand_call_inline): Do not shadow var.
24394 2017-02-28 Richard Biener <rguenther@suse.de>
24396 PR tree-optimization/79723
24397 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24398 address-space properly.
24400 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24402 * doc/optinfo.texi (Optimization groups): Fix option used for
24404 * doc/invoke.texi (-fopt-info): Document "omp".
24405 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24406 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24407 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24408 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24409 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24411 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24413 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24414 OPTGROUP_OMP with "omp".
24416 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24419 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24420 for arithmetic shift of unsigned V2DI.
24422 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24424 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24425 arc/linux.h headers.
24426 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24427 (LINK_SPEC): Likewise.
24428 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24429 (EXTRA_SPECS): Likewise.
24430 (STARTFILE_SPEC): Likewise.
24431 (ENDFILE_SPEC): Likewise.
24432 (LIB_SPEC): Likewise.
24433 (TARGET_SDATA_DEFAULT): Likewise.
24434 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24435 (MULTILIB_DEFAULTS): Likewise.
24436 (DWARF2_UNWIND_INFO): Likewise.
24437 * config/arc/big.h: New file.
24438 * config/arc/elf.h: Likewise.
24439 * config/arc/linux.h: Likewise.
24440 * config/arc/t-uClibc: Remove.
24442 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24444 PR tree-optimization/77536
24445 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24446 (tree_transform_and_unroll_loop): Use above function to compute the
24447 estimated niter of unrolled loop and use it when scaling profile.
24448 Also use count info rather than frequency if it's non-zero.
24449 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24450 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24451 (vect_transform_loop): Call above function.
24453 2017-02-27 Richard Biener <rguenther@suse.de>
24455 PR tree-optimization/45397
24456 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24457 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24458 (visit_nary_op): Add pattern matching for CSEing sign-changed
24459 or truncated operations with wider ones.
24461 2017-02-27 Richard Biener <rguenther@suse.de>
24463 PR tree-optimization/79690
24464 * tree-vect-stmts.c (vectorizable_store): Use vector type
24465 built from the DR with address-space.
24467 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24469 * doc/invoke.texi (Optimize Options): Refine the description
24470 of asan-use-after-return.
24472 2017-02-25 Alan Modra <amodra@gmail.com>
24474 PR rtl-optimization/79584
24475 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24476 base, not ad->base_term, the reg within base. Remove assertion
24477 that ad->base == ad->base_term. Replace gen_int_mode using
24478 bogus mode with const0_rtx.
24480 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24482 PR middle-end/79396
24483 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24484 FMA_EXPR like tcc_binary or tcc_unary.
24486 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24489 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24491 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24492 (output_loc_operands): Handle DW_OP_call_ref and
24493 DW_OP_GNU_variable_value.
24494 (struct variable_value_struct): New type.
24495 (struct variable_value_hasher): Likewise.
24496 (variable_value_hash): New variable.
24497 (string_types): Remove.
24498 (copy_loc_descr): New function.
24499 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24500 (prepend_loc_descr_to_each): New function.
24501 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24502 instead of add_loc_descr_to_each if the first argument is single
24503 location list and the second has multiple.
24504 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24505 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24506 when looking for variable value which doesn't have other location info.
24507 (loc_list_from_tree): Formatting fix.
24508 (gen_array_type_die): Simplify DW_AT_string_length handling.
24509 (adjust_string_types): Remove.
24510 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24511 string_types. Call resolve_variable_values.
24512 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24513 (resolve_addr_in_expr): Likewise. Add A argument.
24514 (copy_deref_exprloc): Remove deref argument. Adjust for the
24515 original expression being DW_OP_GNU_variable_value with optionally
24516 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24517 optionally after it.
24518 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24519 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24520 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24521 (variable_value_hasher::hash, variable_value_hasher::equal): New
24523 (resolve_variable_value_in_expr, resolve_variable_value,
24524 resolve_variable_values, note_variable_value_in_expr,
24525 note_variable_value): New functions.
24526 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24529 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24532 * opts.h (handle_generated_option): Add GENERATED_P argument.
24533 * opts-common.c (handle_option): Adjust function comment.
24534 (handle_generated_option): Add GENERATED_P argument, pass it to
24536 (control_warning_option): Pass false to handle_generated_option
24538 * opts.c (maybe_default_option): Pass true to handle_generated_option
24540 * optc-gen.awk: Likewise.
24542 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24544 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24545 a REG, look at the REG it is a SUBREG of.
24546 (splitter for cmpeqsi_t): Ditto.
24548 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24550 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24551 the special USEs with the pattern of the insn, not the insn itself.
24553 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24556 * doc/invoke.texi: Document -mload-store-pairs.
24558 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24559 Sandra Loosemore <sandra@codesourcery.com>
24561 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24562 argument isn't a CONST_INT.
24563 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24564 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24565 (nios2_validate_compare): Bypass alternate compare logic if *op2
24566 is not a CONST_INT.
24567 (ldstwm_operation_p): Return false if first_base is not a REG or
24568 if first_offset is not a CONST_INT.
24570 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24572 * config/cris/cris.md: Use correct operand in a define_peephole2.
24574 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24576 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24578 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24580 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24581 this_insn if it is an INSN or JUMP_INSN.
24582 (force_offsettable): Look at base, not at addr.
24583 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24584 on things that aren't necessarily CONST_INTs.
24586 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24588 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24589 -mfpmath=sse is the default also for x86-32 targets with SSE2
24590 instruction set when @option{-ffast-math} is enabled
24592 2017-02-24 Jeff Law <law@redhat.com>
24594 PR rtl-optimizatoin/79286
24595 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24598 2017-02-24 Richard Biener <rguenther@suse.de>
24600 PR tree-optimization/79389
24601 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24604 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24606 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24607 function comment to reflect reality.
24608 (loop_exits_before_overflow): Fix typo in function description.
24610 2017-02-24 Richard Biener <rguenther@suse.de>
24612 PR tree-optimization/79389
24613 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24614 properly that a threading opportunity exists. Detect conditional
24615 copy/constant propagation opportunities.
24617 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24619 * config/visium/visium.md (type): Add trap.
24620 (b): New mode attribute.
24621 (*btst): Rename into...
24622 (*btst<mode>): ...this and adjust.
24623 (*cbranchsi4_btst_insn): Rename into...
24624 (*cbranch<mode>4_btst_insn): ...this and adjust.
24625 (trap): New define_insn.
24627 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24629 PR tree-optimization/79389
24630 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24631 (noce_reversed_cond_code): New function.
24632 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24633 reversed_comparison_code. Formatting fix.
24634 (noce_try_store_flag): Test rev_cond != NULL in addition to
24635 reversed_comparison_code.
24636 (noce_try_store_flag_constants): Likewise.
24637 (noce_try_store_flag_mask): Likewise.
24638 (noce_try_addcc): Use rev_cond if non-NULL instead of
24639 reversed_comparison_code.
24640 (noce_try_cmove_arith): Likewise. Formatting fixes.
24641 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24642 (noce_find_if_block): Initialize rev_cond.
24643 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24644 instead of false as last argument never attempt to reverse it
24647 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24649 PR tree-optimization/79663
24650 * tree-predcom.c (combine_chains): Process refs in reverse order
24651 only for ZERO length chains, and add explaining comment.
24653 2017-02-23 Jeff Law <law@redhat.com>
24655 PR tree-optimization/79578
24656 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24657 in call to operand_equal_p.
24659 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24662 * config/i386/cpuid.h: Fix another undefined behavior.
24664 2017-02-23 Richard Biener <rguenther@suse.de>
24666 PR tree-optimization/79683
24667 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24668 vector types for data-refs.
24670 2017-02-23 Martin Liska <mliska@suse.cz>
24672 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24674 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24676 PR middle-end/79665
24677 * internal-fn.c (get_range_pos_neg): Moved to ...
24678 * tree.c (get_range_pos_neg): ... here. No longer static.
24679 * tree.h (get_range_pos_neg): New prototype.
24680 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24681 are known to be in between 0 and signed maximum inclusive, try to
24682 expand both unsigned and signed divmod and use the cheaper one from
24685 2017-02-22 Jeff Law <law@redhat.com>
24687 PR tree-optimization/79578
24688 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24689 to compare base operands.
24691 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24694 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24695 gpc_reg_operand instead of fpr_reg_operand.
24697 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24699 * config/mips/mips.c (mips_return_in_memory): Force FP
24700 vector types to be returned in memory for o32 ABI.
24702 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24704 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24705 instead of DW_TAG_member for static data member declarations and don't
24706 set no_linkage_name for static inline data members.
24707 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24710 2017-02-22 Martin Liska <mliska@suse.cz>
24712 * doc/invoke.texi: Replace inequality signs with square brackets
24715 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24717 PR tree-optimization/68644
24718 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24720 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24723 * lra-constraints.c (simplify_operand_subreg): Handle
24724 WORD_REGISTER_OPERATIONS targets.
24726 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24729 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24730 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24731 elimination by swapping fld*.
24733 2017-02-22 Richard Biener <rguenther@suse.de>
24735 PR tree-optimization/79673
24736 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24737 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24738 irrelevant address-space qualifiers and avoiding a
24739 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24741 2017-02-22 Richard Biener <rguenther@suse.de>
24743 PR tree-optimization/79666
24744 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24745 to not symbolically negate if that may introduce undefined
24748 2017-02-22 Martin Liska <mliska@suse.cz>
24751 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24752 * data-streamer-out.c (streamer_write_gcov_count_stream):
24754 * value-prof.c (stream_out_histogram_value): Make assert more
24755 precise based on type of counter.
24757 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24760 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24761 Use nonimmediate_operand instead of memory_operand for operand 1.
24762 (float-extend standard_x87sse_constant_load splitter): Ditto.
24764 2017-02-21 Jeff Law <law@redhat.com>
24766 PR tree-optimization/79621
24767 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24768 blocks with edges to themselves.
24770 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24773 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24774 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24775 Use gimple_call_builtin_p.
24778 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24779 on temporarily removed DEBUG_INSNs.
24781 PR tree-optimization/79649
24782 * tree-loop-distribution.c (classify_partition): Give up on
24783 non-generic address space loads/stores.
24785 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24787 * doc/loop.texi (Loop manipulation): Remove nonexistent
24788 tree_ssa_loop_version from the documentation.
24789 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24791 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24794 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24795 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24796 * config/rs6000/rs6000.c: Include except.h.
24797 (rs6000_expand_split_stack_prologue): Call
24798 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24800 2017-02-21 Martin Jambor <mjambor@suse.cz>
24803 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24804 have been analyzed.
24806 2017-02-21 Martin Jambor <mjambor@suse.cz>
24808 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24809 for backward compatibility only.
24810 * doc/invoke.texi (Option Summary): Remove all references to
24811 -fipa-cp-alignment.
24813 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24817 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24819 * lra-constraints.c (curr_insn_transform): Handle
24820 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24822 2017-02-21 Martin Liska <mliska@suse.cz>
24824 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24826 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24828 PR translation/79638
24829 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24831 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24834 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24835 (arm_function_ok_for_sibcall): Return false for an indirect call by
24836 descriptor if all the argument registers are used.
24837 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24838 alignment of the function.
24840 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24842 PR tree-optimization/61441
24843 * simplify-rtx.c (simplify_const_unary_operation): For
24844 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24845 the sNaN unmodified.
24847 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24849 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24850 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24851 instead of SYSTEM_HEADER_DIR.
24853 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24854 Martin Liška <mliska@suse.cz>
24856 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24857 Fix typos and grammar, use active voice, and clarify.
24859 2017-02-20 Marek Polacek <polacek@redhat.com>
24861 PR middle-end/79537
24862 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24865 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24867 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24870 * config/i386/i386.c (ix86_expand_builtin): Handle
24871 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24872 ix86_builtins_isa[fcode].isa as a requirement of those
24873 flags and any other flag in the bitmask.
24874 (ix86_init_mmx_sse_builtins): Use 0 instead of
24875 ~OPTION_MASK_ISA_64BIT as mask.
24876 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24877 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24878 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24879 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24881 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24884 * lra-constraints.c (split_reg): Check requested split mode
24885 is supported by the register.
24887 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24889 * lra-constraints.c (simplify_operand_subreg): Remove early
24892 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24895 * lra-constraints.c (curr_insn_transform): Tighten condition
24896 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24898 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24901 * lra-constraints.c (curr_insn_transform): Handle
24902 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24904 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24907 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24909 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24911 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24914 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24917 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24919 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24922 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24924 * final.c (last_columnnum, override_columnnum): New variables.
24925 (final_start_function): Set last_columnnum, pass it to begin_prologue
24926 hook and pass 0 to dwarf2out_begin_prologue.
24927 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24928 to source_line debug hook.
24929 (notice_source_line): Compute last_columnnum and for debug_column_info
24930 return true on column changes.
24931 * debug.h (struct gcc_debug_hooks): Add column argument to
24932 source_line and begin_prologue hooks.
24933 (debug_nothing_int_charstar_int_bool): Remove prototype.
24934 (debug_nothing_int_int_charstar,
24935 debug_nothing_int_int_charstar_int_bool): New prototypes.
24936 (dwarf2out_begin_prologue): Add column argument.
24937 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24938 begin_prologue hooks.
24939 (debug_nothing_int_charstar_int_bool): Remove.
24940 (debug_nothing_int_int_charstar,
24941 debug_nothing_int_int_charstar_int_bool): New functions.
24942 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24943 through to dwarf2out_source_line.
24944 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24945 (dwarf2out_source_line): Add column argument, emit it if requested.
24946 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24948 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24949 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24950 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24951 through to dwarf2out_begin_prologue.
24952 (vmsdbgout_source_line): Add column argument, pass it through to
24953 dwarf2out_source_line.
24954 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24955 dbxout_source_line caller.
24956 (dbxout_source_line): Add column argument.
24958 * common.opt (gno-column-info, gcolumn-info): New options.
24959 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24960 (check_die): Also test for multiple DW_AT_decl_column attributes.
24961 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24962 DW_AT_decl_column if requested.
24963 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24965 (gen_variable_die): Likewise.
24966 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24967 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24970 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24971 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24972 (ix86_handle_option): Handle OPT_m3dnowa.
24973 * doc/invoke.texi (-m3dnowa): Document.
24974 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24975 -m3dnowa instead of -m3dnow -march=athlon.
24978 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24979 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24981 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24984 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24985 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24986 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24987 generator for vsx_xxpermdi_<mode>_be.
24988 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24989 force big-endian semantics.
24990 (vsx_xxpermdi_<mode>_be): New define_expand with same
24991 implementation as previous version of vsx_xxpermdi_<mode>.
24993 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24995 PR tree-optimization/79327
24996 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24997 variable, its initialization and use.
24999 2017-02-17 Julia Koval <julia.koval@intel.com>
25001 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25002 (OPTION_MASK_ISA_PKU_UNSET): New.
25003 (ix86_handle_option): Handle -mrdpid.
25004 * config/i386/cpuid.h (bit_RDPID): New.
25005 * config/i386/driver-i386.c (host_detect_local_cpu):
25006 Detect RDPID feature.
25007 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25008 * config/i386/i386-c.c (ix86_target_macros_internal):
25010 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25011 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25012 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25013 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25014 * config/i386/i386.md (define_insn "rdpid"): New.
25015 * config/i386/i386.opt Add -mrdpid.
25016 * config/i386/immintrin.h (_rdpid_u32): New.
25018 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25020 PR rtl-optimization/79541
25021 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25022 instead of transforming it into USE.
25024 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25026 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25027 If HONOR_SNANS (SFmode) force the input to a register.
25028 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25029 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25030 an frsp or similar insn.
25032 2017-02-17 Martin Liska <mliska@suse.cz>
25034 PR rtl-optimization/79577
25035 * params.def (selsched-max-sched-times): Increase minimum to 1.
25037 2017-02-17 Martin Liska <mliska@suse.cz>
25039 PR rtl-optimization/79574
25040 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25042 2017-02-17 Martin Liska <mliska@suse.cz>
25044 PR tree-optimization/79529
25045 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25046 ssa_defined_default_def_p to handle cases which are implicitly
25048 * tree-ssa.c (ssa_defined_default_def_p): New function.
25049 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25050 which are implicitly defined.
25051 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25053 2017-02-17 Richard Biener <rguenther@suse.de>
25055 PR middle-end/79576
25056 * params.def (max-ssa-name-query-depth): Limit to 10.
25058 2017-02-17 Richard Biener <rguenther@suse.de>
25060 PR tree-optimization/79552
25061 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25064 2017-02-17 Richard Biener <rguenther@suse.de>
25067 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25069 2017-02-17 Marek Polacek <polacek@redhat.com>
25071 PR middle-end/79536
25072 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25073 (fold_negate_expr): New wrapper.
25075 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25077 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25078 Correct terminology and de-emphasize pre-standard behavior.
25080 2017-02-16 Alan Modra <amodra@gmail.com>
25082 PR rtl-optimization/79286
25083 * ira.c (def_dominates_uses): New function.
25084 (update_equiv_regs): Don't create an equivalence for insns that
25085 may trap where the register def does not dominate the use.
25087 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25089 PR rtl-optimization/78127
25090 * lra.c (lra): Call lra_eliminate before finish the loop after
25093 2017-02-16 Richard Biener <rguenther@suse.de>
25095 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25097 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25098 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25099 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25100 (isl_val_int_from_wi): New function.
25101 (extract_affine_gmp): Rename to ...
25102 (extract_affine_wi): ... this, take a widest_int.
25103 (extract_affine_int): Just wrap extract_affine_wi.
25104 (add_param_constraints): Use isl_val_int_from_wi.
25105 (add_loop_constraints): Likewise, and extract_affine_wi.
25107 2017-02-15 Jeff Law <law@redhat.com>
25109 PR middle-end/79521
25110 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25111 ira_init_register_move_cost_if_necessary.
25113 2017-02-15 Martin Sebor <msebor@redhat.com>
25115 PR middle-end/32003
25116 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25117 removed in a prior commit.
25119 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25121 PR tree-optimization/79347
25122 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25123 counters during peeling.
25125 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25127 * Makefile.in (site.exp): Remove "set ISLVER".
25129 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25132 * real.c (real_from_integer): Call real_convert even for decimal.
25134 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25137 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25139 2017-02-14 Andrew Pinski <apinski@cavium.com>
25141 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25142 cores and change the partno/implementer to be correct.
25143 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25144 the 'B" as the implementer.
25145 * config/aarch64/aarch64-tune.md: Regenerate.
25147 2017-02-14 Carl Love <cel@us.ibm.com>
25149 * config/rs6000/rs6000.c: Add case statement entry to make the
25150 xvcvuxdsp built-in argument unsigned.
25151 * config/rs6000/vsx.md: Fix the source and return operand types so they
25152 match the instruction definitions from the ISA document. Fix typo
25153 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25156 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25159 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25160 member early_clobber_alts.
25161 * lra-lives.c (reg_early_clobber_p): New.
25162 (process_bb_lives): Use it.
25163 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25164 (debug_operand_data): Initialize early_clobber_alts.
25165 (setup_operand_alternative): Set up early_clobber_alts.
25166 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25167 alternatives to new_insn_reg.
25168 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25170 (lra_update_insn_regno_info): Pass the new arg.
25172 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25174 PR middle-end/79505
25175 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25176 (new_oacc_loop_raw): Don't clear already cleared fields.
25179 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25180 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25181 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25183 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25186 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25188 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25191 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25192 the extra instruction to the right place to store 128-bit constant
25195 2017-02-14 Martin Sebor <msebor@redhat.com>
25197 PR middle-end/79448
25198 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25199 warning for strings of unknown length.
25201 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25203 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25205 2017-02-14 Jeff Law <law@redhat.com>
25208 * ira-costs.c (scan_one_insn): Initialize register move costs
25209 for pseudos seen in USE/CLOBBER insns.
25211 PR tree-optimization/79095
25212 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25213 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25214 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25215 if the operands are known to be not equal, then the resulting range
25217 (intersect_ranges): If the new range is ~[0,0] and the old range is
25218 wide, then prefer ~[0,0].
25219 * tree-vrp.c (overflow_comparison_p_1): New function.
25220 (overflow_comparison_p): New function.
25221 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25222 if NAME is used in an overflow test.
25223 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25224 overflow check that can be expressed as an equality test, then adjust
25225 ops to be that equality test.
25227 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25229 * config/s390/s390-builtin-types.def: Remove flags argument.
25230 * config/s390/s390.c (s390_init_builtins): Likewise.
25232 2017-02-14 Martin Liska <mliska@suse.cz>
25234 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25235 vector. Fix trailing white spaces.
25237 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25239 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25242 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25244 PR rtl-optimization/68664
25245 * config/arm/arm.c (arm_sched_can_speculate_insn):
25246 New function. Declare prototype.
25247 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25249 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25251 PR rtl-optimization/68664
25252 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25254 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25256 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25258 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25259 max skip bytes for function, loop and jump.
25261 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25263 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25264 ABS_EXPR for gimple dump.
25266 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25269 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25271 PR tree-optimization/79408
25272 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25273 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25274 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25275 also if rhs1 is INTEGER_CST.
25277 2017-02-14 Richard Biener <rguenther@suse.de>
25279 PR middle-end/79432
25280 * tree-into-ssa.c (insert_phi_nodes): When the function can
25281 have abnormal edges rewrite SSA names with broken use-def
25282 dominance out of SSA and register them for PHI insertion.
25284 2017-02-13 Martin Sebor <msebor@redhat.com>
25286 PR middle-end/79496
25287 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25288 clearing info.nowrite flag when snprintf size argument is a range.
25290 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25292 * cprop.c (cprop_jump): Add missing space in string literal.
25293 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25294 (get_constraint_for_component_ref): Likewise.
25295 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25296 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25297 * lra-constraints.c (process_alt_operands): Likewise.
25298 * ipa-inline.c (inline_small_functions): Likewise.
25299 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25300 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25301 * trans-mem.c (diagnose_tm_1_op): Likewise.
25302 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25303 (grid_parallel_clauses_gridifiable): Likewise.
25305 * config/nvptx/mkoffload.c (process): Add space in between
25308 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25309 "MOD4_SSE_REGS" and "ALL_REGS".
25311 * spellcheck.c (test_data): Add , in between "foo" and "food".
25313 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25316 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25317 boundary crossing check and subsequent code generation agree.
25319 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25321 * config/aarch64/aarch64.c (has_memory_op): Delete.
25322 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25325 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25327 PR rtl-optimization/79388
25328 PR rtl-optimization/79450
25329 * combine.c (distribute_notes): When removing TEM_INSN for which
25330 corresponding dest has last value recorded, invalidate that last
25333 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25335 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25336 of explicit '@'. Add missing assembly comment marker on branch costs
25339 2017-02-13 Nathan Sidwell <nathan@acm.org>
25341 * gengtype-lex.l (<in_struct>): Add '/'.
25343 2017-02-13 Martin Liska <mliska@suse.cz>
25346 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25348 2017-02-13 Richard Biener <rguenther@suse.de>
25350 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25352 * configure: Re-generate.
25353 * config.in: Likewise.
25354 * graphite-dependences.c: Simplify as if
25355 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25356 * graphite-isl-ast-to-gimple.c: Likewise.
25357 * graphite-optimize-isl.c: Likewise.
25358 * graphite-poly.c: Likewise.
25359 * graphite-sese-to-poly.c: Likewise.
25360 * graphite.h: Likewise.
25361 * toplev.c: Include isl/version.h and use isl_version () for
25362 printing the ISL version.
25363 * doc/install.texi: Update ISL requirement.
25365 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25367 * doc/standards.texi (Standards): Update reference to
25370 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25372 * doc/extend.texi (Named Address Spaces): sourceware.org now
25374 * doc/install.texi (Binaries): Ditto.
25377 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25379 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25380 standard terminology "stringize"/"stringizing" throughout.
25381 * doc/cppinternals.texi: Likewise.
25383 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25385 * doc/extend.texi: Fix some spelling mistakes and typos.
25386 * doc/invoke.texi: Likewise.
25388 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25391 * params.def (inline-min-speedup) Change from 10 to 8.
25393 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25395 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25398 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25401 * ipa-inline-analysis.c (get_minimal_bb): New function.
25402 (record_modified): Use it.
25403 (remap_edge_change_prob): Handle also ancestor functions.
25405 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25407 * doc/contrib.texi (Contributors): Remove broken link into
25408 the Mauve CVS repository.
25410 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25412 PR middle-end/79454
25413 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25414 result computation whenever lhs doesn't have vector mode, not
25415 just when it has BLKmode.
25417 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25419 * doc/makefile.texi (profiledbootstrap): Refer to the
25420 installation instructions only in textual form.
25422 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25425 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25427 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25429 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25430 (Specific): Update mingw-w64 reference.
25432 (Specific): Remove broken link to Renesas RX processor.
25434 2017-02-10 Richard Biener <rguenther@suse.de>
25436 * toplev.c (process_options): Do not mention obsolete graphite
25437 options when printing sorry message about missing graphite support.
25438 Mention -floop-nest-optimize.
25440 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25442 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25443 (vtst_p16): Likewise.
25444 (vtstq_p8): Likewise.
25445 (vtstq_p16): Likewise.
25447 (vtstq_p64): Likewise.
25448 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25449 (vset_lane_p64): New.
25450 (vsetq_lane_p64): New.
25452 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25454 PR tree-optimization/79411
25455 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25456 stmt operands are SSA_NAMEs used in abnormal phis.
25457 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25460 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25463 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25466 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25468 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25470 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25472 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25476 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25477 not arbitrary TREE_CONSTANT.
25480 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25481 "omp declare target link" attribute unless is_global_var.
25482 * omp-offload.c (find_link_var_op): Likewise.
25484 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25485 Chung-Lin Tang <cltang@codesourcery.com>
25487 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25489 (gimplify_adjust_omp_clauses): Don't delete TILE.
25490 (gimplify_omp_for): Deal with TILE.
25491 * internal-fn.c (expand_GOACC_TILE): New function.
25492 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25494 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25495 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25497 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25498 avoid DIV for outermost collapse var.
25499 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25500 Remove out of date comments, fix whitespace.
25501 * omp-general.c (omp_extract_for_data): Deal with tiling.
25502 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25503 adjust OLF_DIM_BASE value.
25504 (struct omp_for_data): Add tiling field.
25505 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25506 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25507 for auto loops. Remove default auto determining, moved to
25508 oacc_loop_fixed_partitions.
25509 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25510 stmts, add e_mask field.
25511 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25512 (oacc_thread_numbers): Use oacc_dim_call.
25513 (oacc_xform_tile): New.
25514 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25515 (finish_oacc_loop): Adjust for ifns vector.
25516 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25517 add case for GOACC_TILE fns.
25518 (oacc_loop_xform_loop): Delete.
25519 (oacc_loop_process): Iterate over call list directly, and add
25520 handling for GOACC_TILE fns.
25521 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25523 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25524 vector partitioning to outer loops. Assign 2 partitions to loops
25525 when available. Add TILE handling.
25526 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25527 (execite_oacc_device_lower): Process GOACC_TILE fns,
25528 ignore unknown specs.
25529 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25530 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25531 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25533 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25535 * configure.ac (ACX_BUGURL): Update.
25536 * configure: Regenerate.
25538 2017-02-09 Richard Biener <rguenther@suse.de>
25540 PR tree-optimization/69823
25541 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25542 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25544 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25546 * config/arc/arc-c.def: Add __NPS400__ definition.
25547 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25548 (TARGET_NPS400): Define.
25550 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25552 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25554 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25555 pointer, arch_info.
25556 (arc_cpu_types): Fill the arch_info field with a pointer into the
25557 arc_arch_types table.
25558 (arc_selected_cpu): Declare.
25559 * config/arc/arc.c (arc_selected_cpu): Make global.
25560 (arc_selected_arch): Delete.
25561 (arc_base_cpu): Delete.
25562 (arc_override_options): Remove references to deleted variables,
25563 update access to arch information.
25564 (ARC_OPT): Update access to arch information.
25565 (ARC_OPTX): Likewise.
25566 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25567 (TARGET_ARC600): Update access to arch information.
25568 (TARGET_ARC601): Likewise.
25569 (TARGET_ARC700): Likewise.
25570 (TARGET_EM): Likewise.
25571 (TARGET_HS): Likewise.
25572 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25575 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25578 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25579 condition/operands for integer GE/LE/GEU/LEU operations.
25581 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25583 PR translation/79397
25584 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25587 2017-02-08 Martin Jambor <mjambor@suse.cz>
25590 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25591 whether allocation happened.
25592 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25593 nothing was allocated.
25595 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25597 PR tree-optimization/79408
25598 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25599 constant, but SSA_NAME with a known integer range, use the minimum
25600 of that range instead of op1 to determine if modulo can be replaced
25601 with its first operand.
25603 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25605 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25607 2017-02-08 Richard Biener <rguenther@suse.de>
25609 PR tree-optimization/71824
25610 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25611 Check all loops contained in the merged region.
25613 2017-02-07 Andrew Pinski <apinski@cavium.com>
25615 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25617 2017-02-07 Andrew Pinski <apinski@cavium.com>
25619 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25620 (thunderxt88): Likewise.
25621 (thunderxt81): Disable LSE and change v8.1 to v8.
25622 (thunderxt83): Likewise.
25624 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25625 Richard Biener <rguenther@suse.de>
25627 PR middle-end/79399
25628 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25629 type from int to size_t.
25630 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25632 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25634 PR rtl-optimization/79386
25635 * cprop.c (bypass_conditional_jumps): Initialize
25636 bypass_last_basic_block already before splitting bbs after
25637 unconditional traps...
25638 (bypass_conditional_jumps): ... rather than here.
25641 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25642 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25643 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25644 fix -masm=intel patterns.
25646 2017-02-07 Richard Biener <rguenther@suse.de>
25648 PR tree-optimization/79256
25649 PR middle-end/79278
25650 * builtins.c (get_object_alignment_2): Use min_align_of_type
25651 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25652 and ADJUST_FIELD_ALIGN.
25654 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25656 * doc/tm.texi: Regenerate.
25657 * stor-layout.c (layout_decl): Adjust.
25658 (update_alignment_for_field): Likewise.
25659 (place_field): Likewise.
25660 (min_align_of_type): Likewise.
25661 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25662 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25663 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25664 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25665 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25666 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25667 * config/i386/i386.c (x86_field_alignment): Likewise.
25668 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25669 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25670 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25671 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25672 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25673 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25677 2017-01-30 Richard Biener <rguenther@suse.de>
25679 PR tree-optimization/79256
25680 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25681 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25684 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25686 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25687 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25688 builtins to SImode and emit a zero-extend, if necessary.
25690 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25692 * docs/invoke.texi (RISC-V Options): Alphabetize.
25694 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25696 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25699 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25701 * config/riscv/riscv.c: New file.
25702 * common/config/riscv/riscv-common.c: Likewise.
25703 * config.gcc: Likewise.
25704 * config/riscv/constraints.md: Likewise.
25705 * config/riscv/elf.h: Likewise.
25706 * config/riscv/generic.md: Likewise.
25707 * config/riscv/linux.h: Likewise.
25708 * config/riscv/multilib-generator: Likewise.
25709 * config/riscv/peephole.md: Likewise.
25710 * config/riscv/pic.md: Likewise.
25711 * config/riscv/predicates.md: Likewise.
25712 * config/riscv/riscv-builtins.c: Likewise.
25713 * config/riscv/riscv-c.c: Likewise.
25714 * config/riscv/riscv-ftypes.def: Likewise.
25715 * config/riscv/riscv-modes.def: Likewise.
25716 * config/riscv/riscv-opts.h: Likewise.
25717 * config/riscv/riscv-protos.h: Likewise.
25718 * config/riscv/riscv.h: Likewise.
25719 * config/riscv/riscv.md: Likewise.
25720 * config/riscv/riscv.opt: Likewise.
25721 * config/riscv/sync.md: Likewise.
25722 * config/riscv/t-elf-multilib: Likewise.
25723 * config/riscv/t-linux: Likewise.
25724 * config/riscv/t-linux-multilib: Likewise.
25725 * config/riscv/t-riscv: Likewise.
25726 * configure.ac: Likewise.
25727 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25728 Waterman as RISC-V maintainers.
25729 * doc/install.texi: Add RISC-V entries.
25730 * doc/invoke.texi: Add RISC-V options section.
25731 * doc/md.texi: Add RISC-V constraints section.
25732 * configure: Regenerated.
25734 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25737 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25738 false values to be constant vectors with all 0 or all 1 bits set.
25739 (vcondu<mode><mode>): Likewise.
25740 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25742 (fpmask_comparison_operator): Update comment.
25743 (vecint_comparison_operator): New predicate.
25744 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25745 vector conditionals when the true and false values are constant
25746 vectors with all 0 bits or all 1 bits set.
25748 2017-02-06 Martin Sebor <msebor@redhat.com>
25750 PR tree-optimization/79376
25751 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25753 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25755 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25756 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25757 to simplify split condition.
25759 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25761 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25762 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25765 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25767 PR rtl-optimization/68664
25768 * target.def (can_speculate_insn): New hook.
25769 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25770 * doc/tm.texi: Regenerate.
25771 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25772 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25773 (rs6000_sched_can_speculate_insn): New function.
25775 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25777 PR tree-optimization/79284
25778 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25779 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25780 vectorizable_mask_load_store, vectorizable_operation,
25781 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25782 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25783 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25784 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25785 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25786 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25787 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25788 is_gimple_assign (stmt). Replace another such test with
25789 is_gimple_assign (stmt).
25791 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25794 * config/avr/avr.c (rtl-iter.h): Include it.
25795 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25796 (avr_legitimate_combined_insn): ...and implementation.
25798 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25800 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25801 * config/s390/s390.c (s390_const_operand_ok)
25802 (s390_canonicalize_comparison, s390_extract_part)
25803 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25804 (s390_contiguous_bitmask_p, s390_rtx_costs)
25805 (legitimize_pic_address): Likewise.
25806 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25807 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25808 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25809 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25810 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25812 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25814 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25815 REGNO($0) == REGNO($1).
25817 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25819 * config/s390/linux.h(SIZE_TYPE): Add comment.
25821 2017-02-06 Julian Brown <julian@codesourcery.com>
25822 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25823 Virendra Pathak <virendra.pathak@broadcom.com>
25825 * config/aarch64/aarch64-cores.def: Change the scheduler
25827 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25828 * config/aarch64/thunderx2t99.md: New file.
25830 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25832 * doc/standards.texi (Go Language): Update link to language
25835 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25837 * tree-eh.c (lower_resx): Sanitize profile.
25838 (cleanup_empty_eh_move_lp): Likewise.
25840 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25843 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25845 * cfgloopmanip.h (loop_version): Update prototype.
25846 * modulo-sched.c (sms_schedule): Update call of loop_version.
25847 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25848 * tree-parloops.c (gen_parallel_loop): Likewise.
25849 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25850 * tree-ssa-loop-split.c (split_loop): Likewise.
25851 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25852 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25854 2017-02-05 Martin Liska <mliska@suse.cz>
25857 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25859 (print_operand_address): Initialize a struct to zero.
25861 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25863 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25864 garbage collector only in textual form.
25866 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25868 * doc/extend.texi (x86 specific memory model extensions for
25869 transactional memory): Simplify a phrase.
25871 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25874 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25875 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25876 (atomic_storedi_1): Likewise.
25878 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25880 PR tree-optimization/79338
25881 * tree-parloops.c (gather_scalar_reductions): Don't call
25882 vect_analyze_loop_form for loop->inner before destroying loop's
25885 2017-02-03 Martin Sebor <msebor@redhat.com>
25887 PR tree-optimization/79327
25888 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25889 when precision has resulted in leading zeros.
25890 (format_integer): Adjust the likely counter to assume an unknown
25891 argument that may be zero is non-zero.
25893 2017-02-03 Jason Merrill <jason@redhat.com>
25896 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25897 avoid copying non-taken branch.
25899 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25901 PR tree-optimization/79340
25902 * tree-vect-loop.c (vectorizable_reduction): Release
25903 vec_defs elements after safe_splicing them into other vectors.
25906 PR tree-optimization/79327
25907 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25908 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25910 (format_integer): Use wide_int_to_tree instead of build_int_cst
25911 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25912 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25913 of shortest and longest sequence.
25915 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25917 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25918 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25920 2017-02-03 Walter Lee <walt@tilera.com>
25923 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25924 after initial stackframe link reg save.
25925 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25927 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25930 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25931 wu for stxssp alternative.
25933 2017-02-03 Martin Sebor <msebor@redhat.com>
25935 PR tree-optimization/79352
25936 * gimple-fold.c (get_range_strlen): Add argument.
25937 (get_range_strlen): Change return type to bool.
25938 (get_maxval_strlen): Pass in a dummy argument.
25939 * gimple-fold.h (get_range_strlen): Change return type to bool.
25940 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25941 * tree.h (array_at_struct_end_p): Add argument.
25942 * tree.c (array_at_struct_end_p): Handle it.
25944 2017-02-03 Martin Liska <mliska@suse.cz>
25947 * multiple_target.c (create_dispatcher_calls): Redirect edge
25948 from a caller of a dispatcher.
25949 (expand_target_clones): Make the clones local.
25950 (ipa_target_clone): Do both target clones and resolvers.
25951 (ipa_dispatcher_calls): Remove the pass.
25952 (pass_dispatcher_calls::gate): Likewise.
25953 (make_pass_dispatcher_calls): Likewise.
25954 * passes.def (pass_target_clone): Put as very first IPA early
25957 2017-02-03 Martin Liska <mliska@suse.cz>
25959 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25960 in case of a function with ifunc attribute.
25962 2017-02-03 Martin Liska <mliska@suse.cz>
25964 * cgraph.c (cgraph_node::dump): Dump function version info.
25965 * symtab.c (symtab_node::dump_base): Add missing new line.
25967 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25969 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25970 (ifcombine_ifandif): Use it.
25972 2017-02-03 Martin Liska <mliska@suse.cz>
25974 * doc/invoke.texi: Document default value for
25975 use-after-scope-direct-emission-threshold.
25977 2017-02-03 Martin Liska <mliska@suse.cz>
25979 PR tree-optimization/79339
25980 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25981 (format_floating): Likewise.
25983 2017-02-03 Martin Liska <mliska@suse.cz>
25986 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25988 (ipa_node_params_t::remove): Likewise.
25989 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25990 initialization from removed ipa_node_params_t::insert.
25991 (ipa_node_params::~ipa_node_params): Move from removed
25992 ipa_node_params_t::release.
25993 * symbol-summary.h (symbol_summary::m_released): New member.
25994 Do not release a summary twice. Do not allow to call finalizer
25995 for types of a summary that live in GGC memory.
25997 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25999 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26002 2017-02-02 Martin Sebor <msebor@redhat.com>
26004 PR middle-end/79275
26005 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26006 (format_string): Tighten up the range of output for non-constant
26007 strings and correct the expected range for wide non-constant strings.
26009 2017-02-02 Martin Sebor <msebor@redhat.com>
26011 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26013 PR middle-end/32003
26014 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26016 (-fdump-tree-@var): Add to index and document how to come up
26017 with pass-specific option and dump file names.
26018 (-fdump-passes): Clarify where to look for output.
26020 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26022 PR middle-end/77445
26023 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26024 statistics of the analyzed path; allow threading for speed when
26025 any of BBs along the path are optimized for speed.
26027 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26029 PR middle-end/78468
26030 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26031 settings of the virtual registers.
26034 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26036 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26037 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26040 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26042 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26043 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26045 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26047 * config/s390/s390.md: Add missing comments with the expanded
26049 * config/s390/vector.md: Likewise.
26050 * config/s390/vx-builtins.md: Likewise.
26052 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26055 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26056 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26057 conditions on a single line.
26059 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26061 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26062 __S390_VX__ to __VX__.
26064 2017-02-01 Andrew Pinski <apinski@cavium.com>
26066 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26067 stmt_info to record_stmt_cost.
26068 (vect_get_known_peeling_cost): Pass stmt_info if known to
26070 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26071 cpu_vector_cost field into
26072 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26073 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26074 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26075 splitting of scalar_stmt_cost and vec_stmt_cost.
26076 (thunderx_vector_cost): Likewise.
26077 (cortexa57_vector_cost): LIkewise.
26078 (exynosm1_vector_cost): Likewise.
26079 (xgene1_vector_cost): Likewise.
26080 (thunderx2t99_vector_cost): Improve after the splitting of the two
26082 (aarch64_builtin_vectorization_cost): Update for the splitting of
26083 scalar_stmt_cost and vec_stmt_cost.
26085 2017-02-01 Torvald Riegel <triegel@redhat.com>
26086 Richard Henderson <rth@redhat.com>
26088 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26089 conditional on existance of a fast atomic load.
26090 * optabs-query.c (can_atomic_load_p): New function.
26091 * optabs-query.h (can_atomic_load_p): Declare it.
26092 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26093 no fast atomic load is available for the particular size of access.
26094 (expand_atomic_compare_and_swap): Likewise.
26095 (expand_atomic_load): Likewise.
26096 (expand_atomic_store): Likewise.
26097 (expand_atomic_fetch_op): Likewise.
26098 * testsuite/lib/target-supports.exp
26099 (check_effective_target_sync_int_128): Remove x86 because it provides
26100 no fast atomic load.
26101 (check_effective_target_sync_int_128_runtime): Likewise.
26103 2017-02-01 Richard Biener <rguenther@suse.de>
26105 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26106 (graphite_transform_loops): Provide opt-info for optimized nests.
26107 * tree-parloop.c (parallelize_loops): Provide opt-info for
26108 parallelized loops.
26110 2017-02-01 Richard Biener <rguenther@suse.de>
26112 PR middle-end/79315
26113 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26114 was not set before.
26116 2017-02-01 Richard Biener <rguenther@suse.de>
26118 PR tree-optimization/71824
26119 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26120 Verify the loops are valid in the merged SESE region.
26121 (scop_detection::can_represent_loop_1): Check analyzing the
26122 evolution of the number of iterations in the region succeeds.
26124 2017-01-31 Ian Lance Taylor <iant@golang.org>
26126 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26127 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26129 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26131 PR preprocessor/79210
26132 * input.c (get_substring_ranges_for_loc): Replace line_width
26133 assertion with error-handling.
26135 2017-01-31 Richard Biener <rguenther@suse.de>
26137 PR tree-optimization/77318
26138 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26139 (create_pw_aff_from_tree): Take loop parameter.
26140 (add_condition_to_pbb): Pass loop of the condition to
26141 create_pw_aff_from_tree.
26143 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26145 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26146 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26148 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26152 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26154 (convert_int_to_float128): Likewise.
26155 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26156 (convert_int_to_float128): Likewise.
26157 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26158 (UNSPEC_IEEE128_CONVERT): Likewise.
26159 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26160 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26161 Use local variables for IBM extended format.
26162 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26163 (fix_trunc<mode>si2_fprs): Likewise.
26164 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26165 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26166 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26167 to know that we can now have integers of all sizes in vector
26169 (fix<uns>_<mode>di2_hw): Likewise.
26170 (float<uns>_<mode>si2_hw): Likewise.
26171 (fix_<mode>si2_hw): Likewise.
26172 (fixuns_<mode>si2_hw): Likewise.
26173 (float<uns>_<mode>di2_hw): Likewise.
26174 (float_<mode>di2_hw): Likewise.
26175 (float_<mode>si2_hw): Likewise.
26176 (floatuns_<mode>di2_hw): Likewise.
26177 (floatuns_<mode>si2_hw): Likewise.
26178 (xscvqp<su>wz_<mode>): Delete, no longer used.
26179 (xscvqp<su>dz_<mode>): Likewise.
26180 (xscv<su>dqp_<mode>): Likewise.
26181 (ieee128_mfvsrd_64bit): Likewise.
26182 (ieee128_mfvsrd_32bit): Likewise.
26183 (ieee128_mfvsrwz): Likewise.
26184 (ieee128_mtvsrw): Likewise.
26185 (ieee128_mtvsrd_64bit): Likewise.
26186 (ieee128_mtvsrd_32bit): Likewise.
26188 2017-01-31 Martin Liska <mliska@suse.cz>
26191 * ipa-prop.c (ipa_free_all_node_params): Call release method
26192 instead of ~sumbol_summary to not to trigger double times
26195 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26197 PR tree-optimization/71691
26198 * bitmap.h (class auto_bitmap): New.
26199 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26200 is_maybe_undefined instead of ssa_undefined_value_p.
26202 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26204 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26205 __S390_ARCH_LEVEL__ to __ARCH__.
26207 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26209 PR tree-optimization/79267
26210 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26211 if should_remove_lhs_p is true.
26213 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26216 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26217 (add_alignment_attribute): New.
26218 (base_type_die): Add alignment attribute.
26219 (subrange_type_die): Likewise.
26220 (modified_type_die): Likewise.
26221 (gen_array_type_die): Likewise.
26222 (gen_descr_array_type_die: Likewise.
26223 (gen_enumeration_type_die): Likewise.
26224 (gen_subprogram_die): Likewise.
26225 (gen_variable_die): Likewise.
26226 (gen_field_die): Likewise.
26227 (gen_ptr_to_mbr_type_die): Likewise.
26228 (gen_struct_or_union_type_die): Likewise.
26229 (gen_subroutine_type_die): Likewise.
26230 (gen_typedef_die): Likewise.
26231 (base_type_cmp): Compare alignment attribute.
26233 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26236 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26237 (setb_unsigned) New pattern for setb with CCUNS.
26238 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26239 subfc./subfe sequence to avoid overflow problems. Generate a
26240 shorter sequence with cmpld/setb for power9.
26241 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26242 for generating subfc. instruction.
26243 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26244 now uses this instruction.
26246 2017-01-30 Ian Lance Taylor <iant@google.com>
26249 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26250 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26252 2017-01-30 Martin Sebor <msebor@redhat.com>
26254 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26255 Move constant to the right of a relational operator.
26256 (get_mpfr_format_length, format_character, format_string): Ditto.
26257 (should_warn_p, maybe_warn): Same.
26259 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26261 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26264 * asan.c (get_translation_unit_decl): Remove function.
26265 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26267 2017-01-30 Martin Liska <mliska@suse.cz>
26269 PR gcov-profile/79259
26270 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26271 -fprofile-generate.
26273 2017-01-30 Martin Liska <mliska@suse.cz>
26276 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26277 Initialize variables with NULL value.
26279 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26282 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26284 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26286 2017-01-30 Richard Biener <rguenther@suse.de>
26288 PR tree-optimization/79276
26289 * tree-vrp.c (process_assert_insertions): Properly adjust common
26290 when removing a duplicate.
26292 * gcc.dg/torture/pr79276.c: New testcase.
26294 2017-01-30 Richard Biener <rguenther@suse.de>
26296 PR tree-optimization/79256
26297 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26298 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26300 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26302 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26305 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26306 ("*r<noxa>sbg_<mode>_sll_bitmask")
26307 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26308 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26309 Use contiguous_bitmask_nowrap_operand.
26311 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26314 * config/rs6000/altivec.h (vec_xl): Revise #define.
26315 (vec_xst): Likewise.
26317 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26319 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26321 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26323 PR rtl-optimization/79194
26324 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26325 traps before call to bypass_conditional_jumps.
26327 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26329 PR tree-optimization/71374
26330 * lra-constraints.c (check_conflict_input_operands): New.
26331 (match_reload): Use it.
26333 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26336 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26337 account to calculate conflict_set.
26339 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26341 PR rtl-optimization/78559
26342 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26343 other_insn in combine.
26345 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26347 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26348 uint16_type_node for BT_UINT16.
26350 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26352 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26353 "RTL Tests" to menu.
26354 (GIMPLE Tests): New node.
26355 (RTL Tests): New node.
26357 2017-01-27 Richard Biener <rguenther@suse.de>
26359 PR tree-optimization/79245
26360 * tree-loop-distribution.c (distribute_loop): Apply cost
26361 modeling also to detected patterns.
26363 2017-01-27 Richard Biener <rguenther@suse.de>
26365 PR tree-optimization/71433
26366 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26367 (compare_assert_loc): New function.
26368 (process_assert_insertions): Sort and optimize assert locations
26369 to remove duplicates and push down identical assertions on
26370 edges to their destination block.
26372 2017-01-27 Richard Biener <rguenther@suse.de>
26374 PR tree-optimization/79244
26375 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26376 out SSA names even if abnormal.
26378 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26380 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26381 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26382 instead of MPFR_RNDN.
26384 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26387 * arm.c (arm_option_override): Don't call build_target_option_node
26388 until after doing all option overrides.
26389 (arm_valid_target_attribute_tree): Likewise.
26391 2017-01-27 Martin Liska <mliska@suse.cz>
26393 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26394 for {cd}tors and C++ {cd}tors.
26396 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26398 * config/s390/s390.md ("*setmem_long_and")
26399 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26401 2017-01-26 Martin Sebor <msebor@redhat.com>
26403 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26406 2017-01-26 Martin Sebor <msebor@redhat.com>
26408 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26409 HAVE_DFmode before using XFmode or DFmode.
26410 (parse_directive): Avoid using the z length modifier to avoid
26411 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26413 PR middle-end/78703
26414 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26415 to accept adjustment as an array.
26416 (get_int_range): New function.
26417 (struct directive): Make width and prec arrays.
26418 (directive::set_width, directive::set_precision): Call get_int_range.
26419 (format_integer, format_floating): Handle width and precision ranges.
26420 (format_string, parse_directive): Same.
26422 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26425 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26426 comdat_type_p set, just clone them, but keep the children in the
26430 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26431 which have direct callers with -fvar-tracking-assignments enabled
26433 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26434 inside of type units.
26436 2017-01-26 Martin Sebor <msebor@redhat.com>
26438 PR middle-end/78703
26439 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26441 (struct format_result): Replace number_chars, number_chars_min,
26442 and number_chars_max with a single member of struct result_range.
26444 (format_result::operator+=): Adjust.
26445 (struct fmtresult): Remove bounded. Handle likely and unlikely
26447 (fmtresult::adjust_for_width_or_precision): New function.
26448 (fmtresult:type_max_digits): New function.
26449 (bytes_remaining): Handle likely and unlikely counters.
26450 (min_bytes_remaining): Remove.
26451 (format_percent): Simplify.
26452 (format_integer, format_floating): Set likely and unlikely counters.
26453 (get_string_length, format_character, format_string): Same.
26454 (format_plain, should_warn_p): New function.
26455 (maybe_warn): Call should_warn_p. Update diagnostic messages
26456 and handle those for all directives, including plain strings.
26457 (format_directive): Handle likely and unlikely counters.
26458 Remove unnecessary quoting from diagnostics. Add an informational
26460 (add_bytes): Remove.
26461 (pass_sprintf_length::compute_format_length): Simplify.
26462 (try_substitute_return_value): Handle likely and unlikely counters.
26464 2017-01-26 Carl Love <cel@us.ibm.com>
26466 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26467 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26469 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26472 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26473 endianess for subregs into account.
26474 * lra-constraints.c (lra_constraints): Do risky transformations
26475 always on the first iteration.
26476 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26477 last_call_used_reg_set.
26478 (process_bb_lives): Define and use last_call_used_reg_set.
26479 * lra.c (lra): Always continue after lra_constraints on the first
26482 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26484 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26486 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26488 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26490 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26491 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26492 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26493 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26494 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26495 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26496 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26497 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26498 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26500 2017-01-26 Marek Polacek <polacek@redhat.com>
26503 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26504 for the third operand.
26506 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26508 PR middle-end/79236
26509 * omp-low.c (struct omp_context): Add simt_stmt field.
26510 (scan_omp_for): Return omp_context *.
26511 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26512 context to the _simt_ SIMD stmt.
26513 (lower_omp_for): For combined SIMD with sibling _simt_
26514 SIMD, make sure to use the same decls in _looptemp_
26515 clauses as in the sibling.
26517 2017-01-26 David Sherwood <david.sherwood@arm.com>
26519 PR middle-end/79212
26520 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26523 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26526 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26527 emit fld b; fld a; if possible.
26529 * brig-builtins.def: Update copyright years.
26530 * config/arm/arm_acle_builtins.def: Update copyright years.
26532 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26535 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26536 constraint instead of o for the stxsd instruction.
26538 2017-01-25 Carl Love <cel@us.ibm.com>
26540 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26541 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26543 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26545 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26547 2017-01-25 Richard Biener <rguenther@suse.de>
26549 PR tree-optimization/69264
26550 * target.def (vector_alignment_reachable): Improve documentation.
26551 * doc/tm.texi: Regenerate.
26552 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26554 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26555 earlier changes with respect to TYPE_USER_ALIGN.
26556 (vector_alignment_reachable_p): Likewise. Improve dumping.
26558 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26561 * config/arm/arm.md (xordi3): Force constant operand into a register
26564 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26566 * doc/invoke.texi (-fstore-merging): Correct default optimization
26567 levels at which it is enabled.
26568 (-O): Move -fstore-merging from list to...
26571 2017-01-25 Richard Biener <rguenther@suse.de>
26574 * omp-expand.c: Include debug.h.
26575 (expand_omp_taskreg): Make sure to generate early debug before
26576 outlining anything from a function.
26577 (expand_omp_target): Likewise.
26578 (grid_expand_target_grid_body): Likewise.
26580 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26583 * asan.c (get_translation_unit_decl): New function.
26584 (asan_add_global): Extract modules file name from globals
26585 TRANSLATION_UNIT_DECL name.
26587 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26590 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26591 for long calls with APCS frame and VFP.
26593 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26595 * cfg.c (original_copy_tables_initialized_p): New function.
26596 * cfg.h (original_copy_tables_initialized_p): New decl.
26597 * cfgrtl.c (relink_block_chain): Guard the call to
26598 free_original_copy_tables with a call to
26599 original_copy_tables_initialized_p.
26600 * cgraph.h (symtab_node::native_rtl_p): New decl.
26601 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26602 (symtab_node::needed_p): Don't assert for early assembly output
26603 for __RTL functions.
26604 (cgraph_node::finalize_function): Set "force_output" for __RTL
26606 (cgraph_node::analyze): Bail out early for __RTL functions.
26607 (analyze_functions): Update assertion to support __RTL functions.
26608 (cgraph_node::expand): Bail out early for __RTL functions.
26609 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26611 * function.h (struct function): Update comment for field
26613 * gimple-expr.c: Include "tree-pass.h".
26614 (gimple_has_body_p): Return false for __RTL functions.
26615 * Makefile.in (OBJS): Add run-rtl-passes.o.
26616 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26618 (gcc::pass_manager::get_clean_slate): New accessor.
26619 * passes.c: Include "insn-addr.h".
26620 (should_skip_pass_p): Add logging. Update logic for running
26621 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26622 property-provider override so it is only done for gimple passes.
26624 (skip_pass): New function.
26625 (execute_one_pass): Call skip_pass when skipping passes.
26626 * read-md.c (md_reader::read_char): Support filtering
26627 the input to a subset of line numbers.
26628 (md_reader::md_reader): Initialize fields
26629 m_first_line and m_last_line.
26630 (md_reader::read_file_fragment): New function.
26631 * read-md.h (md_reader::read_file_fragment): New decl.
26632 (md_reader::m_first_line): New field.
26633 (md_reader::m_last_line): New field.
26634 * read-rtl-function.c (function_reader::create_function): Only
26635 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26636 curr_properties. Set DECL_INITIAL to a dummy block.
26637 (read_rtl_function_body_from_file_range): New function.
26638 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26640 * run-rtl-passes.c: New file.
26641 * run-rtl-passes.h: New file.
26643 2017-01-24 Jeff Law <law@redhat.com>
26645 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26648 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26650 PR tree-optimization/79159
26651 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26652 (record_nonwrapping_iv): Improve boundary using above function if no
26653 value range information.
26655 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26656 Martin Jambor <mjambor@suse.cz>
26658 * brig-builtins.def: New file.
26659 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26660 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26661 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26662 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26663 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26664 * builtin-types.def (BT_INT8): New.
26665 (BT_INT16): Likewise.
26666 (BT_UINT8): Likewise.
26667 (BT_UINT16): Likewise.
26668 (BT_FN_ULONG): Likewise.
26669 (BT_FN_UINT_INT): Likewise.
26670 (BT_FN_UINT_ULONG): Likewise.
26671 (BT_FN_UINT_LONG): Likewise.
26672 (BT_FN_UINT_PTR): Likewise.
26673 (BT_FN_ULONG_PTR): Likewise.
26674 (BT_FN_INT8_FLOAT): Likewise.
26675 (BT_FN_INT16_FLOAT): Likewise.
26676 (BT_FN_UINT32_FLOAT): Likewise.
26677 (BT_FN_UINT16_FLOAT): Likewise.
26678 (BT_FN_UINT8_FLOAT): Likewise.
26679 (BT_FN_UINT64_FLOAT): Likewise.
26680 (BT_FN_UINT16_UINT32): Likewise.
26681 (BT_FN_UINT32_UINT16): Likewise.
26682 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26683 (BT_FN_INT_PTR_INT): Likewise.
26684 (BT_FN_UINT_PTR_UINT): Likewise.
26685 (BT_FN_LONG_PTR_LONG): Likewise.
26686 (BT_FN_ULONG_PTR_ULONG): Likewise.
26687 (BT_FN_VOID_UINT64_UINT64): Likewise.
26688 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26689 (BT_FN_INT8_INT8_INT8): Likewise.
26690 (BT_FN_INT16_INT16_INT16): Likewise.
26691 (BT_FN_INT_INT_INT): Likewise.
26692 (BT_FN_UINT_FLOAT_UINT): Likewise.
26693 (BT_FN_FLOAT_UINT_UINT): Likewise.
26694 (BT_FN_ULONG_UINT_UINT): Likewise.
26695 (BT_FN_ULONG_UINT_PTR): Likewise.
26696 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26697 (BT_FN_UINT_UINT_UINT): Likewise.
26698 (BT_FN_VOID_UINT_PTR): Likewise.
26699 (BT_FN_UINT_UINT_PTR: Likewise.
26700 (BT_FN_UINT32_UINT64_PTR): Likewise.
26701 (BT_FN_INT_INT_UINT_UINT): Likewise.
26702 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26703 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26704 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26705 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26706 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26707 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26708 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26709 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26710 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26711 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26712 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26713 * doc/frontends.texi: List BRIG FE.
26714 * doc/install.texi (Testing): Add BRIG tesring requirements.
26715 * doc/invoke.texi (Overall Options): Mention BRIG.
26716 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26718 2017-01-24 Richard Biener <rguenther@suse.de>
26720 PR translation/79208
26721 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26723 2017-01-24 Martin Jambor <mjambor@suse.cz>
26726 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26727 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26728 and known_contexts.
26730 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26732 PR middle-end/79123
26733 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26734 casts from signed to unsigned really don't have a range.
26736 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26738 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26739 GMP_RNDx for compatiblity.
26741 2017-01-24 Martin Liska <mliska@suse.cz>
26744 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26745 that would prevent us to call alloca with -1 as argument.
26747 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26749 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26750 Avoid -Wformat-security warning.
26752 2017-01-23 Andrew Pinski <apinski@cavium.com>
26754 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26757 2017-01-23 Martin Sebor <msebor@redhat.com>
26759 PR middle-end/78703
26760 * gimple-ssa-sprintf.c (warn_level): New global.
26761 (format_integer): Use it here and throughout the rest of the file.
26762 Use the same switch to compute sign as base.
26763 (maybe_warn): New function.
26764 (format_directive): Factor out warnings into maybe_warn.
26765 Add debugging output. Use warn_level.
26766 (add_bytes): Use warn_level.
26767 (pass_sprintf_length::compute_format_length): Add debugging output.
26768 (try_substitute_return_value): Same.
26769 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26771 PR middle-end/78703
26772 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26773 (struct fmtresult, format_integer, format_floating): Adjust.
26774 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26775 (get_string_length, format_string,format_directive): Same.
26776 (pass_sprintf_length::compute_format_length): Same.
26777 (try_substitute_return_value): Simplify slightly.
26779 PR middle-end/78703
26780 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26781 (fmtresult::operator+=): Outlined.
26782 (struct fmtresult): Add ctors.
26783 (struct conversion_spec): Rename...
26784 (struct directive): ...to this. Add and remove data members.
26785 (directive::set_width, directive::set_precision): New functions.
26786 (format_percent): Use fmtresult ctor.
26787 (get_width_and_precision): Remove.
26788 (format_integer): Make naming changes. Avoid computing width and
26790 (format_floating): Same. Adjust indentation.
26791 (format_character, format_none): New functions.
26792 (format_string): Moved character handling to format_character.
26793 (format_directive): Remove arguments, change return type.
26794 (parse_directive): New function.
26795 (pass_sprintf_length::compute_format_length): Move directive
26796 parsing to parse_directive.
26798 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26800 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26801 (assign_assembler_name_if_needed): ... this.
26802 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26803 (assign_assembler_name_if_needed): ... this.
26804 (free_lang_data_in_cgraph): Adjust callers.
26805 * cgraphunit.c (cgraph_node::analyze): Likewise.
26806 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26809 2017-01-23 Richard Biener <rguenther@suse.de>
26811 PR tree-optimization/79088
26812 PR tree-optimization/79188
26813 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26814 resetting loop bounds after last path deletion. Reset loop
26815 bounds of the target loop, make code match the comments.
26816 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26817 Make sure loops need no fixups.
26819 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26821 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26822 exponent support with double type for first argument.
26823 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26824 type returned by __builtin_vec_extract_sig,
26825 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26826 functions from "vector int" to "vector unsigned int" or from
26827 "vector long long int" to "vector unsigned long long int".
26828 Changed type returned by __builtin_vec_extract_exp,
26829 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26830 functions from "vector int" to "vector unsigned int" or from
26831 "vector long long int" to "vector unsigned long long int".
26832 Changed return type of __builtin_vec_test_data_class,
26833 __builtin_vec_test_data_class_sp, and
26834 __builtin_vec_test_data_class_dp from "vector int" to
26835 "vector bool int" or from "vector long long int" to "vector bool
26836 long long int" and changed second argument type from "unsigned
26837 int" to "int". Added new overloaded function forms "vector float
26838 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26839 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26840 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26841 double, vector unsigned long long int)" and "vector double
26842 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26843 long int)". Changed return type of
26844 __builtin_scalar_test_data_class and
26845 __builtin_scalar_test_data_class_sp and
26846 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26847 int" and changed second argument from "unsigned int" to "int".
26848 Changed type returned by __builtin_scalar_test_neg,
26849 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26850 from "int" to "bool int". Added new overloaded function form
26851 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26852 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26853 exponent double-precision with floating point first argument.
26854 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26855 documentation of scalar_test_data_class, scalar_test_neg,
26856 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26857 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26858 vec_test_data_class built-in functions to reflect refinements in
26859 their type signatures.
26861 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26863 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26865 (aarch64_elf_asm_destructor): Likewise.
26867 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26869 PR rtl-optimization/78634
26870 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26871 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26872 * ifcvt.c (noce_try_cmove): Add missing cost check.
26874 PR rtl-optimization/71724
26875 * combine.c (if_then_else_cond): Look for situations where it is
26876 beneficial to undo the work of one of the recursive calls.
26878 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26880 PR tree-optimization/70754
26881 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26882 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26883 combined stmt before it if not NULL.
26884 (combine_chains): Process refs reversely and compute dominance point
26887 2017-01-23 Martin Liska <mliska@suse.cz>
26889 PR tree-optimization/79196
26890 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26891 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26893 (strlen_optimize_stmt): Call the renamed function.
26895 2017-01-23 Michael Matz <matz@suse.de>
26897 PR tree-optimization/78384
26898 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26900 2017-01-23 Richard Biener <rguenther@suse.de>
26902 PR tree-optimization/79186
26903 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26904 both incoming edges before moving an assert.
26906 2017-01-23 Martin Jambor <mjambor@suse.cz>
26908 * ipa-prop.c (load_from_param_1): Removed.
26909 (load_from_unmodified_param): Bits from load_from_param_1 put back
26911 (load_from_param): Removed.
26912 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26913 with stmt. Reverted back to use of load_from_unmodified_param.
26915 2017-01-23 Martin Jambor <mjambor@suse.cz>
26918 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26919 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26920 field a pointer to garbage collected vector, mark lattices and
26921 ipcp_orig_node with GTY((skip)).
26922 (ipa_get_param_count): Adjust to descriptors being a pointer.
26923 (ipa_get_param): Likewise.
26924 (ipa_get_type): Likewise.
26925 (ipa_get_param_move_cost): Likewise.
26926 (ipa_set_param_used): Likewise.
26927 (ipa_get_controlled_uses): Likewise.
26928 (ipa_set_controlled_uses): Likewise.
26929 (ipa_is_param_used): Likewise.
26930 (ipa_node_params_t): Move into garbage collector. New methods insert
26932 (ipa_node_params_sum): Annotate wth GTY(()).
26933 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26935 (ipa_load_from_parm_agg): Adjust declaration.
26936 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26937 * ipa-profile.c (ipa_profile): Likewise.
26938 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26939 (ipa_populate_param_decls): Make descriptors parameter garbage
26941 (ipa_dump_param): Adjust to descriptors being a pointer.
26942 (ipa_alloc_node_params): Likewise.
26943 (ipa_initialize_node_params): Likewise.
26944 (load_from_param_1): Make descriptors parameter garbage collected.
26945 (load_from_unmodified_param): Likewise.
26946 (load_from_param): Likewise.
26947 (ipa_load_from_parm_agg): Likewise.
26948 (ipa_node_params::~ipa_node_params): Removed.
26949 (ipa_free_all_node_params): Remove call to delete operator.
26950 (ipa_node_params_t::insert): New.
26951 (ipa_node_params_t::remove): Likewise.
26952 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26953 copy known_csts and known_contexts vectors.
26954 (ipa_read_node_info): Adjust to descriptors being a pointer.
26955 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26957 (ipcp_transform_function): Make descriptors variable garbage
26960 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26962 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26963 * config/i386/avx512dqintrin.h: Ditto.
26964 * config/i386/avx512fintrin.h: Ditto.
26965 * config/i386/i386.c: Handle new builtins.
26966 * config/i386/i386-builtin.def: Add new builtins.
26967 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26968 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26970 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26971 Martin Liska <mliska@suse.cz>
26973 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26974 * asan.c (asan_expand_poison_ifn): Support stores and use
26975 appropriate ASAN report function.
26976 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26977 * internal-fn.def (ASAN_POISON_USE): Declare.
26978 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26979 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26980 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26981 ASAN_POISON calls w/o LHS.
26982 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26983 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26984 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26985 * gimplify.c (asan_poison_variables): Add attribute
26986 use_after_scope_memory to variables that really needs to live
26988 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26989 having the attribute.
26991 2017-01-23 Martin Liska <mliska@suse.cz>
26993 * asan.c (create_asan_shadow_var): New function.
26994 (asan_expand_poison_ifn): Likewise.
26995 * asan.h (asan_expand_poison_ifn): New declaration.
26996 * internal-fn.c (expand_ASAN_POISON): Likewise.
26997 * internal-fn.def (ASAN_POISON): New builtin.
26998 * sanopt.c (pass_sanopt::execute): Expand
26999 asan_expand_poison_ifn.
27000 * tree-inline.c (copy_decl_for_dup_finish): Make function
27002 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27003 * tree-ssa.c (is_asan_mark_p): New function.
27004 (execute_update_addresses_taken): Rewrite local variables
27005 (identified just by use-after-scope as addressable) into SSA.
27007 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27009 * doc/install.texi (Specific): opensource.apple.com uses https
27010 now. Remove trailing slash.
27012 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27014 * README.Portability: Remove note on an Irix compatibility issue.
27016 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27018 * gcov.c (INCLUDE_ALGORITHM): Define.
27019 (INCLUDE_VECTOR): Define.
27020 No longer include <vector> and <algorithm> directly.
27022 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27024 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27026 * doc/invoke.texi (Code Gen Options): Ditto.
27028 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27031 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27033 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27035 rtl-optimization/79125
27036 * cprop.c (local_cprop_pass): Handle cases where we make an
27037 unconditional trap.
27039 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27043 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27044 read from, for big endian.
27046 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27048 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27049 register pauth builtins for LP64 only.
27051 2017-01-20 Marek Polacek <polacek@redhat.com>
27054 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27057 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27059 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27061 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27062 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27063 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27065 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27068 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27069 in big-endian mode when they are not a single duplicated value.
27071 2017-01-20 Richard Biener <rguenther@suse.de>
27073 * BASE-VER: Bump to 7.0.1.
27075 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27077 * omp-low.c (omplow_simd_context): New struct. Use it...
27078 (lower_rec_simd_input_clauses): ...here and...
27079 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27080 references to idx, lane, max_vf, is_simt.
27082 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27084 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27087 2017-01-20 Martin Jambor <mjambor@suse.cz>
27089 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27090 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27092 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27093 (GTFILES): Rename hsa.c to hsa-common.c.
27094 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27095 * hsa-dump.c: Likewise.
27096 * hsa-gen.c: Likewise.
27097 * hsa-regalloc.c: Likewise.
27098 * ipa-hsa.c: Likewise.
27099 * omp-expand.c: Likewise.
27100 * omp-low.c: Likewise.
27101 * toplev.c: Likewise.
27103 2017-01-20 Marek Polacek <polacek@redhat.com>
27106 * doc/invoke.texi: Document -Wduplicated-branches.
27107 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27108 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27109 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27110 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27111 return 0 only when not OEP_LEXICOGRAPHIC.
27112 (fold_build_cleanup_point_expr): Use the expression
27113 location when building CLEANUP_POINT_EXPR.
27114 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27115 * tree.c (add_expr): Handle error_mark_node.
27117 2017-01-20 Martin Liska <mliska@suse.cz>
27120 * tree-profile.c (init_ic_make_global_vars): Do not call
27122 (gimple_init_gcov_profiler): Likewise.
27124 2017-01-20 Martin Liska <mliska@suse.cz>
27127 * cgraph.h (maybe_create_reference): Remove argument and
27129 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27131 * ipa-cp.c (create_specialized_node): Likewise.
27132 * symtab.c (symtab_node::maybe_create_reference): Handle
27133 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27135 2017-01-20 Martin Liska <mliska@suse.cz>
27137 * read-rtl-function.c (function_reader::create_function): Use
27138 build_decl instread of build_decl_stat.
27140 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27142 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27143 * config/i386/avx512dqintrin.h: Ditto.
27144 * config/i386/avx512fintrin.h: Ditto.
27145 * config/i386/i386-builtin-types.def: Add new types.
27146 * config/i386/i386.c: Handle new types.
27147 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27148 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27149 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27150 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27151 (__builtin_ia32_kshiftridi): New.
27152 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27154 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27158 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27159 define to rs6000_init_stack_protect_guard.
27160 (rs6000_init_stack_protect_guard): New function.
27162 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27163 Yunqiang Su <yunqiang.su@imgtec.com>
27165 * config.gcc (supported_defaults): Add madd4.
27166 (with_madd4): Add validation.
27167 (all_defaults): Add madd4.
27168 * config/mips/mips.opt (mmadd4): New option.
27169 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27171 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27173 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27174 (ISA_HAS_FUSED_MADD4): Likewise.
27175 * doc/invoke.texi (-mmadd4): Document the new option.
27176 * doc/install.texi (--with-madd4): Document the new option.
27178 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27180 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27181 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27182 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27183 (aarch64_init_pauth_hint_builtins): New.
27184 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27185 (aarch64_expand_builtin): Expand new builtins.
27187 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27189 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27190 * combine-stack-adj.c (no_unhandled_cfa): Handle
27191 REG_CFA_TOGGLE_RA_MANGLE.
27192 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27193 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27194 info for return address signing.
27195 (aarch64_expand_epilogue): Likewise.
27197 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27199 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27200 * config/aarch64/aarch64-protos.h
27201 (aarch64_return_address_signing_enabled): New declaration.
27202 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27204 (aarch64_expand_prologue): Sign return address before it's pushed onto
27206 (aarch64_expand_epilogue): Authenticate return address fetched from
27208 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27209 (aarch64_attributes): New function attributes for "sign-return-address".
27210 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27211 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27212 ("*do_return"): Generate combined instructions according to key index.
27213 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27214 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27216 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27217 * config/aarch64/aarch64.opt (msign-return-address=): New.
27218 * doc/extend.texi (AArch64 Function Attributes): Documents
27219 "sign-return-address=".
27220 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27222 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27224 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27225 overall option summary.
27227 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27229 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27230 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27231 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27232 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27234 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27236 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27237 -mpower9-minmax by default for -mcpu=power9.
27238 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27239 128-bit floating point.
27241 2017-01-20 Alan Modra <amodra@gmail.com>
27243 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27244 optimizing for size.
27246 2017-01-20 Alan Modra <amodra@gmail.com>
27249 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27250 for strcmp and strncmp from corresponding builtin decl.
27252 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27254 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27255 instead of i386/rtems-64.h.
27256 * config/i386/rtems-64.h: Remove.
27258 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27262 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27264 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27266 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27268 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27269 Change int to HOST_WIDE_INT.
27270 * config/aarch64/aarch64-protos.h
27271 (aarch64_simd_gen_const_vector_dup): Likewise.
27272 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27274 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27276 * langhooks-def.h (lhd_type_for_size): New decl.
27277 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27278 * langhooks.c (lhd_type_for_size): New function, taken from
27281 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27283 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27284 define_bypass for CR latency.
27285 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27286 (power9-alu2): Add define_bypass for CR latency.
27288 (power9-mul): Update insn latency.
27289 (power9-mul-compare): Update insn latency, bypass latency and remove
27292 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27294 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27296 * config/aarch64/aarch64.md
27297 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27298 aarch64_nopcrelative_literal_loads.
27299 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27301 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27303 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27304 TARGET_LOONGSON_3A.
27305 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27307 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27310 * config.gcc (supported_defaults): Add lxc1-sxc1.
27311 (with_lxc1_sxc1): Add validation.
27312 (all_defaults): Add lxc1-sxc1.
27313 * config/mips/mips.opt (mlxc1-sxc1): New option.
27314 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27316 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27317 __mips_no_lxc1_sxc1.
27318 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27319 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27320 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27322 2017-01-19 Richard Biener <rguenther@suse.de>
27324 PR tree-optimization/72488
27325 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27326 sure to restore SSA info.
27327 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27329 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27331 PR rtl-optimization/79121
27332 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27333 of the inner type when shifting an extended value.
27335 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27338 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27339 interposable aliases.
27341 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27344 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27345 Use the evmergelohi instruction.
27346 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27347 (mov_sitf_e500_subreg8_2_be): Likewise.
27348 (mov_sitf_e500_subreg12_2_le): Likewise.
27349 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27350 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27351 (mov_sitf_e500_subreg8_2_le): Likewise.
27352 (mov_sitf_e500_subreg12_2_be): Likewise.
27354 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27356 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27357 attribute from vecsimple to vecperm.
27358 (altivec_vbpermq2): Likewise.
27360 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27363 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27365 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27366 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27367 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27368 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27369 case where N arg is SIZE_MAX.
27370 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27371 (cmpstrsi): Add pattern.
27373 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27375 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27376 __builtin_vec_revb builtins.
27377 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27378 built-in functions to support generation of the ISA 3.0 XXBR<x>
27379 vector byte reverse instructions.
27380 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27381 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27382 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27383 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27384 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27385 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27386 (P9V_BUILTIN_VEC_REVB): Likewise.
27387 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27388 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27389 (p9_xxbrq_v16qi): Likewise.
27390 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27391 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27392 (p9_xxbrh_v8hi): Likewise.
27393 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27394 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27395 vec_revb built-in functions.
27397 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27399 PR rtl-optimization/78952
27400 * config/i386/i386.md (any_extract): New code iterator.
27401 (*insvqi_2): Use any_extract for source operand.
27402 (*insvqi_3): Use any_shiftrt for source operand.
27404 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27406 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27408 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27410 2017-01-18 Matthias Klose <doko@ubuntu.com>
27412 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27414 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27416 * config/rs6000/altivec.h (vec_bperm): Change #define.
27417 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27418 (altivec_vbpermq2): New define_insn.
27419 (altivec_vbpermd): Likewise.
27420 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27421 function interface.
27422 (VBPERMD): Likewise.
27423 (VBPERM): New polymorphic function interface.
27424 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27425 Add entries for P9V_BUILTIN_VEC_VBPERM.
27426 * doc/extend.texi: Add interfaces for vec_bperm.
27428 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27430 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27431 first letter of error messages.
27432 (s390_resolve_overloaded_builtin): Likewise.
27433 * config/s390/s390.c (s390_expand_builtin): Likewise.
27434 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27435 (s390_valid_target_attribute_inner_p): Likewise.
27436 * config/s390/s390.md ("tabort"): Likewise.
27438 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27440 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27441 (ISA_AVOID_DIV_HILO): New macro.
27442 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27443 (ISA_HAS_DDIV): Likewise.
27445 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27447 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27449 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27451 * doc/invoke.texi (fabi-version): Spelling fix.
27453 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27456 * doc/invoke.texi (fabi-version): Mention mangling fix for
27459 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27462 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27464 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27467 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27468 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27470 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27472 (stack_protect_test): Ditto.
27473 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27474 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27476 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27477 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27478 -mstack-protector-guard-offset=.
27479 (RS/6000 and PowerPC Options): Ditto.
27481 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27483 * config/i386/i386.h (MASK_CLASS_P): New define.
27484 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27485 there are no registers from different register sets also when
27486 mask registers are used. Update function comment.
27487 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27488 to (*k/*r) and (*k/*km) alternatives.
27490 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27492 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27493 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27494 (EH_RETURN_HANDLER_RTX): New define.
27495 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27496 Force frame pointer in EH return functions.
27497 (aarch64_expand_epilogue): Add barrier for eh_return.
27498 (aarch64_final_eh_return_addr): Remove.
27499 (aarch64_eh_return_handler_rtx): New function.
27500 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27502 (aarch64_eh_return_handler_rtx): New prototype.
27504 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27506 * config/rs6000/altivec.h (vec_rlmi): New #define.
27507 (vec_vrlnm): Likewise.
27508 (vec_rlnm): Likewise.
27509 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27510 (UNSPEC_VRLNM): Likewise.
27511 (VIlong): New mode iterator.
27512 (altivec_vrl<VI_char>mi): New define_insn.
27513 (altivec_vrl<VI_char>nm): Likewise.
27514 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27516 (VRLDNM): Likewise.
27517 (RLNM): New polymorphic function entry.
27518 (VRLWMI): New monomorphic function entry.
27519 (VRLDMI): Likewise.
27520 (RLMI): New polymorphic function entry.
27521 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27522 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27523 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27526 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27529 * dwarf2out.c (field_byte_offset): Restore the
27530 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27531 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27532 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27535 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27538 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27540 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27543 * dwarf2out.c (add_data_member_location_attribute): For constant
27544 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27545 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27546 DW_AT_byte_size attributes.
27548 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27550 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27551 after forcing to constant memory when the code model is medium.
27553 2017-01-17 Julia Koval <julia.koval@intel.com>
27556 * config/i386/avx512fintrin.h
27557 (_mm512_i32gather_ps): Change __addr type to void const*.
27558 (_mm512_mask_i32gather_ps): Ditto.
27559 (_mm512_i32gather_pd): Ditto.
27560 (_mm512_mask_i32gather_pd): Ditto.
27561 (_mm512_i64gather_ps): Ditto.
27562 (_mm512_mask_i64gather_ps): Ditto.
27563 (_mm512_i64gather_pd): Ditto.
27564 (_mm512_mask_i64gather_pd): Ditto.
27565 (_mm512_i32gather_epi32): Ditto.
27566 (_mm512_mask_i32gather_epi32): Ditto.
27567 (_mm512_i32gather_epi64): Ditto.
27568 (_mm512_mask_i32gather_epi64): Ditto.
27569 (_mm512_i64gather_epi32): Ditto.
27570 (_mm512_mask_i64gather_epi32): Ditto.
27571 (_mm512_i64gather_epi64): Ditto.
27572 (_mm512_mask_i64gather_epi64): Ditto.
27573 (_mm512_i32scatter_ps): Change __addr type to void*.
27574 (_mm512_mask_i32scatter_ps): Ditto.
27575 (_mm512_i32scatter_pd): Ditto.
27576 (_mm512_mask_i32scatter_pd): Ditto.
27577 (_mm512_i64scatter_ps): Ditto.
27578 (_mm512_mask_i64scatter_ps): Ditto.
27579 (_mm512_i64scatter_pd): Ditto.
27580 (_mm512_mask_i64scatter_pd): Ditto.
27581 (_mm512_i32scatter_epi32): Ditto.
27582 (_mm512_mask_i32scatter_epi32): Ditto.
27583 (_mm512_i32scatter_epi64): Ditto.
27584 (_mm512_mask_i32scatter_epi64): Ditto.
27585 (_mm512_i64scatter_epi32): Ditto.
27586 (_mm512_mask_i64scatter_epi32): Ditto.
27587 (_mm512_i64scatter_epi64): Ditto.
27588 (_mm512_mask_i64scatter_epi64): Ditto.
27589 * config/i386/avx512pfintrin.h
27590 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27591 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27592 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27593 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27594 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27595 (_mm512_prefetch_i32scatter_ps): Ditto.
27596 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27597 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27598 (_mm512_prefetch_i64scatter_pd): Ditto.
27599 (_mm512_prefetch_i64scatter_ps): Ditto.
27600 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27601 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27602 * config/i386/avx512vlintrin.h
27603 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27604 (_mm_mmask_i32gather_ps): Ditto.
27605 (_mm256_mmask_i32gather_pd): Ditto.
27606 (_mm_mmask_i32gather_pd): Ditto.
27607 (_mm256_mmask_i64gather_ps): Ditto.
27608 (_mm_mmask_i64gather_ps): Ditto.
27609 (_mm256_mmask_i64gather_pd): Ditto.
27610 (_mm_mmask_i64gather_pd): Ditto.
27611 (_mm256_mmask_i32gather_epi32): Ditto.
27612 (_mm_mmask_i32gather_epi32): Ditto.
27613 (_mm256_mmask_i32gather_epi64): Ditto.
27614 (_mm_mmask_i32gather_epi64): Ditto.
27615 (_mm256_mmask_i64gather_epi32): Ditto.
27616 (_mm_mmask_i64gather_epi32): Ditto.
27617 (_mm256_mmask_i64gather_epi64): Ditto.
27618 (_mm_mmask_i64gather_epi64): Ditto.
27619 (_mm256_i32scatter_ps): Change __addr type to void*.
27620 (_mm256_mask_i32scatter_ps): Ditto.
27621 (_mm_i32scatter_ps): Ditto.
27622 (_mm_mask_i32scatter_ps): Ditto.
27623 (_mm256_i32scatter_pd): Ditto.
27624 (_mm256_mask_i32scatter_pd): Ditto.
27625 (_mm_i32scatter_pd): Ditto.
27626 (_mm_mask_i32scatter_pd): Ditto.
27627 (_mm256_i64scatter_ps): Ditto.
27628 (_mm256_mask_i64scatter_ps): Ditto.
27629 (_mm_i64scatter_ps): Ditto.
27630 (_mm_mask_i64scatter_ps): Ditto.
27631 (_mm256_i64scatter_pd): Ditto.
27632 (_mm256_mask_i64scatter_pd): Ditto.
27633 (_mm_i64scatter_pd): Ditto.
27634 (_mm_mask_i64scatter_pd): Ditto.
27635 (_mm256_i32scatter_epi32): Ditto.
27636 (_mm256_mask_i32scatter_epi32): Ditto.
27637 (_mm_i32scatter_epi32): Ditto.
27638 (_mm_mask_i32scatter_epi32): Ditto.
27639 (_mm256_i32scatter_epi64): Ditto.
27640 (_mm256_mask_i32scatter_epi64): Ditto.
27641 (_mm_i32scatter_epi64): Ditto.
27642 (_mm_mask_i32scatter_epi64): Ditto.
27643 (_mm256_i64scatter_epi32): Ditto.
27644 (_mm256_mask_i64scatter_epi32): Ditto.
27645 (_mm_i64scatter_epi32): Ditto.
27646 (_mm_mask_i64scatter_epi32): Ditto.
27647 (_mm256_i64scatter_epi64): Ditto.
27648 (_mm256_mask_i64scatter_epi64): Ditto.
27649 (_mm_i64scatter_epi64): Ditto.
27650 (_mm_mask_i64scatter_epi64): Ditto.
27651 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27652 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27653 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27654 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27655 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27656 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27657 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27658 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27659 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27660 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27661 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27662 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27663 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27664 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27665 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27666 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27667 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27668 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27669 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27670 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27671 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27672 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27673 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27674 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27675 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27676 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27677 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27678 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27679 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27680 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27681 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27682 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27683 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27684 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27685 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27686 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27687 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27688 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27689 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27690 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27691 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27692 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27693 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27694 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27695 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27696 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27697 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27698 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27699 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27700 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27701 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27702 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27703 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27704 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27705 definitions accordingly.
27707 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27708 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27711 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27714 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27717 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27718 hard regs for inner regno.
27720 2017-01-17 Martin Liska <mliska@suse.cz>
27723 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27724 assumption and add comment.
27726 2017-01-17 Nathan Sidwell <nathan@acm.org>
27728 * ipa-visibility.c (localize_node): New function, broken out of ...
27729 (function_and_variable_visibility): ... here. Call it.
27731 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27733 PR middle-end/77445
27734 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27735 correctly set frequency of oudgoing edge.
27736 (duplicate_thread_path): Fix profile updating.
27738 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27741 * configure.ac: Add GCC_BASE_VER.
27742 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27743 version from BASE-VER file.
27744 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27745 (gcc.o): Depend on $(BASEVER).
27746 * common.opt (dumpfullversion): New option.
27747 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27748 * doc/invoke.texi: Document -dumpfullversion.
27749 * doc/install.texi: Document --with-gcc-major-version-only.
27750 * configure: Regenerated.
27752 2017-01-17 Richard Biener <rguenther@suse.de>
27754 PR tree-optimization/71433
27755 * tree-vrp.c (register_new_assert_for): Merge same asserts
27756 on all incoming edges.
27757 (process_assert_insertions_for): Handle insertions at the
27760 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27762 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27763 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27765 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27768 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27771 2017-01-17 Alan Modra <amodra@gmail.com>
27774 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27775 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27776 symbolic stack limit when pic.
27778 2017-01-16 Martin Sebor <msebor@redhat.com>
27780 PR tree-optimization/78608
27781 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27783 2017-01-16 Jeff Law <law@redhat.com>
27786 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27787 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27788 for several include directories that may be relative to sysroot.
27789 * config/i386/x-mingw32 (gplus_includedir): Define.
27790 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27791 (native_system_includedir): Likewise.
27792 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27793 override if TARGET_SYSTEM_ROOT is defined.
27794 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27796 PR tree-optimization/79090
27797 PR tree-optimization/33562
27798 PR tree-optimization/61912
27799 PR tree-optimization/77485
27800 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27801 and computed trims into the dump file.
27803 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27805 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27807 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27810 * gimplify.c (gimplify_init_constructor): If want_value and
27811 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27815 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27816 sequence. Formatting fixes.
27817 (doloop_optimize): Formatting fixes.
27820 * gcc.c (debug_level_greater_than_spec_func): New function.
27821 (static_spec_functions): Add debug-level-gt spec function.
27822 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27824 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27825 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27826 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27827 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27828 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27829 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27831 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27833 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27834 QImode fixups to general and mask registers only.
27836 2017-01-16 Carl Love <cel@us.ibm.com>
27838 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27839 for built-in functions
27840 vector signed char vec_nabs (vector signed char)
27841 vector signed short vec_nabs (vector signed short)
27842 vector signed int vec_nabs (vector signed int)
27843 vector signed long long vec_nabs (vector signed long long)
27844 vector float vec_nabs (vector float)
27845 vector double vec_nabs (vector double)
27846 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27848 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27849 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27850 * doc/extend.texi: Update the documentation file for the new built-in
27853 2017-01-16 Martin Sebor <msebor@redhat.com>
27855 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27858 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27860 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27861 UNSPEC_VSX__XXSPLTD to require special splat handling.
27863 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27866 * system.h: Poison strndup.
27868 2017-01-16 Alan Modra <amodra@gmail.com>
27871 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27874 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27876 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27878 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27880 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27881 call recog here. Assert that INSN_CODE (insn) is non-negative.
27883 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27886 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27888 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27889 in the currently scheduled RTL fragment.
27891 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27893 PR rtl-optimization/78751
27894 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27897 2017-01-14 Jeff Law <law@redhat.com>
27899 PR tree-optimization/79090
27900 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27901 variable length stores.
27902 (compute_trims): Delete dead assignment to *trim_tail.
27903 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27906 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27908 PR rtl-optimization/78626
27909 PR rtl-optimization/78727
27910 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27911 of a block, and split such blocks after everything else is finished.
27913 2017-01-14 Alan Modra <amodra@gmail.com>
27916 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27917 target legitimate_combined_insn.
27918 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27919 (rs6000_legitimate_combined_insn): New function.
27920 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27922 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27923 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27924 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27926 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27928 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27930 2017-01-13 Jeff Law <law@redhat.com>
27932 PR tree-optimization/33562
27933 PR tree-optimization/61912
27934 PR tree-optimization/77485
27935 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27937 (delete_dead_assignment): Likewise.
27938 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27939 statement to delete_dead_call and delete_dead_assignment.
27941 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27944 * substring-locations.c (format_warning_va): Strengthen case 1 so
27945 that both endpoints of the substring must be within the format
27946 range for just the substring to be printed.
27948 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27950 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27951 * config/i386/i386.c (ix86_target_string): Add missing options
27952 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27953 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27954 flags_other and ix86_target_other to flags2_other. Display unknown
27956 (ix86_valid_target_attribute_inner_p): Add missing options and
27957 reorder options by implied ISAs, as in ix86_target_string.
27959 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27961 * hash-table.h (hash_table::too_empty_p): New function.
27962 (hash_table::expand): Use it.
27963 (hash_table::traverse): Likewise.
27964 (hash_table::empty_slot): Use sizeof (value_type) instead of
27965 sizeof (PTR) to convert bytes to elements. Shrink the table
27966 if the current size is excessive for the current number of
27969 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27971 * ira-costs.c (record_reg_classes): Break from the inner loop
27972 early once alt_fail is known to be true. Update outer loop
27973 handling accordingly.
27975 2017-01-13 Jeff Law <law@redhat.com>
27977 * tree-ssa-dse.c (decrement_count): New function.
27978 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27979 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27980 when we know the partially dead statement is a mem* function.
27982 PR tree-optimization/61912
27983 PR tree-optimization/77485
27984 * tree-ssa-dse.c: Include expr.h.
27985 (maybe_trim_constructor_store): New function.
27986 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27988 PR tree-optimization/33562
27989 PR tree-optimization/61912
27990 PR tree-optimization/77485
27991 * doc/invoke.texi: Document new dse-max-object-size param.
27992 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27993 * tree-ssa-dse.c: Include params.h.
27994 (dse_store_status): New enum.
27995 (initialize_ao_ref_for_dse): New, partially extracted from
27997 (valid_ao_ref_for_dse, normalize_ref): New.
27998 (setup_live_bytes_from_ref, compute_trims): Likewise.
27999 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28000 (maybe_trim_partially_dead_store): Likewise.
28001 (maybe_trim_complex_store): Likewise.
28002 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28003 Track what bytes live from the original store. Return tri-state
28004 for dead, partially dead or live.
28005 (dse_dom_walker): Add constructor, destructor and new private members.
28006 (delete_dead_call, delete_dead_assignment): New extracted from
28008 (dse_optimize_stmt): Make a member of dse_dom_walker.
28009 Use initialize_ao_ref_for_dse.
28011 PR tree-optimization/33562
28012 PR tree-optimization/61912
28013 PR tree-optimization/77485
28014 * sbitmap.h (bitmap_count_bits): Prototype.
28015 (bitmap_clear_range, bitmap_set_range): Likewise.
28016 * sbitmap.c (bitmap_clear_range): New function.
28017 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28019 2017-01-13 Martin Liska <mliska@suse.cz>
28022 * function.c (set_cfun): Add new argument force.
28023 * function.h (set_cfun): Likewise.
28024 * ipa-inline-transform.c (inline_call): Use the function when
28025 strict alising from is dropped for function we inline to.
28027 2017-01-13 Richard Biener <rguenther@suse.de>
28029 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28030 for dumping GIMPLE INTEGER_CSTs.
28032 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28034 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28035 to 201112L since C++17.
28037 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28040 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28041 if -fsanitize=kernel-address is present.
28043 2017-01-13 Richard Biener <rguenther@suse.de>
28045 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28046 as _Literal ( type ) number in case usual suffixes do not
28047 preserve all information.
28049 2017-01-13 Richard Biener <rguenther@suse.de>
28051 PR tree-optimization/77283
28052 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28053 and ssa-iterators.h.
28054 (is_feasible_trace): Implement a cost model based on joiner
28057 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28060 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28061 char or short to __float128/_Float128 directly.
28063 2017-01-12 Martin Sebor <msebor@redhat.com>
28065 to -Wformat-overflow.
28066 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28067 (min_bytes_remaining): Same.
28068 (get_string_length): Same.
28069 (format_string): Same.
28070 (format_directive): Same.
28072 (pass_sprintf_length::handle_gimple_call): Same.
28074 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28076 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28077 info.nowrite calls with no lhs that can't throw. Return bool
28078 whether gsi_remove has been called or not.
28079 (pass_sprintf_length::handle_gimple_call): Return bool whether
28080 try_substitute_return_value called gsi_remove. Formatting fix.
28081 (pass_sprintf_length::execute): Don't use gsi_remove if
28082 handle_gimple_call returned true.
28085 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28086 be removed due to side-effects, don't remove following barrier nor
28087 turn the successor edge into fallthru edge.
28089 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28092 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28093 element-reversing loads and stores as not swappable.
28095 2017-01-12 Nathan Sidwell <nathan@acm.org>
28096 Nicolai Stange <nicstange@gmail.com>
28098 * combine.c (try_combine): Don't ignore result of overlap checking
28099 loop. Combine overlap & asm check into single loop.
28101 2017-01-12 Richard Biener <rguenther@suse.de>
28103 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28104 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28106 2017-01-12 Richard Biener <rguenther@suse.de>
28108 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28109 and TS_TARGET_OPTION directly derive from TS_BASE.
28110 * tree-core.h (tree_optimization_option): Derive from tree_base.
28111 (tree_target_option): Likewise.
28113 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28115 * config/i386/i386.c (memory_address_length): Increase len
28116 only when rip_relative_addr_p returns false.
28118 2017-01-11 Julia Koval <julia.koval@intel.com>
28120 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28121 (OPTION_MASK_ISA_SGX_SET): New.
28122 (ix86_handle_option): Handle OPT_msgx.
28123 * config.gcc: Added sgxintrin.h.
28124 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28125 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28126 * config/i386/i386.c (ix86_target_string): Add -msgx.
28128 (ix86_option_override_internal): Handle new options.
28129 (ix86_valid_target_attribute_inner_p): Add sgx.
28130 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28131 * config/i386/i386.opt: Add msgx.
28132 * config/i386/sgxintrin.h: New file.
28133 * config/i386/x86intrin.h: Add sgxintrin.h.
28135 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28138 * fold-const.c (maybe_nonzero_address): Return 1 for function
28140 (tree_single_nonzero_warnv_p): Don't handle function local objects
28144 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28147 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28150 * opts.c: Include "spellcheck.h"
28151 (struct string_fragment): New struct.
28152 (struct edit_distance_traits<const string_fragment &>): New
28154 (get_closest_sanitizer_option): New function.
28155 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28157 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28159 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28161 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28162 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28163 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28164 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28165 for initial die_offset if dwarf_split_debug_info.
28166 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28167 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28169 (output_skeleton_debug_sections): Formatting fix. Use
28170 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28171 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28173 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28175 * config/arm/cortex-a53.md: Add bypasses for
28176 cortex_a53_r2f_cvt.
28177 (cortex_a53_r2f): Only use for transfers.
28178 (cortex_a53_f2r): Likewise.
28179 (cortex_a53_r2f_cvt): Add reservation for conversions.
28180 (cortex_a53_f2r_cvt): Likewise.
28182 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28184 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28185 to all inlined functions, change static to extern.
28187 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28190 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28192 (arm_assemble_integer): Likewise.
28194 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28196 * config.gcc: Use new awk script to check CPU, FPU and architecture
28197 parameters for --with-... options.
28198 * config/arm/parsecpu.awk: New file
28199 * config/arm/arm-cpus.in: New file.
28200 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28202 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28204 * config/arm/t-arm: Update dependency rules.
28205 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28206 of processing .def files.
28207 * config/arm/genopt.sh: Deleted.
28208 * config/arm/gentune.sh: Deleted.
28209 * config/arm/arm-cores.def: Deleted.
28210 * config/arm/arm-arches.def: Deleted.
28211 * config/arm/arm-fpus.def: Deleted.
28212 * config/arm/arm-tune.md: Regenerated.
28213 * config/arm/arm-tables.opt: Regenerated.
28214 * config/arm/arm-cpu.h: New generated file.
28215 * config/arm/arm-cpu-data.h: New generated file.
28216 * config/arm/arm-cpu-cdata.h: New generated file.
28218 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28221 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28223 (input_varpool_node): Unpack dynamically_initialized bit.
28225 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28227 PR rtl-optimization/79032
28228 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28229 the alignment of the adjusted memory reference against that of MODE,
28230 instead of the alignment of the original memory reference.
28232 2017-01-11 Martin Jambor <mjambor@suse.cz>
28234 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28236 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28237 decorated functions.
28239 2017-01-11 Richard Biener <rguenther@suse.de>
28241 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28242 set range/nonnull info for PHI results. Do not set it on
28243 stmts marked for removal.
28245 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28247 * expr.c (store_field): In the bitfield case, fetch the return value
28248 from the registers before applying a single big-endian adjustment.
28249 Always do a final load for a BLKmode value not larger than a word.
28251 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28254 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28255 that we correctly handle column numbers greater than
28256 LINE_MAP_MAX_COLUMN_NUMBER.
28258 2017-01-10 Martin Sebor <msebor@redhat.com>
28260 PR middle-end/78245
28261 * gimple-ssa-sprintf.c (get_destination_size): Call
28262 {init,fini}object_sizes.
28263 * tree-object-size.c (addr_object_size): Adjust.
28264 (pass_through_call): Adjust.
28265 (pass_object_sizes::execute): Adjust.
28266 * tree-object-size.h (fini_object_sizes): Declare.
28268 2017-01-10 Martin Sebor <msebor@redhat.com>
28270 PR tree-optimization/78775
28271 * builtins.c (get_size_range): Move...
28272 * calls.c: ...to here.
28273 (alloc_max_size): Accept zero argument.
28274 (operand_signed_p): Remove.
28275 (maybe_warn_alloc_args_overflow): Call get_size_range.
28276 * calls.h (get_size_range): Declare.
28278 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28280 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28281 from TI's devices.csv file as of September 2016.
28282 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28284 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28286 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28287 * doc/invoke.texi: Likewise.
28288 * doc/md.texi: Likewise.
28289 * doc/objc.texi: Likewise.
28291 2017-01-10 Joshua Conner <joshconner@google.com>
28293 * config/arm/fuchsia-elf.h: New file.
28294 * config/fuchsia.h: New file.
28295 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28296 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28298 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28300 2016-01-10 Richard Biener <rguenther@suse.de>
28302 PR tree-optimization/79034
28303 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28304 Propagate out degenerate PHIs in the joiner.
28306 2017-01-10 Martin Liska <mliska@suse.cz>
28308 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28309 (sort_congruence_classes_by_decl_uid): Likewise.
28310 (sort_congruence_class_groups_by_decl_uid): Likewise.
28311 (sem_item_optimizer::merge_classes): Sort class, groups in these
28312 classes and members in the groups by DECL_UID of declarations.
28313 This would make merge operations stable.
28315 2017-01-10 Martin Liska <mliska@suse.cz>
28317 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28318 usage of m_classes_vec.
28319 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28320 (sem_item_optimizer::get_group_by_hash): Likewise.
28321 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28322 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28323 (sem_item_optimizer::verify_classes): Likewise.
28324 (sem_item_optimizer::process_cong_reduction): Likewise.
28325 (sem_item_optimizer::dump_cong_classes): Likewise.
28326 (sem_item_optimizer::merge_classes): Likewise.
28327 * ipa-icf.h (congruence_class_hash): Rename from
28328 congruence_class_group_hash. Remove declaration of m_classes_vec.
28330 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28332 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28333 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28334 * config.gcc: Add avx512vpopcntdqintrin.h.
28335 * config/i386/avx512vpopcntdqintrin.h: New.
28336 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28337 * config/i386/i386-builtin-types.def: Add new types.
28338 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28339 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28340 __builtin_ia32_vpopcountq_v8di_mask): New.
28341 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28342 __AVX512VPOPCNTDQ__.
28343 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28344 (PTA_AVX512VPOPCNTDQ): Define.
28345 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28346 TARGET_AVX512VPOPCNTDQ_P): Define.
28347 * config/i386/i386.opt: Add mavx512vpopcntdq.
28348 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28349 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28351 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28353 PR middle-end/77484
28354 * predict.def (PRED_CALL): Set to 67.
28356 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28358 * expr.c (store_field): In the bitfield case, if the value comes from
28359 a function call and is of an aggregate type returned in registers, do
28360 not modify the field mode; extract the value in all cases if the mode
28361 is BLKmode and the size is not larger than a word.
28363 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28366 * config/i386/cpuid.h: Fix undefined behavior.
28368 2017-01-04 Jeff Law <law@redhat.com>
28370 PR tree-optimization/79007
28371 PR tree-optimization/67955
28372 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28373 conservative for pt.null when flag_non_call_exceptions is on.
28375 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28377 PR translation/79019
28378 PR translation/79020
28379 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28380 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28381 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28383 * config/avr/avr.opt (maccumulate-args): Likewise.
28384 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28385 * common.opt (freport-bug): Likewise.
28386 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28387 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28388 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28389 translatable string.
28390 * config/i386/i386.c (function_value_32): Likewise.
28391 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28392 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28394 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28395 * common/config/msp430/msp430-common.c (msp430_handle_option):
28397 * symtab.c (symtab_node::verify_base): Likewise.
28398 * opts.c (set_debug_level): Likewise.
28399 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28400 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28401 missing whitespace to translatable strings.
28402 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28403 * config/sh/superh.h: Likewise.
28404 * config/i386/xopintrin.h: Likewise.
28405 * config/i386/znver1.md: Likewise.
28406 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28407 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28408 * double-int.h (struct double_int): Likewise.
28409 * double-int.c (div_and_round_double): Likewise.
28410 * wide-int.cc: Likewise.
28411 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28412 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28413 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28414 (crossjumps_occurred): ... this.
28415 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28418 PR tree-optimization/78899
28419 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28420 returning bool return struct loop *, NULL for failure and the new
28422 (versionable_outer_loop_p): Don't version outer loop if it has
28423 dont_vectorized bit set.
28424 (tree_if_conversion): When versioning outer loop, ensure
28425 tree_if_conversion is performed also on the inner loop of the
28426 non-vectorizable outer loop copy.
28427 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28428 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28429 prevent vectorization of it.
28430 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28431 the outer loop vectorization of the non-scalar version is attempted
28432 before vectorization of the inner loop in scalar version. If
28433 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28434 vectorization of its inner loop.
28435 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28436 has 2 inner loops, rename also on edges from bb whose single pred
28437 is outer_loop->header. Fix typo in function comment.
28439 2017-01-09 Martin Sebor <msebor@redhat.com>
28442 * asan.c (asan_emit_stack_protection): Increase local buffer size
28443 to avoid snprintf truncation warning.
28445 2017-01-09 Andrew Pinski <apinski@cavium.com>
28447 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28448 to reference thunderx2t99 for the tuning structure
28449 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28451 (thunderx2t99_extra_costs): This.
28452 * config/aarch64/aarch64-tune.md: Regenerate.
28453 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28454 (vulcan_addrcost_table): This.
28455 (vulcan_regmove_cost): Rename to ...
28456 (thunderx2t99_regmove_cost): This.
28457 (vulcan_vector_cost): Rename to ...
28458 (thunderx2t99_vector_cost): this.
28459 (vulcan_branch_cost): Rename to ...
28460 (thunderx2t99_branch_cost): This.
28461 (vulcan_tunings): Rename to ...
28462 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28463 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28465 2017-01-09 Martin Jambor <mjambor@suse.cz>
28469 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28470 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28471 (propagate_vr_accross_jump_function): Use the above function for all
28472 value range computations for pass-through jump functions and type
28473 converasion from explicit value range values.
28474 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28475 parameters from TYPE_ARG_TYPES.
28476 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28477 (ipa_write_node_info): Stream type of the actual argument.
28478 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28480 2017-01-09 Martin Liska <mliska@suse.cz>
28483 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28484 (lookup_compiler): Do not show error message with have_E.
28486 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28488 PR tree-optimization/78938
28489 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28490 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28491 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28492 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28495 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28497 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28500 2017-01-09 Richard Biener <rguenther@suse.de>
28502 PR tree-optimization/78997
28503 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28504 name condition properly.
28506 2017-01-09 Richard Biener <rguenther@suse.de>
28509 * dwarf2out.c (is_cxx): New overload with context.
28510 (is_naming_typedef_decl): Use it.
28512 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28514 * invoke.texi (Option Summary): Correct spacing in option lists
28515 and add line breaks to fix over-long lines.
28517 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28519 PR middle-end/17660
28521 * extend.texi (Common Variable Attributes): Add xref to GCC
28522 Internals manual to explain mode attribute keywords.
28524 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28527 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28528 and Preprocessor Options.
28529 (Options for Linking): Document -pthread here....
28530 (RS/6000 and PowerPC Options): ...not here.
28531 (Solaris 2 Options): ...or here.
28532 * doc/cppopts.texi: Document -pthread.
28534 2017-01-08 Martin Sebor <msebor@redhat.com>
28536 PR middle-end/77708
28537 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28538 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28539 New member functions.
28540 (format_directive): Used them.
28542 (pass_sprintf_length::handle_gimple_call): Same.
28543 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28544 to avoid truncation for any argument.
28545 (extract_affine_mul): Same.
28546 * tree.c (get_file_function_name): Same.
28548 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28550 PR middle-end/77484
28551 * predict.def (PRED_INDIR_CALL): Set to 86.
28553 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28555 PR preprocessor/54124
28556 * doc/cppopts.texi: Reformat -d subtable to list the full name
28557 of the options. Add cross-reference to the docs for the general
28558 compiler -d options.
28559 * doc/invoke.texi (Developer Options): Add cross-reference to the
28560 preprocessor-specific -d option documentation.
28562 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28564 PR preprocessor/13498
28565 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28566 redudant material, and reflect new command-line options.
28567 (System Headers): Likewise.
28569 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28571 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28572 -isystem, and -idirafter. Copy-edit.
28573 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28574 default for -ftrack-macro-expansion. Delete obsolete and
28575 badly-formatted implementation details about -fdebug-cpp output.
28576 * doc/cppwarnopts.texi: Copy-edit.
28578 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28581 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28582 that the transition from a max line width >= 1<<10 to narrower
28583 lines works correctly.
28585 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28587 * doc/options.texi (PerFunction): New.
28588 * opt-functions.awk (switch_flags): Map both Optimization and
28589 PerFunction to CL_OPTIMIZATION.
28590 * opth-gen.awk: Test for PerFunction flag along with
28592 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28593 it only when the latter is present. Skip those that don't in
28594 the hash function generator.
28595 * common.opt (fvar-tracking): Mark as PerFunction instead of
28597 (fvar-tracking-assignments): Likewise.
28598 (fvar-tracking-assignments-toggle): Likewise.
28599 (fvar-tracking-uninit): Likewise.
28601 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28603 PR translation/79018
28604 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28607 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28610 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28611 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28612 TARGET_LONG_JUMP_TABLE_OFFSETS.
28613 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28614 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28615 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28616 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28617 * config/m68k/m68k.md (tablejump expander): Likewise.
28618 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28619 TARGET_LONG_JUMP_TABLE_OFFSETS.
28620 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28621 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28623 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28624 David Holsgrove <david.holsgrove@xilinx.com>
28626 * common/config/microblaze/microblaze-common.c
28627 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28628 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28630 * config/microblaze/microblaze.c (microblaze_must_save_register)
28631 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28633 (microblaze_eh_return): New function.
28634 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28635 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28636 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28637 * config/microblaze/microblaze.md (eh_return): New pattern.
28639 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28641 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28642 GCC_DIAGNOSTIC_STRINGIFY): Define.
28644 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28646 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28648 * config/arm/arm.md (<mcrr>): New.
28650 * config/arm/arm.c (arm_arch5te): New.
28651 (arm_option_override): Set arm_arch5te.
28652 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28654 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28655 (arm_mcrr_qualifiers): ... this. New.
28656 (MRRC_QUALIFIERS): Define to...
28657 (arm_mrrc_qualifiers): ... this. New.
28658 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28660 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28661 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28662 (MRRCI, mrrc, MRRC): New.
28663 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28664 VUNSPEC_MRRC2): New.
28666 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28668 * config/arm/arm.md (<mcr>): New.
28670 * config/arm/arm.c (arm_coproc_builtin_available): Add
28671 support for mcr, mrc, mcr2 and mrc2.
28672 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28673 (arm_mcr_qualifiers): ... this. New.
28674 (MRC_QUALIFIERS): Define to ...
28675 (arm_mrc_qualifiers): ... this. New.
28676 (MCR_QUALIFIERS): Define to ...
28677 (arm_mcr_qualifiers): ... this. New.
28678 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28680 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28681 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28682 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28683 VUNSPEC_MRC2): New.
28685 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28687 * config/arm/arm.md (*ldc): New.
28691 * config/arm/arm.c (arm_coproc_builtin_available): Add
28692 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28693 (arm_coproc_ldc_stc_legitimate_address): New.
28694 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28695 'qualifier_const_pointer'.
28696 (LDC_QUALIFIERS): Define to...
28697 (arm_ldc_qualifiers): ... this. New.
28698 (STC_QUALIFIERS): Define to...
28699 (arm_stc_qualifiers): ... this. New.
28700 * config/arm/arm-protos.h
28701 (arm_coproc_ldc_stc_legitimate_address): New.
28702 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28703 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28704 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28705 stc2, stcl, stc2l): New.
28706 * config/arm/constraints.md (Uz): New.
28707 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28708 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28709 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28710 VUNSPEC_STC2L): New.
28712 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28714 * config/arm/arm.md (<cdp>): New.
28715 * config/arm/arm.c (neon_const_bounds): Rename this ...
28716 (arm_const_bounds): ... this.
28717 (arm_coproc_builtin_available): New.
28718 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28719 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28720 (CDP_QUALIFIERS): Define to...
28721 (arm_cdp_qualifiers): ... this. New.
28723 (arm_expand_builtin_args): Add case for 6 arguments.
28724 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28725 (arm_const_bounds): ... this.
28726 (arm_coproc_builtin_available): New.
28727 * config/arm/arm_acle.h (__arm_cdp): New.
28729 * config/arm/arm_acle_builtins.def (cdp): New.
28731 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28732 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28733 'arm_const_bounds'.
28734 * config/arm/types.md (coproc): New.
28735 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28736 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28737 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28738 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28740 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28742 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28743 (UBINOP_QUALIFIERS): New.
28745 (acle_builtin_data): New. Change comment.
28746 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28747 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28748 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28749 arm_acle_builtins.def.
28750 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28751 (arm_init_acle_builtins): New.
28752 (CRC32_BUILTIN): Remove.
28753 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28754 crc32cb, crc32ch and crc32cw.
28755 (arm_init_crc32_builtins): Remove.
28756 (arm_init_builtins): Use arm_init_acle_builtins rather
28757 than arm_init_crc32_builtins.
28758 (arm_expand_acle_builtin): New.
28759 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28760 * config/arm/arm_acle_builtins.def: New.
28762 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28764 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28765 (arm_builtin_datum): ... this.
28766 (arm_init_neon_builtin): Rename to ...
28767 (arm_init_builtin): ... this. Add a new parameters PREFIX
28768 and USE_SIG_IN_NAME.
28769 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28770 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28771 'arm_builtin_datum'.
28772 (arm_init_vfp_builtins): Likewise.
28773 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28774 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28775 (arm_expand_neon_args): Rename to ...
28776 (arm_expand_builtin_args): ... this. Rename builtin_arg
28777 enum values and differentiate between ARG_BUILTIN_MEMORY
28778 and ARG_BUILTIN_NEON_MEMORY.
28779 (arm_expand_neon_builtin_1): Rename to ...
28780 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28781 values, arm_expand_builtin_args and add bool parameter NEON.
28782 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28783 (arm_expand_vfp_builtin): Likewise.
28784 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28786 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28788 PR middle-end/77484
28789 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28790 * predict.c (tree_estimate_probability_bb): Reverse direction of
28791 polymorphic call predictor.
28793 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28795 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28796 (determine_pass_name_match): ...this new function and...
28797 (should_skip_pass_p): ...this new function.
28799 2017-01-06 Nathan Sidwell <nathan@acm.org>
28801 * ipa-visibility.c (function_and_variable_visibility): Reformat
28802 comments and long lines. Remove extrneous if.
28803 * symtab.c (symtab_node::make_decl_local): Fix code format.
28804 (symtab_node::set_section_for_node): Fix comment typo.
28806 2017-01-06 Martin Liska <mliska@suse.cz>
28809 * lra-constraints.c: Rename invariant to lra_invariant.
28810 * predict.c (set_even_probabilities): Initialize e to NULL.
28812 2017-01-05 Martin Sebor <msebor@redhat.com>
28814 PR tree-optimization/78910
28815 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28816 (format_integer): Correct off-by-one error in the handling
28817 of precision with negative numbers in signed conversions..
28819 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28821 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28823 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28825 PR tree-optimization/71016
28826 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28827 factor_out_conditional_conversion. Formatting fix.
28828 (factor_out_conditional_conversion): Add cond_stmt argument.
28829 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28830 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28833 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28835 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28836 read-rtl-function.o, and selftest-rtl.o.
28837 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28838 (selftest::aarch64_test_loading_full_dump): New function.
28839 (selftest::aarch64_run_selftests): New function.
28840 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28841 selftest::aarch64_run_selftests.
28842 * config/i386/i386.c
28843 (selftest::ix86_test_loading_dump_fragment_1): New function.
28844 (selftest::ix86_test_loading_call_insn): New function.
28845 (selftest::ix86_test_loading_full_dump): New function.
28846 (selftest::ix86_test_loading_unspec): New function.
28847 (selftest::ix86_run_selftests): Call the new functions.
28848 * emit-rtl.c (maybe_set_max_label_num): New function.
28849 * emit-rtl.h (maybe_set_max_label_num): New decl.
28850 * function.c (instantiate_decls): Guard call to
28851 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28852 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28854 * gensupport.c (gen_reader::gen_reader): Pass "false"
28855 for new "compact" param of rtx_reader.
28856 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28857 rather than an empty string for NULL strings.
28858 * read-md.c: Potentially include config.h rather than bconfig.h.
28859 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28860 (have_error): New global, copied from errors.c.
28861 (md_reader::read_name): Rename to...
28862 (md_reader::read_name_1): ...this, adding "out_loc" param,
28863 and converting "missing name or number" to returning false, rather
28865 (md_reader::read_name): Reimplement in terms of read_name_1.
28866 (md_reader::read_name_or_nil): New function.
28867 (md_reader::read_string): Handle "(nil)" by returning NULL.
28868 (md_reader::md_reader): Add new param "compact".
28869 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28870 (md_reader::read_file): New method.
28871 * read-md.h (md_reader::md_reader): Add new param "compact".
28872 (md_reader::read_file): New method.
28873 (md_reader::is_compact): New accessor.
28874 (md_reader::read_name): Convert return type from void to file_location.
28875 (md_reader::read_name_or_nil): New decl.
28876 (md_reader::read_name_1): New decl.
28877 (md_reader::m_compact): New field.
28878 (noop_reader::noop_reader): Pass "false" for new "compact" param
28880 (rtx_reader::rtx_reader): Add new "compact" param.
28881 (rtx_reader::read_rtx_operand): Make virtual and convert return
28882 type from void to rtx.
28883 (rtx_reader::read_until): New decl.
28884 (rtx_reader::handle_any_trailing_information): New virtual function.
28885 (rtx_reader::postprocess): New virtual function.
28886 (rtx_reader::finalize_string): New virtual function.
28887 (rtx_reader::m_in_call_function_usage): New field.
28888 (rtx_reader::m_reuse_rtx_by_id): New field.
28889 * read-rtl-function.c: New file.
28890 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28891 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28892 (selftest::verify_three_block_rtl_cfg): New decl.
28893 * read-rtl-function.h: New file.
28894 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28895 For host, include function.h, memmodel.h, and emit-rtl.h.
28896 (one_time_initialization): New function.
28897 (struct compact_insn_name): New struct.
28898 (compact_insn_names): New array.
28899 (find_code): Handle insn codes in compact dumps.
28900 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28901 (bind_subst_iter_and_attr): Likewise.
28902 (add_condition_to_string): Likewise.
28903 (add_condition_to_rtx): Likewise.
28904 (apply_attribute_uses): Likewise.
28905 (add_current_iterators): Likewise.
28906 (apply_iterators): Likewise.
28907 (initialize_iterators): Guard usage of apply_subst_iterator with
28908 #ifdef GENERATOR_FILE.
28909 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28910 (md_reader::read_mapping): Likewise.
28911 (add_define_attr_for_define_subst): Likewise.
28912 (add_define_subst_attr): Likewise.
28913 (read_subst_mapping): Likewise.
28914 (check_code_iterator): Likewise.
28915 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28917 (one_time_initialization): New function.
28918 (rtx_reader::read_until): New method.
28919 (read_flags): New function.
28920 (parse_reg_note_name): New function.
28921 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28922 Handle reuse_rtx ids.
28923 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28924 Add parsing support for RTL dumps, mirroring the special-cases in
28925 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28926 values, and calling handle_any_trailing_information.
28927 (rtx_reader::read_rtx_operand): Convert return type from void
28928 to rtx, returning return_rtx. Handle case 'e'. Call
28929 finalize_string on XSTR and XTMPL fields.
28930 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28931 "(nil)" values were omitted. Call the postprocess vfunc on the
28933 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28934 class ctor. Initialize m_in_call_function_usage. Call
28935 one_time_initialization.
28936 * rtl-tests.c (selftest::test_uncond_jump): Call
28937 set_new_first_and_last_insn.
28938 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28939 * selftest-rtl.c: New file.
28940 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28941 (selftest::get_insn_by_uid): New decl.
28942 * selftest-run-tests.c (selftest::run_tests): Call
28943 read_rtl_function_c_tests.
28944 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28945 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28948 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28950 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28951 operands in a special way. Assert that pos+len <= mode precision.
28953 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28955 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28956 3 argument Alias with unlimited for the negative form.
28957 (fno-vect-cost-model): Removed.
28959 2017-01-05 Martin Liska <mliska@suse.cz>
28961 * hsa-gen.c (gen_hsa_divmod): New function.
28962 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28964 2017-01-05 Martin Liska <mliska@suse.cz>
28967 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28970 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28972 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28973 small constant length operands.
28975 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28977 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28978 between loop iterations.
28980 2017-01-05 Martin Liska <mliska@suse.cz>
28983 * gimplify.c (gimplify_decl_expr): Compare to
28984 asan_poisoned_variables instread of checking flags.
28985 (gimplify_target_expr): Likewise.
28986 (gimplify_expr): Likewise.
28987 (gimplify_function_tree): Conditionally initialize
28988 asan_poisoned_variables.
28990 2017-01-04 Jeff Law <law@redhat.com>
28992 PR tree-optimizatin/78812
28993 * rtl.h (contains_mem_rtx_p): Prototype.
28994 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28995 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28996 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28997 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28998 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29000 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29002 * input.c (assert_char_at_range): Default-initialize actual_range.
29004 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29006 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29007 to match the caller.
29009 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29011 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29012 insns after final jump in test to emit dummy move.
29014 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29016 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29017 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29019 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29021 * multiple_target.c (create_dispatcher_calls): Init e_next.
29022 * tree-ssa-loop-split.c (split_loop): Init border.
29023 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29026 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29031 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29032 (altivec_register_operand): Do not return true if the operand
29033 contains a SUBREG mixing SImode and SFmode.
29034 (vsx_register_operand): Likewise.
29035 (vsx_reg_sfsubreg_ok): New predicate.
29036 (vfloat_operand): Do not return true if the operand contains a
29037 SUBREG mixing SImode and SFmode.
29038 (vint_operand): Likewise.
29039 (vlogical_operand): Likewise.
29040 (gpc_reg_operand): Likewise.
29041 (int_reg_operand): Likewise.
29042 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29043 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29044 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29046 (rs6000_emit_move_si_sf_subreg): New helper function.
29047 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29048 fixup SUBREGs involving SImode and SFmode.
29049 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29050 numbers for the new peephole2 optimization.
29051 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29052 the GLIBC math library that do AND/IOR/XOR operations on single
29053 precision floating point.
29054 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29055 target macros to say whether we need to avoid SUBREGs mixing
29057 (TARGET_ALLOW_SF_SUBREG): Likewise.
29058 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29059 (UNSPEC_SI_FROM_SF): Likewise.
29060 (iorxor): Change spacing.
29061 (and_ior_xor): New iterator for AND, IOR, and XOR.
29062 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29063 (movdi_from_sf_zero_ext): Likewise.
29064 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29065 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29066 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29067 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29068 (fms<mode>4): Likewise.
29069 (fnma<mode>4): Likewise.
29070 (fnms<mode>4): Likewise.
29071 (nfma<mode>4): Likewise.
29072 (nfms<mode>4): Likewise.
29074 2017-01-04 Marek Polacek <polacek@redhat.com>
29077 * doc/invoke.texi: Document -Wpointer-compare.
29079 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29081 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29084 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29085 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29088 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29091 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29092 documentation of the powerpc_popcntb_ok attribute.
29093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29094 code to issue warning messages if a requested CPU configuration is
29095 not supported by the binary (assembler and loader) toolchain.
29096 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29097 made to define a built-in function that has been disabled.
29098 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29099 made to define a built-in function that has been disabled.
29100 (altivec_init_builtins): Add comment explaining why definition
29101 of the DST built-in functions is not preceded by an assertion
29102 check. Add assertions to prevent ICE if attempts are made to
29103 define an altivec predicate or an abs* built-in function that has
29105 (htm_init_builtins): Add comment explaining why definition of the
29106 htm built-in functions is not preceded by an assertion check.
29108 2017-01-04 Jeff Law <law@redhat.com>
29110 PR tree-optimizatin/67955
29111 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29112 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29113 the points-to solution does not include pt_null. Use DECL_PT_UID
29116 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29118 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29119 Use gen_int_mode instead of gen_lopwart for const_int operands.
29121 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29123 PR tree-optimization/71563
29124 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29125 out of range value - has low bits known to be zero.
29127 2017-01-04 Alan Modra <amodra@gmail.com>
29129 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29130 * configure: Regenerate.
29131 * config.in: Regenerate.
29133 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29136 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29137 a substring of the message, but strcmp with the whole message. Ifdef
29138 ENABLE_NLS, translate the message first using dgettext.
29140 2017-01-03 Jeff Law <law@redhat.com>
29142 PR tree-optimizatin/78856
29143 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29144 (mark_threaded_blocks): Remove code to truncate thread paths that
29145 cross multiple loop headers. Instead invalidate the cached loop
29146 iteration information and handle case of a thread path walking
29147 into an irreducible region.
29149 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29152 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29153 assertions. Add support for doing the signbit if the IEEE 128-bit
29154 floating point value is in a GPR.
29155 * config/rs6000/rs6000.md (Fsignbit): Delete.
29156 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29157 Update the length attribute if the value is in a GPR.
29158 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29159 the sign or zero extension instruction, since the value is always 0/1.
29160 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29163 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29164 extracting SImode to a GPR register so that we can generate a
29165 store, limit the vector to be in a traditional Altivec register
29166 for the vextuwrx instruction.
29168 2017-01-03 Ian Lance Taylor <iant@google.com>
29170 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29172 2017-01-03 Martin Sebor <msebor@redhat.com>
29174 PR tree-optimization/78696
29175 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29176 precision. Use MPFR for %f for greater fidelity. Correct handling
29178 (pass_sprintf_length::compute_format_length): Set width and precision
29179 specified by asrerisk to void_node for vararg functions.
29180 (try_substitute_return_value): Adjust dump output.
29182 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29184 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29186 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29188 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29189 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29190 -mlra/-mno-lra was passed to the compiler.
29192 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29194 PR rtl-optimization/65618
29195 * emit-rtl.c (try_split): Move initialization of "before" and
29196 "after" to just before the call to emit_insn_after_setloc.
29198 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29200 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29202 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29204 * dwarf2out.c (gen_enumeration_type_die): When
29205 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29207 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29209 PR tree-optimization/78965
29210 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29211 Change first argument from const call_info & to call_info &. For %n
29212 set info.nowrite to false.
29214 PR middle-end/78901
29215 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29216 possibly throwing calls.
29218 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29219 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29220 and fns handling, rather than in a separate case SSA_NAME.
29222 2017-01-02 Jeff Law <law@redhat.com>
29224 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29225 fixes for first_period and second_period variables.
29227 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29230 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29231 (*insvqi_1): New insn pattern.
29232 (*insvqi_1_mem_rex64): Ditto.
29233 (*insvqi_2): Ditto.
29234 (*insvqi_3): Rename from *insvqi.
29236 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29238 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29240 * doc/cfg.texi (Edges): Remove reference to Java.
29241 (Maintaining the CFG): Ditto.
29243 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29245 PR middle-end/77674
29246 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29247 transparent aliases.
29249 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29251 PR middle-end/77484
29252 * predict.def (PRED_CALL): Update hitrate.
29253 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29254 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29255 into direct/indirect/polymorphic variants.
29257 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29259 Update copyright years.
29261 * gcc.c (process_command): Update copyright notice dates.
29262 * gcov-dump.c (print_version): Ditto.
29263 * gcov.c (print_version): Ditto.
29264 * gcov-tool.c (print_version): Ditto.
29265 * gengtype.c (create_file): Ditto.
29266 * doc/cpp.texi: Bump @copying's copyright year.
29267 * doc/cppinternals.texi: Ditto.
29268 * doc/gcc.texi: Ditto.
29269 * doc/gccint.texi: Ditto.
29270 * doc/gcov.texi: Ditto.
29271 * doc/install.texi: Ditto.
29272 * doc/invoke.texi: Ditto.
29274 Copyright (C) 2017 Free Software Foundation, Inc.
29276 Copying and distribution of this file, with or without modification,
29277 are permitted in any medium without royalty provided the copyright
29278 notice and this notice are preserved.