1 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
7 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12 2017-10-09 Jakub Jelinek <jakub@redhat.com>
15 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
16 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
18 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
20 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
21 (WI_BINARY_PREDICATE_RESULT): Likewise.
22 (wi::binary_traits::operator_result): New type.
23 (wi::binary_traits::predicate_result): Likewise.
24 (generic_wide_int::operator~, unary generic_wide_int::operator-)
25 (generic_wide_int::operator==, generic_wide_int::operator!=)
26 (generic_wide_int::operator&, generic_wide_int::and_not)
27 (generic_wide_int::operator|, generic_wide_int::or_not)
28 (generic_wide_int::operator^, generic_wide_int::operator+
29 (binary generic_wide_int::operator-, generic_wide_int::operator*):
31 (operator~, unary operator-, operator==, operator!=, operator&)
32 (operator|, operator^, operator+, binary operator-, operator*): New
34 * expr.c (get_inner_reference): Use wi::bit_and_not.
35 * fold-const.c (fold_binary_loc): Likewise.
36 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
37 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
38 (bit_value_binop): Likewise.
39 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
40 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
41 (extract_range_from_binary_expr_1): Likewise.
42 (masked_increment): Likewise.
43 (simplify_bit_ops_using_ranges): Likewise.
45 2017-10-09 Martin Jambor <mjambor@suse.cz>
48 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
49 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
50 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
52 (hsa_fixup_mov_insn_type): New function.
53 (hsa_op_with_type::get_in_type): Use it.
54 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
55 immediates in an assert.
56 (hsa_op_with_type::extend_int_to_32bit): New method.
57 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
58 types. Convert to dest type if necessary.
59 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
60 (reg_for_gimple_ssa): Pass false as min32int to
61 hsa_type_for_scalar_tree_type.
62 (gen_hsa_addr): Fixup type when creating addresable temporary.
63 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
64 (gen_hsa_unary_operation): Extend operands and convert to dest type if
65 necessary. Call hsa_fixup_mov_insn_type.
66 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
67 extend operands and convert to dest type if necessary.
68 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
69 to dest type if necessary.
70 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
71 if conversion nt necessary and size matches.
72 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
73 to dest type if necessary.
74 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
75 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
77 (gen_hsa_clrsb): Likewise.
78 (gen_hsa_ffs): Likewise.
79 (gen_hsa_divmod): Extend operands and convert to dest type if
81 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
83 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
85 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
86 Remove empty default arguments. Use a brace block as output
88 (conditional return): Ditto.
90 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
91 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
93 (group_ending_nop): Ditto.
95 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
96 (splitters for those): Ditto.
98 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
100 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
101 a conditional jump (and the compare for it) so that pc_rtx is the
103 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
104 for the deleted and renamed ctr<mode>_internal[234] patterns.
105 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
106 Delete second conditional return pattern.
107 (ctr<mode>_internal2): Delete this second bdnz pattern.
108 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
109 (ctr<mode>_internal4): Delete this second bdz pattern.
111 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
113 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
114 (always_initialized_rtx_for_ssa_name_p): New predicate.
115 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
116 (finish_out_of_ssa): Free new field of SA.
117 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
118 * tree-ssa-coalesce.c: Include tree-ssa.h.
119 (get_parm_default_def_partitions): Remove extern keyword.
120 (get_undefined_value_partitions): New function.
121 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
122 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
124 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
126 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
128 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
130 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
132 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
135 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
137 * i386.c (has_dispatch): Disable for Ryzen.
139 2017-10-08 Olivier Hainque <hainque@adacore.com>
141 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
142 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
144 (thumb_set_return_address): Likewise.
146 2017-10-08 Olivier Hainque <hainque@adacore.com>
148 * common/config/arm/arm-common.c (arm_except_unwind_info):
149 Handle DWARF2_UNWIND_INFO.
151 2017-10-07 Michael Collison <michael.collison@arm.com>
153 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
156 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
158 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
159 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
160 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
161 defined, force the creation of a new block for a dispatch label.
163 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
165 * invoke.texi (Wsuggest-attribute=cold): Document.
166 * common.opt (Wsuggest-attribute=cold): New
167 * ipa-pure-const.c (warn_function_cold): New function.
168 * predict.c (compute_function_frequency): Use it.
169 * predict.h (warn_function_cold): Declare.
171 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
173 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
176 2017-10-06 Martin Liska <mliska@suse.cz>
178 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
179 keyword for member functions.
180 (struct sanopt_tree_couple): New struct.
181 (struct sanopt_tree_couple_hash): New function.
182 (struct sanopt_ctx): Add new hash_map.
183 (has_dominating_ubsan_ptr_check): New function.
184 (record_ubsan_ptr_check_stmt): Likewise.
185 (maybe_optimize_ubsan_ptr_ifn): Likewise.
186 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
187 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
189 2017-10-06 Sudakshina Das <sudi.das@arm.com>
192 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
193 aarch64_simd_valid_immediate on CONST_VECTORs.
194 (aarch64_reg_or_bic_imm): Likewise.
196 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
198 PR rtl-optimization/82396
199 * haifa-sched.c (ready_sort_real): Disable qsort checking.
201 2017-10-06 Sebastian Pop <sebpop@gmail.com>
203 * graphite-dependences.c (scop_get_reads): Move code to...
204 (scop_get_must_writes): Move code to...
205 (scop_get_may_writes): Move code to...
206 (scop_get_reads_and_writes): ... here.
207 (scop_get_dependences): Call scop_get_reads_and_writes.
209 2017-10-06 Jakub Jelinek <jakub@redhat.com>
211 PR tree-optimization/82434
212 * fold-const.h (can_native_encode_type_p,
213 can_native_encode_string_p): Remove.
214 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
215 don't encode anything, just return what would be otherwise returned.
216 (native_encode_fixed, native_encode_complex, native_encode_vector):
218 (native_encode_string): Likewise. Inline by hand
219 can_native_encode_string_p.
220 (can_native_encode_type_p): Remove.
221 (can_native_encode_string_p): Remove.
222 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
223 STRING_CSTs using can_native_encode_string_p, test all
224 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
225 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
226 argument from native_encode_expr.
227 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
228 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
231 2017-10-06 Richard Biener <rguenther@suse.de>
233 PR tree-optimization/82397
234 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
235 operand_equal_p but rely on data_ref_compare_tree for detecting
237 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
238 to match up with dr_group_sort_cmp.
240 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
243 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
245 * config/s390/s390-builtin-types.def: Regenerate.
247 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
250 * config/s390/s390-builtin-types.def: Regenerate.
251 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
252 Change flag from B_VXE to B_VX.
253 (s390_vec_min_dbl): Remove B_VXE flag.
255 2017-10-06 Richard Biener <rguenther@suse.de>
257 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
258 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
259 translate_isl_ast_to_gimple::is_valid_rename,
260 translate_isl_ast_to_gimple::get_rename,
261 translate_isl_ast_to_gimple::get_def_bb_for_const,
262 translate_isl_ast_to_gimple::get_new_name,
263 translate_isl_ast_to_gimple::collect_all_ssa_names,
264 translate_isl_ast_to_gimple::copy_loop_phi_args,
265 translate_isl_ast_to_gimple::collect_all_ssa_names,
266 translate_isl_ast_to_gimple::copy_loop_phi_args,
267 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
268 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
269 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
270 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
271 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
272 translate_isl_ast_to_gimple::copy_cond_phi_args,
273 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
274 translate_isl_ast_to_gimple::edge_for_new_close_phis,
275 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
276 translate_isl_ast_to_gimple::rename_uses,
277 translate_isl_ast_to_gimple::rename_all_uses): Remove.
278 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
279 (set_rename_for_each_def): Likewise.
280 (graphite_copy_stmts_from_block): Handle debug stmt resetting
281 here. Handle rewriting SCEV analyzable uses here.
282 (copy_bb_and_scalar_dependences): Generate code for PHI
284 (graphite_regenerate_ast_isl): Adjust.
285 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
286 (add_write, add_read): New functions.
287 (build_cross_bb_scalars_def): Use it and simplify.
288 (build_cross_bb_scalars_use): Likewise.
289 (graphite_find_cross_bb_scalar_vars): Inline into...
290 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
291 simulating out-of-SSA. Compute liveout and add dependencies.
292 (build_scops): Force an empty entry block.
293 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
295 (sese_build_liveouts): Declare.
296 (sese_trivially_empty_bb_p): Likewise.
297 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
298 compute liveout and debug_liveout.
299 (sese_bad_liveouts_use): Remove.
300 (sese_reset_debug_liveouts_bb): Likewise.
301 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
302 (sese_build_liveouts): Build liveout and debug_liveout and store
304 (new_sese_info): Adjust.
305 (free_sese_info): Likewise.
306 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
307 do not build liveout here.
308 (move_sese_in_condition): Adjust region entry.
309 (scev_analyzable_p): Match up with chrec_apply requirements.
310 (sese_trivially_empty_bb_p): New.
311 * tree-into-ssa.c (get_reaching_def): Properly support generating
312 default-defs for incremental rewrite of anonymous names.
314 2017-10-06 Richard Biener <rguenther@suse.de>
316 * graphite-sese-to-poly.c (extract_affine): For casts increasing
317 precision do not perform modulo reduction.
319 2017-10-06 Richard Biener <rguenther@suse.de>
321 PR tree-optimization/82436
322 * tree-vect-slp.c (vect_supported_load_permutation_p): More
323 conservatively choose the vectorization factor when checking
324 whether we can perform the required load permutation.
325 (vect_transform_slp_perm_load): Assert when we may not fail.
327 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
329 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
330 message for incompatible -msdata=* and -mcall-* options.
332 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
334 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
335 for post-reload scheduling.
337 2017-10-05 Tamar Christina <tamar.christina@arm.com>
339 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
341 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
343 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
344 to improve monte carlo in scimark.
346 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
348 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
349 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
350 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
351 pentium4_cost, nocona_cost): Set reassociation width to 1.
352 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
353 width to 2 for fp operations and 1 otherwise.
354 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
356 (atom_cost): Set reassociation width to 2.
357 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
358 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
359 (core_cost): Set fp reassociation width to 4 and vector to 2.
360 (ix86_reassociation_width): Rewrite using cost table; special case
361 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
362 and TARGET_AVX128_OPTIMAL.
363 * i386.h (processor_costs): Add
364 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
365 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
366 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
367 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
368 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
369 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
371 2017-10-05 Nathan Sidwell <nathan@acm.org>
373 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
375 2017-10-05 Tamar Christina <tamar.christina@arm.com>
377 * config/arm/arm.c (arm_test_fpu_data): New.
378 (arm_run_selftests): Call arm_test_fpu_data.
380 2017-10-04 Nathan Sidwell <nathan@acm.org>
382 * toplev.c (toplev::main): Remove excess parens on pretty_printer
384 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
386 2017-10-04 Sudakshina Das <sudi.das@arm.com>
388 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
389 check type for aarch64_simd_valid_immediate.
390 (aarch64_output_simd_mov_immediate): Update prototype.
391 (aarch64_simd_valid_immediate): Update prototype.
392 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
393 support for ORR-immediate.
394 (and<mode>3): modified pattern to add support for BIC-immediate.
395 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
396 now checks for valid immediate for BIC and ORR based on new enum
398 (aarch64_output_simd_mov_immediate): Function now used to output
399 BIC/ORR imm as well based on new enum argument.
400 * config/aarch64/constraints.md (Do): New vector immediate constraint.
402 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
403 (aarch64_reg_or_bic_imm): Likewise.
405 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
407 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
408 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
410 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
414 PR rtl-optimization/82396
415 * haifa-sched.c (autopref_multipass_init): Simplify
417 (autopref_rank_data): Simplify sort order.
418 * sched-int.h (autopref_multipass_data_): Remove
419 multi_mem_insn_p, min_offset and max_offset.
421 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
423 * doc/sourcebuild.texi: Document vect_peeling_profitable.
425 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
427 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
430 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
432 * doc/sourcebuild.texi: Document vect_long_mult.
434 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
436 PR tree-optimization/82413
437 * fold-const.c (build_range_check): Use widest_int when comparing
438 the maximum ETYPE value with HIGH.
440 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
442 PR rtl-optimization/82396
443 * haifa-sched.c (autopref_multipass_init): Simplify
445 (autopref_rank_data): Simplify sort order.
446 * sched-int.h (autopref_multipass_data_): Remove
447 multi_mem_insn_p, min_offset and max_offset.
449 2017-10-04 Jakub Jelinek <jakub@redhat.com>
451 PR tree-optimization/82381
452 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
453 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
456 PR tree-optimization/82374
457 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
458 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
459 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
460 current_function_decl to the new decl.
462 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
464 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
465 helper macro for IEEE float128 hardware built-in functions.
466 (SQRTF128_ODD): Add built-in functions with the round-to-odd
468 (TRUNCF128_ODD): Likewise.
469 (ADDF128_ODD): Likewise.
470 (SUBF128_ODD): Likewise.
471 (MULF128_ODD): Likewise.
472 (DIVF128_ODD): Likewise.
473 (FMAF128_ODD): Likewise.
474 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
475 UNSPEC_TRUNC_ROUND_TO_ODD.
476 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
477 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
478 floating point round to odd instructions.
479 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
480 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
481 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
482 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
483 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
484 (trunc<mode>sf2_hw): Change the truncate with round to odd
485 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
486 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
487 to odd hardware instructions.
488 (sub<mode>3_odd): Likewise.
489 (mul<mode>3_odd): Likewise.
490 (div<mode>3_odd): Likewise.
491 (sqrt<mode>2_odd): Likewise.
492 (fma<mode>4_odd): Likewise.
493 (fms<mode>4_odd): Likewise.
494 (nfma<mode>4_odd): Likewise.
495 (nfms<mode>4_odd): Likewise.
496 (trunc<mode>df2_odd): Change the truncate with round to odd
497 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
499 * doc/extend.texi (PowerPC built-in functions): Update documentation
500 for existing IEEE float128-bit built-in functions. Add built-in
501 functions that generate the IEEE 128-bit floating point round to
504 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
506 PR rtl-optimization/77729
507 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
508 to (X&(C1&~C2))|C2 transformations.
510 2017-10-03 Martin Jambor <mjambor@suse.cz>
512 PR tree-optimization/82363
513 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
514 mismatch, mark lacc written regardless of racc.
516 2017-10-03 Jakub Jelinek <jakub@redhat.com>
518 PR tree-optimization/82381
519 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
520 stmt_to_insert nor wheather SSA_NAMEs are default defs.
521 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
522 fallthrough into reassoc_stmt_dominates_stmt_p.
525 * combine.c (combine_instructions): Don't combine in unreachable
528 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
531 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
532 function to not use the have_cpu variable. Do not set cpu_index,
533 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
535 (rs6000_valid_attribute_p): Remove duplicate initializations of
536 old_optimize and func_optimize.
537 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
538 (rs6000_activate_target_options): Make global.
539 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
542 2017-10-02 Jakub Jelinek <jakub@redhat.com>
544 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
545 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
546 Set *poffset to 0 and *psize and *pmax_size to -1 if
547 *poffset + *psize overflows in HOST_WIDE_INT.
549 PR tree-optimization/82387
550 PR tree-optimization/82388
551 PR tree-optimization/82389
552 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
553 instead of live_bytes non-NULL.
555 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
558 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
561 2017-10-02 Richard Biener <rguenther@suse.de>
563 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
564 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
565 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
567 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
569 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
570 requested precision matches the type's.
571 * calls.c (alloc_max_size): Calculate the new candidate size as
572 a widest_int and use wi::to_widest when comparing it with the
573 current candidate size.
574 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
575 zero rather than integer_zero_node.
576 * match.pd: Check for a no-op conversion before using wi::add
577 rather than after. Use tree_to_uhwi when summing small shift
578 counts into an unsigned int.
580 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
581 Alan Hayward <alan.hayward@arm.com>
582 David Sherwood <david.sherwood@arm.com>
585 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
586 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
587 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
590 2017-10-02 Richard Biener <rguenther@suse.de>
592 PR tree-optimization/82355
593 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
594 a mapping for the enclosing loop but avoid generating one for
596 (copy_bb_and_scalar_dependences): Remove premature codegen
597 error on PHIs in blocks duplicated into multiple places.
598 * graphite-scop-detection.c
599 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
600 in the region use it as loop and nest to analyze the DR in.
601 (try_generate_gimple_bb): Likewise.
602 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
603 (add_loop_constraints): For blocks in a loop not in the region
604 create a dimension with a single iteration.
605 * sese.h (gbb_loop_at_index): Remove assert.
607 2017-10-01 Kevin Buettner <kevinb@redhat.com>
609 * omp-expand.c (adjust_context_scope): New function.
610 (expand_parallel_call): Call adjust_context_scope.
612 2017-10-01 Jeff Law <law@redhat.com>
614 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
615 dom_opt_dom_walker class with direct access to private members.
616 Add comments. Call test_for_singularity.
617 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
618 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
619 m_dummy_cond anymore.
620 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
622 (pass_dominator:execute): Build the dummy_cond here and pass it
623 to the dom_opt_dom_walker ctor.
624 (test_for_singularity): New function.
626 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
627 Maya Rashish <coypu@sdf.org>
629 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
630 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
631 (alpha*-*-netbsd*) Use nbsd_tm_file.
632 (arm*-*-netbsdelf*) Likewise.
633 (i[34567]86-*-netbsdelf*) Likewise.
634 (x86_64-*-netbsd*) Likewise.
635 (mips*-*-netbsd*) Likewise.
636 (powerpc-*-netbsd*) Likewise.
637 (sh*-*-netbsd*) Likewise.
638 (sparc-*-netbsdelf*) Likewise.
639 (sparc64-*-netbsd*) Likewise.
640 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
642 (vax-*-netbsdelf*) Likewise.
643 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
644 (UINT_FAST8_TYPE) Likewise.
645 (INT_FAST16_TYPE) Check CHAR_FAST16.
646 (UINT_FAST16_TYPE) Likewise.
648 2017-09-30 Jakub Jelinek <jakub@redhat.com>
651 * config/i386/i386.md
652 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
653 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
654 *divmodsi4_zext_2): New define_insn_and_split.
655 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
656 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
657 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
658 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
659 New define_insn_and_split.
660 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
661 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
662 operands[1] having DImode when mode is SImode.
664 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
665 always SImode for DIV and MOD in REG_EQUAL notes.
667 2017-09-29 Yury Gribov <tetra2005@gmail.com>
670 * match.pd: Fix handling of NaNs in pattern.
672 2017-09-29 Jeff Law <law@redhat.com>
674 * sbitmap.c (bitmap_bit_in_range_p): New function.
675 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
676 * tree-ssa-dse.c (live_bytes_read): New function.
677 (dse_classify_store): Ignore reads of dead bytes.
679 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
680 typos and whitespace errors.
681 * config/i386/predicates.md (address_no_seg_operand): Likewise.
682 * config/s390/s390.c (s390_emit_prologue): Likewise.
684 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
687 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
688 with a symbol for LRA.
690 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
692 PR rtl-optimization/82338
693 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
695 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
697 * genmodes.c (calc_wider_mode): Suppress qsort macro.
698 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
699 (qsort_chk): Declare.
700 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
701 (qsort_chk): New function.
703 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
705 PR tree-optimization/82337
706 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
707 phi definition if the PHI result appears in an abnormal PHI.
708 (find_basis_for_base_expr): Don't record a basis if the LHS of the
709 basis appears in an abnormal PHI.
711 2017-09-29 Richard Biener <rguenther@suse.de>
713 * graphite-isl-ast-to-gimple.c
714 (translate_isl_ast_to_gimple::set_codegen_error): New function.
715 (binary_op_to_tree): Use it.
716 (get_rename_from_scev): Likewise.
717 (copy_loop_phi_nodes): Likewise.
718 (copy_bb_and_scalar_dependences): Likewise.
719 (translate_pending_phi_nodes): Likewise.
721 2017-09-29 Jakub Jelinek <jakub@redhat.com>
724 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
725 for movabsq $(i32 << shift), r64.
727 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
729 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
730 index when encoding %esp as %rsp to avoid 0x67 prefix.
732 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
734 * config/i386/i386.md (*movsf_internal, *movdf_internal):
735 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
737 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
739 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
740 Extensions with more than 16 double VFP registers.
741 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
742 to_clear_mask and all code related to it. Replace the remaining
743 entry by a sbitmap and adapt code accordingly.
745 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
747 * brig-builtins.def: Change pure attributes to const.
749 2017-09-28 Joseph Myers <joseph@codesourcery.com>
751 * config.gcc (default_gnu_indirect_function): Default to yes for
752 sparc*-*-linux* with glibc.
754 2017-09-28 Joseph Myers <joseph@codesourcery.com>
756 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
757 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
758 when creating .init_array and .fini_array sections with priority
761 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
764 * config/aarch64/aarch64.c
765 (aarch64_builtin_support_vector_misalignment): Always return false
766 when misalignment is unknown.
768 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
770 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
771 this function to return false if the definition used by the swap
772 instruction is artificial, or if the memory address from which the
773 constant value is loaded is not represented by a base address held
774 in a register or if the base address register is a frame or stack
775 pointer. Additionally, return false if the base address of the
776 loaded constant is a SYMBOL_REF but is not considered to be a
778 (replace_swapped_load_constant): New function.
779 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
780 loaded constant vector with a load of a swapped constant vector.
782 2017-09-27 Carl Love <cel@us.ibm.com>
784 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
785 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
786 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
789 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
791 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
792 first, always call autopref_rank_data otherwise.
794 2017-09-27 Richard Biener <rguenther@suse.de>
796 * graphite-scop-detection.c (find_scop_parameters): Move
797 loop bound handling ...
798 (gather_bbs::before_dom_children): ... here, avoiding the need
799 to build scop_info->loop_nest.
800 (record_loop_in_sese): Remove.
801 * sese.h (sese_info_t::loop_nest): Remove.
802 * sese.c (new_sese_info): Do not allocate loop_nest.
803 (free_sese_info): Do not free loop_nest.
805 2017-09-27 Jakub Jelinek <jakub@redhat.com>
808 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
809 lhs from calls if the lhs has addressable type.
811 2017-09-27 Richard Biener <rguenther@suse.de>
813 * graphite.h (scop::max_alias_set): New member.
814 * graphite-scop-detection.c: Remove references to non-existing
816 (build_alias_sets): Record the maximum alias set used for drs.
817 (build_scops): Support zero as unlimited for
818 --param graphite-max-arrays-per-scop.
819 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
821 (build_poly_sr_1): ... here. Compute alias set based on the
822 maximum alias set used for drs rather than
823 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
825 2017-09-27 Richard Biener <rguenther@suse.de>
827 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
828 --param loop-block-tile-size=0 to disable tiling.
830 2017-09-27 Richard Biener <rguenther@suse.de>
832 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
833 (graphite-max-nb-scop-params): Document special value zero.
834 * domwalk.h (dom_walker::STOP): New symbolical constant.
835 (dom_walker::dom_walker): Add optional parameter for bb to
837 (dom_walker::~dom_walker): Declare.
838 (dom_walker::before_dom_children): Document STOP return value.
839 (dom_walker::m_user_bb_to_rpo): New member.
840 (dom_walker::m_bb_to_rpo): Likewise.
841 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
842 mapping here if not provided by the user.
843 (dom_walker::~dom_walker): Free bb to RPO mapping if not
844 provided by the user.
845 (dom_walker::STOP): Define.
846 (dom_walker::walk): Do not compute bb to RPO mapping here.
847 Support STOP return value from before_dom_children to stop
849 * graphite-optimize-isl.c (optimize_isl): If the schedule
850 is the same still generate code if -fgraphite-identity
851 or -floop-parallelize-all are given.
852 * graphite-scop-detection.c: Include cfganal.h.
853 (gather_bbs::gather_bbs): Get and pass through bb to RPO
855 (gather_bbs::before_dom_children): Return STOP for BBs
857 (build_scops): Compute bb to RPO mapping and pass it to
858 the domwalk. Treat --param graphite-max-nb-scop-params=0
859 as not limiting the number of params.
860 * graphite.c (graphite_initialize): Remove limit on the
861 number of basic-blocks in a function.
862 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
863 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
866 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
868 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
869 Adjust code to eliminate needing to do the shift right 32-bits
870 operation after XSCVDPSPN.
872 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
874 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
875 ((X / Y) != 0 -> X >= Y): Likewise.
877 2017-09-26 Carl Love <cel@us.ibm.com>
879 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
880 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
881 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
882 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
883 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
884 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
885 definitions and overloading.
886 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
887 statement for P9V_BUILTIN_XST_LEN_R.
888 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
889 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
890 define_expand and define_insn for the instructions and builtins.
891 * doc/extend.texi: Update the built-in documentation file for the new
893 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
894 define_insn for the instructions
896 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
899 * gcc/config/netbsd-protos.h: New file.
900 * gcc/config/netbsd.c: New file.
901 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
902 * gcc/config/t-netbsd: New file.
903 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
904 (tmake_file) Add t-netbsd.
905 (extra_objs) Add netbsd.o.
907 2017-09-26 Janus Weil <janus@gcc.gnu.org>
911 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
913 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
915 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
916 sign extension from a vector register to a GPR by doing a 32-bit
917 direct move and then an EXTSW.
918 (extendsi<mode>2 splitter): Likewise.
919 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
920 right or vector extract after doing XSCVDPSPN. Use
921 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
923 (movdi_from_sf_zero_ext): Likewise.
924 (reload_gpr_from_vsxsf): Likewise.
925 (p8_mfvsrd_4_disf): Delete, no longer used.
926 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
927 then needing to move the SFmode to a GPR to use the XSCVDPSP
928 instruction instead of FRSP and XSCVDPSPN.
929 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
930 it is adjacent to the other XSCVSPDP insns.
931 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
932 SFmode to be in traditional Altivec registers.
933 (vsx_xscvdpspn): Eliminate useless alternative constraint.
934 (vsx_xscvspdpn): Likewise.
935 (vsx_xscvspdpn_scalar): Likewise.
937 2017-09-26 Martin Jambor <mjambor@suse.cz>
939 * tree-sra.c (compare_access_positions): Put integral types first,
940 stabilize sorting of integral types, remove conditions putting
941 non-full-precision integers last.
942 (sort_and_splice_var_accesses): Disable scalarization if a
943 non-integert would be represented by a non-full-precision integer.
945 2017-09-26 Joseph Myers <joseph@codesourcery.com>
947 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
948 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
949 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
950 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
951 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
952 conditionals inside the function instead of around it. Call
953 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
954 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
956 2017-09-26 Richard Biener <rguenther@suse.de>
958 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
960 (scop_detection::build_scop_breadth): ... this. Removed.
961 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
962 (scop_detection::harmful_stmt_in_bb): Likewise.
963 (scop_detection::graphite_can_represent_stmt): Likewise.
964 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
965 (scop_detection::can_represent_loop): Remove recursion, fold in ...
966 (scop_detection::can_represent_loop_1): ... this. Removed.
967 (scop_detection::harmful_loop_in_region): Simplify after inlining
968 the above and remove more quadraticness.
969 (build_scops): Adjust.
970 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
973 2017-09-26 Jakub Jelinek <jakub@redhat.com>
976 * config/i386/i386.c (ix86_print_operand_address_as): Only test
977 REGNO (base) == SP_REG if base is a REG.
980 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
981 if it is different SSA_NAME.
982 (optimize_range_tests_cmp_bitwise): New function.
983 (optimize_range_tests): Call it.
985 2017-09-26 Richard Biener <rguenther@suse.de>
987 PR tree-optimization/82321
988 * graphite.c (canonicalize_loop_closed_ssa): Properly check
989 for the def being inside the loop.
991 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
993 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
995 * config/s390/s390-builtins.def: Fix constraint on op4.
997 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
999 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1000 independent expanders.
1001 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1002 ("vec_ordered", "vec_unordered"): New expanders.
1004 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1006 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1009 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1011 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1012 vec_unpacks_lo_v16qi.
1013 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1015 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1017 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1018 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1019 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1021 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1023 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1025 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1027 ("*vec_slb<mode>"): New insn pattern.
1028 ("vec_shr_<mode>"): New expander.
1029 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1030 and force the shift count operand to V16QImode.
1031 ("vec_srb<mode>"): Set shift count mode to V16QI.
1033 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1035 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1036 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1037 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1039 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1042 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1044 2017-09-26 Richard Biener <rguenther@suse.de>
1046 PR tree-optimization/82320
1047 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1050 2017-09-25 Jeff Law <law@redhat.com>
1052 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1053 prototype for new argument.
1054 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1055 mostly extracted from rs6000_emit_allocate_stack.
1056 (rs6000_emit_probe_stack_range_stack_clash): New function.
1057 (rs6000_emit_allocate_stack): Call
1058 rs6000_emit_probe_stack_range_stack_clash as needed.
1059 (rs6000_emit_probe_stack_range): Add additional argument
1060 to call to gen_probe_stack_range{si,di}.
1061 (output_probe_stack_range): New.
1062 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1063 (output_probe_stack_range_stack_clash): New.
1064 (rs6000_emit_prologue): Emit notes into dump file as requested.
1065 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1066 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1067 Add additional operand and pass it to output_probe_stack_range.
1069 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1071 PR tree-optimization/82163
1072 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1073 (checking_verify_loop_closed_ssa): New parameter.
1074 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1075 (check_loop_closed_ssa_stmt): Delete.
1076 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1077 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1078 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1081 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1083 * brig-builtins.def: Treat HSAIL barrier builtins as
1084 setjmp/longjump style functions.
1086 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1088 * target.def (constant_alignment): New hook.
1089 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1090 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1091 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1092 * doc/tm.texi: Regenerate.
1093 * targhooks.h (default_constant_alignment): Declare.
1094 (constant_alignment_word_strings): Likewise.
1095 * targhooks.c (default_constant_alignment): New function.
1096 (constant_alignment_word_strings): Likewise.
1097 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1098 instead of CONSTANT_ALIGNMENT.
1099 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1100 (force_const_mem): Likewise.
1101 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1102 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1103 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1104 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1105 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1107 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1108 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1109 constant_alignment_word_strings.
1110 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1111 (CONSTANT_ALIGNMENT): Likewise.
1112 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1113 (arm_constant_alignment): New function.
1114 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1115 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1116 constant_alignment_word_strings.
1117 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1118 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1119 constant_alignment_word_strings.
1120 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1121 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1122 (cris_constant_alignment): New function.
1123 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1124 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1125 (epiphany_constant_alignment): New function.
1126 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1127 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1128 constant_alignment_word_strings.
1129 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1130 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1131 constant_alignment_word_strings.
1132 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1133 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1134 constant_alignment_word_strings.
1135 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1136 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1137 * config/i386/i386.c (ix86_constant_alignment): Make static.
1138 Use the same interface as the target hook.
1139 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1140 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1141 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1142 constant_alignment_word_strings.
1143 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1144 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1145 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1146 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1147 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1148 constant_alignment_word_strings.
1149 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1150 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1151 constant_alignment_word_strings.
1152 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1153 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1154 constant_alignment_word_strings.
1155 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1156 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1158 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1159 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1160 * config/mips/mips.c (mips_constant_alignment): New function.
1161 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1162 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1163 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1164 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1165 (mmix_constant_alignment): Make static. Use the same interface
1167 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1168 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1169 constant_alignment_word_strings.
1170 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1171 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1172 constant_alignment_word_strings.
1173 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1174 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1175 constant_alignment_word_strings.
1176 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1177 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1178 (rs6000_constant_alignment): New function.
1179 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1180 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1181 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1182 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1183 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1184 (rs6000_constant_alignment): New function.
1185 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1186 * config/s390/s390.c (s390_constant_alignment): New function.
1187 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1188 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1189 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1190 constant_alignment_word_strings.
1191 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1192 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1193 (sparc_constant_alignment): New function.
1194 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1195 * config/spu/spu.c (spu_constant_alignment): New function.
1196 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1197 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1198 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1199 constant_alignment_word_strings.
1200 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1201 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1202 constant_alignment_word_strings.
1203 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1204 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1205 constant_alignment_word_strings.
1206 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1207 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1208 (visium_constant_alignment): New function.
1209 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1210 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1211 (xtensa_constant_alignment): New function.
1212 * system.h (CONSTANT_ALIGNMENT): Poison.
1214 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1216 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1217 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1218 (rs6000_builtin_valid_without_lhs): New helper function.
1219 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1220 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1222 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1224 * target.h (vec_perm_indices): Use unsigned short rather than
1226 (auto_vec_perm_indices): Likewise.
1227 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1228 Use unsigned int rather than unsigned char.
1229 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1231 2017-09-25 Richard Biener <rguenther@suse.de>
1233 * cfgloop.h (sort_sibling_loops): Declare.
1234 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1235 (sort_sibling_loops): New function sorting the sibling loop list
1237 * graphite.c (graphite_transform_loops): Sort sibling loops.
1239 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1241 * target.def (vec_perm_const_ok): Change sel parameter to
1243 * optabs-query.c (can_vec_perm_p): Update accordingly.
1244 * doc/tm.texi: Regenerate.
1245 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1246 auto_vec_perm_indices and remove separate nelt field.
1247 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1248 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1249 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1250 (aarch64_expand_vec_perm_const): Update accordingly.
1251 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1252 to vec_perm_indices.
1253 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1254 auto_vec_perm_indices and remove separate nelt field.
1255 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1256 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1257 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1259 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1260 to vec_perm_indices.
1261 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1262 sel to vec_perm_indices.
1263 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1264 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1265 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1267 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1270 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1273 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1274 on the FUNCTION_DECL function context if it has a DIE that is a
1277 2017-09-25 Richard Biener <rguenther@suse.de>
1279 PR tree-optimization/82285
1280 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1283 2017-09-25 Tom de Vries <tom@codesourcery.com>
1287 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1290 2017-09-25 Richard Biener <rguenther@suse.de>
1292 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1293 ISL errors other than isl_error_quota happen. Dump if the
1294 schedule is the same.
1295 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1296 errors instead of aborting inside ISL.
1298 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1301 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1302 of libgcc_eh for m64.
1303 * config/i386/darwin64.h: Likewise.
1305 2017-09-25 Richard Biener <rguenther@suse.de>
1308 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1309 attribute for incomplete types nor twice for complete ones.
1311 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1314 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1315 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1318 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1321 * config/i386/i386.opt (mprefer-avx256): Use
1322 ix86_target_flags variable.
1323 * config/i386/i386.c (ix86_target_string): Move
1324 -mprefer-avx256 to flag2_opts.
1326 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1329 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1330 and x != -1 | y != -1 into (x & y) != -1.
1332 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1334 * config.gcc: Add new case statement to set
1335 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1336 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1337 s390x-*-linux* case statements. Added aarch64 to the list of
1338 supported architectures.
1340 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1342 PR tree-optimization/82289
1343 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1344 STMT_VINFO_RELEVANT_P.
1346 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1347 Alan Hayward <alan.hayward@arm.com>
1348 David Sherwood <david.sherwood@arm.com>
1350 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1351 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1352 (extract_range_from_binary_expr_1): Don't call
1353 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1355 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1356 Alan Hayward <alan.hayward@arm.com>
1357 David Sherwood <david.sherwood@arm.com>
1359 * target.def (preferred_vector_alignment): New hook.
1360 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1362 * doc/tm.texi: Regenerate.
1363 * targhooks.h (default_preferred_vector_alignment): Declare.
1364 * targhooks.c (default_preferred_vector_alignment): New function.
1365 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1367 (DR_TARGET_ALIGNMENT): New macro.
1368 (aligned_access_p): Update commentary.
1369 (vect_known_alignment_in_bytes): New function.
1370 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1372 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1373 Calculate the misalignment based on the target alignment rather than
1375 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1376 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1377 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1378 the target alignment, rather than masking the element misalignment
1379 with the number of elements in a vector. Also use the target
1380 alignment when calculating the maximum number of peels.
1381 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1382 instead of TYPE_ALIGN_UNIT.
1383 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1384 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1385 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1386 (vect_create_data_ref_ptr): Likewise.
1387 (vect_setup_realignment): Realign by ANDing with
1388 -DR_TARGET_MISALIGNMENT.
1389 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1390 the number of peels based on DR_TARGET_ALIGNMENT.
1391 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1392 with the guaranteed alignment boundary when deciding whether
1394 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1395 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1396 (ensure_base_align): Remove stmt_info parameter. Get the
1397 target base alignment from DR_TARGET_ALIGNMENT.
1398 (vectorizable_store): Update call accordingly. Interpret
1399 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1401 (vectorizable_load): Likewise.
1403 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1404 Alan Hayward <alan.hayward@arm.com>
1405 David Sherwood <david.sherwood@arm.com>
1407 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1408 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1409 (vect_enhance_data_refs_alignment): Likewise.
1411 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1413 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1414 error. Only quit immediately if parsing is complete.
1415 (BEGIN): Initialize fatal_err and parse_done.
1416 (begin fpu, end fpu): Check number of arguments.
1417 (begin arch, end arch): Likewise.
1418 (begin cpu, end cpu): Likewise.
1419 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1420 (optalias): Likewise.
1422 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1424 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1425 * config/arm/arm-isa.h: Delete. Move definitions to ...
1426 * arm-cpus.in: ... here. Use new feature and fgroup values.
1427 * config/arm/arm.c (arm_option_override): Use lower case for feature
1429 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1430 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1431 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1433 (print_isa_bits_for): New function.
1434 (gen_isa): New function.
1435 (gen_comm_data): Use print_isa_bits_for.
1436 (define feature): New keyword.
1437 (define fgroup): New keyword.
1438 * config/arm/t-arm (TM_H): Remove.
1439 (GTM_H): Add arm-isa.h.
1440 (arm-isa.h): Add rule to generate file.
1441 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1442 case for feature bit names.
1444 2017-09-22 Richard Biener <rguenther@suse.de>
1446 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1448 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1449 print of no dependency loops ...
1450 * graphite.c (graphite_transform_loops): ... here.
1451 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1453 (same_close_phi_node, remove_duplicate_close_phi,
1454 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1455 (canonicalize_loop_closed_ssa): ... here and simplify.
1456 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1457 (optimize_isl): Use dump_printf_loc to tell when we stopped
1458 optimizing because of an ISL timeout.
1460 2017-09-22 Richard Biener <rguenther@suse.de>
1462 PR tree-optimization/82291
1463 * tree-if-conv.c (predicate_mem_writes): Make sure to
1464 remove writes in blocks predicated with false.
1466 2017-09-22 Richard Biener <rguenther@suse.de>
1468 * sese.c: Include cfganal.h.
1469 (if_region_set_false_region): Remove.
1470 (create_if_region_on_edge): Likewise.
1471 (move_sese_in_condition): Re-implement without destroying
1474 2017-09-22 Richard Biener <rguenther@suse.de>
1476 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1477 Verify both BBs contain loop PHI nodes before dispatching to
1479 (graphite_regenerate_ast_isl): Do not recompute dominators,
1480 do not verify three times. Restructure for clarity.
1481 * graphite-scop-detection.c (same_close_phi_node,
1482 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1483 defined_in_loop_p, canonicalize_loop_closed_ssa,
1484 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1485 checking and SSA rewrite, move to ...
1486 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1487 (graphite_initialize): Do not pass in ctx, do not reset the
1488 SCEV cache, compute only dominators.
1489 (graphite_transform_loops): Allocate ISL ctx after
1490 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1491 Maintain post-dominators only around build_scops.
1492 * sese.c (if_region_set_false_region): Make static. Free
1493 and recompute dominators.
1494 (move_sese_in_condition): Assert we don't get called with
1495 post-dominators computed.
1496 * sese.h (if_region_set_false_region): Remove.
1498 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1500 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1501 mode attribute for TARGET_AVX512VL.
1503 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1505 * config/i386/i386.opt (mprefer-avx256): New option.
1506 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1508 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1509 for TARGET_PREFER_AVX256.
1510 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1512 2017-09-21 Jeff Law <law@redhat.com>
1514 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1515 Fix dump output if the only stack space is for pushed registers.
1517 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1519 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1522 2017-09-21 Martin Sebor <msebor@redhat.com>
1525 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1526 code (in C++) or code that triggers warnings.
1528 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1530 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1532 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1534 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1535 * sched-rgn.c: Ditto.
1536 * sel-sched-ir.c: Ditto.
1538 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1540 * toplev.h (set_random_seed): Adjust return type.
1541 * toplev.c (init_local_tick): Move eager initialization of random_seed
1542 to get_random_seed. Adjust comment.
1543 (init_random_seed): Inline to get_random_seed, delete.
1544 (get_random_seed): Initialize random_seed lazily.
1545 (set_random_seed): Do not return previous value.
1546 (print_switch_value): Do not call get_random_seed.
1548 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1550 * cgraph.c (delete_function_version): New, broken out from...
1551 (cgraph_node::delete_function_version): ...here. Rename to
1552 cgraph_node::delete_function_version_by_decl. Update all uses.
1553 (cgraph_node::remove): Call delete_function_version.
1555 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1558 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1559 VAR_DECLs to which addressable non-volatile parameters are mapped
1560 and for id->retvar after the return value assignment. Clear
1561 id->retval and id->retbnd after inlining.
1563 2017-09-21 Richard Biener <rguenther@suse.de>
1565 PR tree-optimization/82276
1566 PR tree-optimization/82244
1567 * tree-vrp.c (build_assert_expr_for): Set
1568 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1570 (remove_range_assertions): Revert earlier change.
1572 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1575 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1577 2017-09-21 Richard Biener <rguenther@suse.de>
1579 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1580 Restore valid IL after code generation errors.
1581 * graphite.c (graphite_transform_loops): Diagnose code
1582 generation issues as MSG_MISSED_OPTIMIZATION and continue
1583 with processing SCOPs.
1585 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1586 Alan Hayward <alan.hayward@arm.com>
1587 David Sherwood <david.sherwood@arm.com>
1589 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1590 rather than choosing between plus_constant and gen_rtx_<CODE>.
1591 * expr.c (emit_push_insn): Likewise.
1592 (expand_expr_real_2): Likewise.
1594 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1595 Alan Hayward <alan.hayward@arm.com>
1596 David Sherwood <david.sherwood@arm.com>
1598 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1600 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1601 Alan Hayward <alan.hayward@arm.com>
1602 David Sherwood <david.sherwood@arm.com>
1604 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1605 calling tree_to_uhwi.
1607 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1608 Alan Hayward <alan.hayward@arm.com>
1609 David Sherwood <david.sherwood@arm.com>
1611 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1612 INTEGER_CST rather than a negative test for ADDR_EXPR.
1614 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1615 Alan Hayward <alan.hayward@arm.com>
1616 David Sherwood <david.sherwood@arm.com>
1618 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1619 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1621 2017-09-21 Richard Biener <rguenther@suse.de>
1623 PR tree-optimization/71351
1624 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1625 graphite_create_new_loop_guard): Remove, fold remaining parts
1627 (translate_isl_ast_node_for): ... here and simplify.
1629 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1632 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1633 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1634 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1635 alternative always use QI mode, for -Os imov (=R,R) alternative
1636 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1639 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1640 Jeff Law <law@redhat.com>
1642 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1643 (allocate_stack_space): New function, partially extracted from
1645 (s390_emit_prologue): Track offset to most recent stack probe.
1646 Code to allocate space moved into allocate_stack_space.
1647 Dump actions when no stack is allocated.
1648 (s390_prologue_plus_offset): New function.
1649 (s390_emit_stack_probe): Likewise.
1651 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1653 * common.opt (Wa, Wl, Wp, g, gz=): Add
1655 (gno-column-info): Remove.
1656 (gcolumn-info): Drop RejectNegative.
1658 (gno-record-gcc-switches): Remove.
1659 (grecord-gcc-switches): Drop RejectNegative.
1660 (gno-split-dwarf): Remove.
1661 (gsplit-dwarf): Drop RejectNegative.
1662 (gno-strict-dwarf): Remove.
1663 (gstrict-dwarf): Drop RejectNegative.
1664 * config/darwin.opt (gfull, gused): Add RejectNegative.
1665 * dwarf2out.c (gen_producer_string): Drop
1666 gno-record-gcc-switches handler.
1667 * optc-gen.awk: Add g to prefixes with negative forms.
1668 * opts-common.c (remapping_prefix_p): New.
1669 (find_opt): Check it.
1670 (generate_canonical_option): Test g prefix.
1671 (option_map): Add -gno- mapping.
1672 (add_misspelling_candidates): Check remapping_prefix_p.
1674 2017-09-20 Jeff Law <law@redhat.com>
1676 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1677 thinko in stack clash protection support.
1679 * explow.c (compute_stack_clash_protection_loop_data): Use
1680 CONST_INT_P instead of explicit test. Verify object is a
1681 CONST_INT_P before looking at INTVAL.
1682 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1683 instead of explicit test.
1685 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1688 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1689 address instead of to r1 and r11.
1691 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1693 * config.gcc: Support "knm".
1694 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1695 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1697 * config/i386/i386.c (m_KNM): Define.
1698 (processor_target_table): Add "knm".
1700 (ix86_option_override_internal): Add "knm".
1701 (ix86_issue_rate): Add PROCESSOR_KNM.
1702 (ix86_adjust_cost): Ditto.
1703 (ia32_multipass_dfa_lookahead): Ditto.
1704 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1705 (fold_builtin_cpu): Add M_INTEL_KNM.
1706 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1707 (processor_type): Add PROCESSOR_KNM.
1708 * config/i386/x86-tune.def: Add m_KNM.
1709 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1711 2017-09-20 Richard Biener <rguenther@suse.de>
1713 PR tree-optimization/80213
1714 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1715 are allowed in empty BBs as well.
1716 (canonicalize_loop_closed_ssa): Also look for other complex
1718 (scop_detection::get_sese): Include the loop-closed PHI block
1720 (scop_detection::merge_sese): Remove code adding extra blocks.
1721 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1722 (build_scops): Assert the final returned scop is invalid.
1724 2017-09-20 Richard Biener <rguenther@suse.de>
1726 PR tree-optimization/82264
1727 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1729 (vn_phi_lookup): Likewise.
1730 (vn_phi_insert): Likewise.
1732 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1734 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1735 that fits into uhwi or shwi, add DW_AT_const_value regardless
1736 of early_dwarf without going through RTL, using add_AT_unsigned
1739 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1740 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1741 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1742 (DEBUG_MACRO_SECTION): Likewise.
1743 (DEBUG_LTO_MACRO_SECTION): Likewise.
1744 (DEBUG_STR_DWO_SECTION): Likewise.
1745 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1746 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1747 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1748 (DEBUG_LTO_LINE_STR_SECTION): Define.
1749 (init_sections_and_labels): Initialize debug_line_str_section
1750 variable. Initialize debug_loc_section for -gdwarf-5 to
1751 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1753 2017-09-20 Richard Biener <rguenther@suse.de>
1755 * graphite-sese-to-poly.c (extract_affine): Properly handle
1756 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1758 2017-09-20 Richard Biener <rguenther@suse.de>
1760 PR tree-optimization/81373
1761 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1762 Force SESE live-out defs to be handled even if they are
1765 2017-09-19 Jeff Law <law@redhat.com>
1767 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1768 nothing for stack adjustments with REG_STACK_CHECK.
1769 * sched-deps.c (parse_add_or_inc): Reject insns with
1770 REG_STACK_CHECK from dependency breaking.
1771 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1772 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1773 * reg-notes.def (STACK_CHECK): New note.
1775 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1776 (ix86_expand_prologue): Dump stack clash info as needed.
1777 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1779 * function.c (dump_stack_clash_frame_info): New function.
1780 * function.h (dump_stack_clash_frame_info): Prototype.
1781 (enum stack_clash_probes): New enum.
1783 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1784 flag_stack_clash_protection.
1785 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1786 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1787 (arm_frame_pointer_required): Likewise.
1788 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1789 (ia64_expand_prologue): Likewise.
1790 * config/mips/mips.c (mips_expand_prologue): Likewise.
1791 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1792 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1793 (sparc_flat_expand_prologue): Likewise.
1794 * config/spu/spu.c (spu_expand_prologue): Likewise.
1796 * explow.c: Include "params.h".
1797 (anti_adjust_stack_and_probe_stack_clash): New function.
1798 (get_stack_check_protect): Likewise.
1799 (compute_stack_clash_protection_loop_data): Likewise.
1800 (emit_stack_clash_protection_loop_start): Likewise.
1801 (emit_stack_clash_protection_loop_end): Likewise.
1802 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1803 Use anti_adjust_stack_and_probe_stack_clash.
1804 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1805 (emit_stack_clash_protection_loop_start): Likewise.
1806 (emit_stack_clash_protection_loop_end): Likewise.
1807 * rtl.h (get_stack_check_protect): Prototype.
1808 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1809 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1810 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1812 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1814 * doc/tm.texi: Rebuilt.
1815 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1816 get_stack_check_protect.
1817 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1818 * config/arm/arm.c (arm_expand_prologue): Likewise.
1819 (arm_frame_pointer_required): Likewise.
1820 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1821 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1822 * config/mips/mips.c (mips_expand_prologue): Likewise.
1823 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1824 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1825 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1826 (sparc_flat_expand_prologue): Likewise.
1828 * common.opt (-fstack-clash-protection): New option.
1829 * flag-types.h (enum stack_check_type): Note difference between
1830 -fstack-check= and -fstack-clash-protection.
1831 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1832 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1833 * toplev.c (process_options): Issue warnings/errors for cases
1834 not handled with -fstack-clash-protection.
1835 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1836 (-fstack-check): Note additional problem with -fstack-check=generic.
1837 Note that -fstack-check is primarily for Ada and refer users
1838 to -fstack-clash-protection for stack-clash-protection.
1839 Document new params for stack clash protection.
1841 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1843 * config/i386/i386.c (ix86_split_long_move): Do not handle
1844 address used for LEA in a special way.
1846 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1848 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1850 2017-09-19 Martin Sebor <msebor@redhat.com>
1853 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1854 of incompatible types.
1856 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1858 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1859 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1860 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1861 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1863 2017-09-19 Richard Biener <rguenther@suse.de>
1865 PR tree-optimization/82244
1866 * tree-vrp.c (remove_range_assertions): Do not propagate
1867 a constant to abnormals but replace the assert with a copy.
1869 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1871 PR rtl-optimization/57878
1872 PR rtl-optimization/68988
1873 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1874 avoidance test involving non_reload_pseudos. Move frequency test
1875 below the general fragmentation avoidance test.
1877 2017-09-19 Richard Biener <rguenther@suse.de>
1879 PR tree-optimization/69728
1880 * graphite-sese-to-poly.c (schedule_error): New global.
1881 (add_loop_schedule): Handle empty domain by failing the
1883 (build_original_schedule): Handle schedule_error.
1885 2017-09-19 Richard Biener <rguenther@suse.de>
1887 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1888 Do not iterate to sibling loops but only to siblings of inner
1891 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1894 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1895 that the registers are different.
1897 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1899 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1900 to processor_model and "amdfam17h" to arch_names_table.
1901 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1903 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1906 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1908 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1909 Alan Hayward <alan.hayward@arm.com>
1910 David Sherwood <david.sherwood@arm.com>
1912 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1914 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1916 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1917 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1918 vect_schedule_slp_instance.
1919 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1920 Update call to vect_slp_analyze_node_operations. Simplify return
1922 (vect_slp_analyze_bb_1): Update call accordingly.
1923 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1924 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1925 (vect_schedule_slp): Update call accordingly.
1927 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1928 Alan Hayward <alan.hayward@arm.com>
1929 David Sherwood <david.sherwood@arm.com>
1931 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1932 with types that aren't in fact scalar.
1934 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1936 * tree-vect-slp.c (vect_record_max_nunits): New function,
1938 (vect_build_slp_tree_1): ...here.
1939 (vect_build_slp_tree_2): Call it for phis too.
1941 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1943 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1944 to vect_get_vec_def_for_operand when getting the mask operand.
1946 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1947 Alan Hayward <alan.hayward@arm.com>
1948 David Sherwood <david.sherwood@arm.com>
1950 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1953 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1954 Alan Hayward <alan.hayward@arm.com>
1955 David Sherwood <david.sherwood@arm.com>
1957 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1958 calculation for vector booleans.
1960 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1961 Alan Hayward <alan.hayward@arm.com>
1962 David Sherwood <david.sherwood@arm.com>
1964 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1966 (vect_transform_stmt): ...here.
1967 (vect_analyze_stmt): Use it instead of calling
1968 vectorizable_live_operation directly.
1970 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1972 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1973 non-SIMT targets in acc vector loops.
1975 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1977 * configure.ac: Add arc and check if assembler supports gdwarf2.
1978 * configure: Regenerate.
1980 2017-09-18 Richard Biener <rguenther@suse.de>
1982 PR tree-optimization/82220
1983 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1984 epilogue niters from the min_profitable_iters compute.
1986 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1989 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1990 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1991 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1993 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1996 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1997 switching to a new text section.
1999 2017-09-18 Richard Biener <rguenther@suse.de>
2001 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2003 (build_alias_set): Reject aliases with no access function.
2005 2017-09-18 Richard Biener <rguenther@suse.de>
2007 PR tree-optimization/79622
2008 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2010 (build_cross_bb_scalars_use): Likewise.
2012 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2014 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2016 2017-09-18 Alan Modra <amodra@gmail.com>
2019 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2020 stack_pointer_rtx for count 0. Update comments. Break up
2021 large rtl expression.
2023 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2025 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2026 Increase to 20 bytes.
2027 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2028 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2029 or avx version of the stub.
2031 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2034 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2035 compute the minimum stack alignment. Also update preferred stack
2036 boundary for leaf functions.
2038 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2040 PR tree-optimization/82228
2041 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2044 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2046 * common/config/nds32/nds32-common.c
2047 (nds32_option_optimization_table): Refine formatting.
2048 (nds32_option_optimization_table): Use -fsched-pressure and
2049 -fomit-frame-pointer for specific optimization level.
2051 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2053 * config/nds32/nds32.c: Refine formatting and comments.
2054 * config/nds32/nds32.h: Likewise.
2055 * config/nds32/nds32.md: Likewise.
2056 * config/nds32/nds32-cost.c: Likewise.
2057 * config/nds32/nds32-isr.c: Likewise.
2058 * config/nds32/nds32-md-auxiliary.c: Likewise.
2059 * config/nds32/nds32-multiple.md: Likewise.
2060 * config/nds32/nds32-predicates.c: Likewise.
2062 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2063 Jakub Jelinek <jakub@redhat.com>
2065 Add support for -std=c++2a.
2066 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2068 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2070 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2073 * doc/extend.texi (Common Function Attributes): Add
2074 references to ARM, AArch64, and S/390 specific attributes.
2075 (Function Specific Option Pragmas): Add AArch64 and S/390
2076 to list of back ends that support the target pragma.
2078 2017-09-15 Nathan Sidwell <nathan@acm.org>
2080 * doc/standards.texi: Fix C++17 description. Update URLs for
2083 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2085 * common.opt (Wcast-align=strict): New warning option.
2086 * doc/invoke.texi: Document -Wcast-align=strict.
2088 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2090 * cgraph.h (cgraph_thunk_info): Add comments.
2091 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2092 assert for VIRTUAL_* arguments stricter.
2094 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2096 PR tree-optimization/71026
2097 * match.pd: Move RDIV patterns from fold-const.c
2098 * fold-const.c (distribute_real_division): Removed.
2099 (fold_binary_loc): Remove calls to distribute_real_divison.
2101 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2103 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2104 c++1z and gnu++1z as deprecated. Change other references to
2105 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2106 Change -Wc++1z-compat to -Wc++17-compat.
2107 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2108 * dwarf2out.c (highest_c_language): Handle C++17.
2109 (gen_compile_unit_die): Likewise.
2111 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2113 PR rtl-optimization/82192
2114 * combine.c (make_extraction): Don't look through non-paradoxical
2115 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2118 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2119 Alan Hayward <alan.hayward@arm.com>
2120 David Sherwood <david.sherwood@arm.com>
2122 * target.def (function_arg_offset): New hook.
2123 * targhooks.h (default_function_arg_offset): Declare.
2124 * targhooks.c (default_function_arg_offset): New function.
2125 * function.c (locate_and_pad_parm): Use
2126 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2127 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2128 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2129 * doc/tm.texi: Regenerate.
2130 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2131 * config/spu/spu.c (spu_function_arg_offset): New function.
2132 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2133 * system.h (FUNCTION_ARG_OFFSET): Poison.
2135 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2136 Alan Hayard <alan.hayward@arm.com>
2137 David Sherwood <david.sherwood@arm.com>
2139 * target.def (truly_noop_truncation): New hook.
2140 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2141 than TRULY_NOOP_TRUNCATION.
2142 * hooks.h (hook_bool_uint_uint_true): Declare.
2143 * hooks.c (hook_bool_uint_uint_true): New function.
2144 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2145 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2146 * doc/tm.texi: Regenerate.
2147 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2148 rather than TRULY_NOOP_TRUNCATION in comments.
2149 (simplify_comparison): Likewise.
2150 (record_truncated_value): Likewise.
2151 * expmed.c (extract_bit_field_1): Likewise.
2152 (extract_split_bit_field): Likewise.
2153 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2154 instead of TRULY_NOOP_TRUNCATION.
2155 * function.c (assign_parm_setup_block): Likewise.
2156 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2157 * rtlhooks.c: Include target.h.
2158 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2159 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2160 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2161 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2162 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2163 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2164 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2165 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2166 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2167 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2168 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2169 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2170 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2171 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2172 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2173 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2174 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2175 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2176 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2177 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2178 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2179 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2180 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2181 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2182 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2183 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2184 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2185 rather than TRULY_NOOP_TRUNCATION in comments.
2186 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2187 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2188 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2189 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2190 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2191 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2192 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2193 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2194 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2195 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2196 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2197 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2198 rather than TRULY_NOOP_TRUNCATION in comments.
2199 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2200 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2201 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2202 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2203 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2204 TRULY_NOOP_TRUNCATION condition.
2205 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2206 (TRULY_NOOP_TRUNCATION): Delete.
2207 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2208 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2209 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2210 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2211 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2212 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2213 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2214 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2215 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2216 rather than TRULY_NOOP_TRUNCATION in comments.
2217 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2218 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2219 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2220 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2221 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2222 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2224 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2227 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2228 (*cmp_ior): Likewise.
2229 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2230 (*ior_scc_scc_cmp): Likewise.
2231 (*and_scc_scc): Likewise.
2232 (*and_scc_scc_cmp): Likewise.
2234 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2235 Alan Hayard <alan.hayward@arm.com>
2236 David Sherwood <david.sherwood@arm.com>
2238 * target.def (can_change_mode_class): New hook.
2239 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2240 (hard_regno_nregs): Likewise.
2241 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2242 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2243 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2244 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2245 (LOAD_EXTEND_OP): Update accordingly.
2246 * doc/tm.texi: Regenerate.
2247 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2248 CANNOT_CHANGE_MODE_CLASS.
2249 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2250 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2251 * combine.c (simplify_set): Update accordingly.
2252 * emit-rtl.c (validate_subreg): Likewise.
2253 * recog.c (general_operand): Likewise.
2254 * regcprop.c (mode_change_ok): Likewise.
2255 * reload1.c (choose_reload_regs): Likewise.
2256 (inherit_piecemeal_p): Likewise.
2257 * rtlanal.c (simplify_subreg_regno): Likewise.
2258 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2259 instead of CANNOT_CHANGE_MODE_CLASS.
2260 (reload_cse_simplify_operands): Likewise.
2261 * reload.c (push_reload): Use targetm.can_change_mode_class
2262 instead of CANNOT_CHANGE_MODE_CLASS.
2263 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2264 REG_CANNOT_CHANGE_MODE_P.
2265 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2266 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2267 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2268 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2269 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2270 (arm_can_change_mode_class): New function.
2271 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2272 than CANNOT_CHANGE_MODE_CLASS in comments.
2273 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2274 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2275 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2276 (ix86_can_change_mode_class): ...this new function, inverting the
2277 sense of the return value.
2278 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2279 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2280 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2281 (ia64_can_change_mode_class): New function.
2282 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2283 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2284 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2285 (m32c_can_change_mode_class): ...this new function, inverting the
2286 sense of the return value.
2287 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2288 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2289 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2290 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2291 (mips_can_change_mode_class): ...this new function, inverting the
2292 sense of the return value.
2293 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2294 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2295 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2296 (msp430_can_change_mode_class): New function.
2297 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2298 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2299 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2300 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2301 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2302 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2303 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2304 (pa_cannot_change_mode_class): Replace with...
2305 (pa_can_change_mode_class): ...this new function, inverting the
2306 sense of the return value.
2307 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2308 than CANNOT_CHANGE_MODE_CLASS in comments.
2309 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2310 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2311 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2312 (pdp11_cannot_change_mode_class): Replace with...
2313 (pdp11_can_change_mode_class): ...this new function, inverting the
2314 sense of the return value.
2315 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2316 * config/powerpcspe/powerpcspe-protos.h
2317 (rs6000_cannot_change_mode_class_ptr): Delete.
2318 * config/powerpcspe/powerpcspe.c
2319 (rs6000_cannot_change_mode_class_ptr): Delete.
2320 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2321 (rs6000_option_override_internal): Assign to
2322 targetm.can_change_mode_class instead of
2323 rs6000_cannot_change_mode_class_ptr.
2324 (rs6000_cannot_change_mode_class): Replace with...
2325 (rs6000_can_change_mode_class): ...this new function, inverting the
2326 sense of the return value.
2327 (rs6000_debug_cannot_change_mode_class): Replace with...
2328 (rs6000_debug_can_change_mode_class): ...this new function.
2329 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2330 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2331 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2332 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2333 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2335 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2336 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2337 (rs6000_option_override_internal): Assign to
2338 targetm.can_change_mode_class instead of
2339 rs6000_cannot_change_mode_class_ptr.
2340 (rs6000_cannot_change_mode_class): Replace with...
2341 (rs6000_can_change_mode_class): ...this new function, inverting the
2342 sense of the return value.
2343 (rs6000_debug_cannot_change_mode_class): Replace with...
2344 (rs6000_debug_can_change_mode_class): ...this new function.
2345 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2346 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2347 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2348 (s390_can_change_mode_class): ...this new function, inverting the
2349 sense of the return value.
2350 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2351 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2352 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2353 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2354 (sh_cannot_change_mode_class): Replace with...
2355 (sh_can_change_mode_class): ...this new function, inverting the
2356 sense of the return value.
2357 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2358 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2359 (sparc_can_change_mode_class): New function.
2360 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2361 * config/spu/spu.c (spu_can_change_mode_class): New function.
2362 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2363 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2364 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2365 (visium_can_change_mode_class): New function.
2366 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2368 2017-09-15 Richard Biener <rguenther@suse.de>
2370 PR tree-optimization/82217
2371 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2372 but not undefined case.
2374 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2377 * postreload.c (reload_cse_simplify_operands): Skip
2378 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2380 2017-09-15 Richard Biener <rguenther@suse.de>
2382 PR tree-optimization/68823
2383 * graphite-scop-detection.c (build_alias_set): If we have a
2384 possible dependence check whether we can handle them by just
2385 looking at the DRs DR_ACCESS_FNs.
2386 (build_scops): If build_alias_set fails, fail the SCOP.
2388 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2390 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2391 to support float128 built-in functions that require the ISA 3.0
2393 (BU_FLOAT128_3_HW): Likewise.
2394 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2396 (FMAF128): Likewise.
2398 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2399 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2400 floating point instructions.
2401 (rs6000_invalid_builtin): Likewise.
2402 (rs6000_builtin_mask_names): Likewise.
2403 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2404 (RS6000_BTM_FLOAT128_HW): Likewise.
2405 (RS6000_BTM_COMMON): Likewise.
2406 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2408 * doc/extend.texi (RS/6000 built-in functions): Document the
2409 IEEE 128-bit floating point square root and fused multiply-add
2412 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2414 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2415 reg (r2) isn't in the set of registers defined in the prologue.
2417 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2418 Alan Hayward <alan.hayward@arm.com>
2419 David Sherwood <david.sherwood@arm.com>
2421 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2422 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2423 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2424 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2425 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2427 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2428 max_vectorization_factor.
2429 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2431 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2432 Alan Hayward <alan.hayward@arm.com>
2433 David Sherwood <david.sherwood@arm.com>
2435 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2436 (vect_worthwhile_without_simd_p): Declare.
2437 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2438 (vectorizable_reduction): Use it.
2439 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2440 (vectorizable_operation): Likewise.
2442 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2443 Alan Hayward <alan.hayward@arm.com>
2444 David Sherwood <david.sherwood@arm.com>
2446 * tree-vectorizer.h (vect_get_num_copies): New function.
2447 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2448 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2449 (vectorizable_induction): Likewise.
2450 (vectorizable_live_operation): Likewise.
2451 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2452 (vectorizable_bswap): Likewise.
2453 (vectorizable_call): Likewise.
2454 (vectorizable_conversion): Likewise.
2455 (vectorizable_assignment): Likewise.
2456 (vectorizable_shift): Likewise.
2457 (vectorizable_operation): Likewise.
2458 (vectorizable_store): Likewise.
2459 (vectorizable_load): Likewise.
2460 (vectorizable_condition): Likewise.
2461 (vectorizable_comparison): Likewise.
2462 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2464 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2465 Alan Hayward <alan.hayward@arm.com>
2466 David Sherwood <david.sherwood@arm.com>
2468 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2469 of vect_init_vector.
2471 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2472 Alan Hayward <alan.hayward@arm.com>
2473 David Sherwood <david.sherwood@arm.com>
2475 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2476 an inline wrapper that provides a location.
2477 (gimple_build_vector): Likewise.
2478 * gimple-fold.c (gimple_build_vector_from_val): New function.
2479 (gimple_build_vector): Likewise.
2480 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2481 functions to build the initial value. Always return a gimple value.
2482 (get_initial_defs_for_reduction): Likewise. Only compute
2484 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2485 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2486 (vectorizable_induction): Use gimple_build_vector rather than
2489 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2490 Alan Hayward <alan.hayward@arm.com>
2491 David Sherwood <david.sherwood@arm.com>
2493 * target.h (vec_perm_indices): New typedef.
2494 (auto_vec_perm_indices): Likewise.
2495 * optabs-query.h: Include target.h
2496 (can_vec_perm_p): Take a vec_perm_indices *.
2497 * optabs-query.c (can_vec_perm_p): Likewise.
2498 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2499 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2500 * tree-vect-generic.c (lower_vec_perm): Likewise.
2501 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2502 (vect_grouped_load_supported): Likewise.
2503 (vect_shift_permute_load_chain): Likewise.
2504 (vect_permute_store_chain): Use auto_vec_perm_indices.
2505 (vect_permute_load_chain): Likewise.
2506 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2507 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2508 Update uses of can_vec_perm_p.
2509 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2510 mode with a number of elements. Take a vec_perm_indices *.
2511 (vect_create_epilog_for_reduction): Update accordingly.
2512 Use auto_vec_perm_indices.
2513 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2514 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2515 (vect_transform_slp_perm_load): Likewise.
2516 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2517 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2518 (vect_gen_perm_mask_checked): Likewise.
2519 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2520 (vect_gen_perm_mask_checked): Likewise.
2521 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2522 (vectorizable_store): Likewise.
2523 (vectorizable_load): Likewise.
2524 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2525 (vectorizable_bswap): Likewise.
2527 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2528 Alan Hayward <alan.hayward@arm.com>
2529 David Sherwood <david.sherwood@arm.com>
2531 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2532 * tree.c (build_vector): Likewise.
2533 (build_vector_from_ctor): Update accordingly.
2534 (build_vector_from_val): Likewise.
2535 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2536 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2537 * tree-vect-generic.c (add_rshift): Likewise.
2538 (expand_vector_divmod): Likewise.
2539 (optimize_vector_constructor): Likewise.
2540 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2541 (vect_transform_slp_perm_load): Likewise.
2542 (vect_schedule_slp_instance): Likewise.
2543 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2544 (vectorizable_call): Likewise.
2545 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2546 * expmed.c (make_tree): Likewise.
2547 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2548 a vector passed to build_vector.
2549 (fold_convert_const): Likewise.
2550 (exact_inverse): Likewise.
2551 (fold_ternary_loc): Likewise.
2552 (fold_relational_const): Likewise.
2553 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2554 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2555 (const_unop): Likewise. Store the reduction accumulator in a
2556 variable rather than an array.
2557 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2558 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2559 the new vector, rather than constructing it after the input arrays.
2560 (native_interpret_vector): Use auto_vec<tree> when building
2561 a vector passed to build_vector. Add elements in order.
2562 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2563 auto_vec<tree> when building a vector passed to build_vector.
2564 (vect_create_epilog_for_reduction): Likewise.
2565 (vectorizable_induction): Likewise.
2566 (get_initial_def_for_reduction): Likewise. Fix indentation of
2568 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2570 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2571 passed to build_vector.
2573 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2574 Alan Hayward <alan.hayward@arm.com>
2575 David Sherwood <david.sherwood@arm.com>
2577 * tree-core.h (tree_base::u): Add an "nelts" field.
2578 (tree_vector): Use VECTOR_CST_NELTS as the length.
2579 * tree.c (tree_size): Likewise.
2580 (make_vector): Initialize VECTOR_CST_NELTS.
2581 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2582 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2583 TYPE_VECTOR_SUBPARTS.
2584 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2585 as the number of units, setting it from VECTOR_CST_NELTS.
2586 (const_vector_from_tree): Likewise.
2587 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2588 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2589 (fold_negate_expr_1): Likewise.
2590 (fold_convert_const): Likewise.
2591 (const_binop): Likewise. Differentiate the number of output and
2593 (const_unop): Likewise.
2594 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2595 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2596 in cases that did the opposite.
2598 2017-09-14 Richard Biener <rguenther@suse.de>
2600 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2603 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2605 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2607 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2610 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2611 if and where to split a bb, except for splitting before debug insn
2612 sequences followed by non-label real insn. Delete debug insns
2613 in between basic blocks.
2615 * combine.c (make_compound_operation_int): Formatting fixes.
2617 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2618 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2619 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2620 * config/sol2.h (LINK_EH_SPEC): Likewise.
2621 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2622 * config/s390/linux.h (LINK_SPEC): Likewise.
2623 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2624 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2625 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2626 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2627 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2628 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2629 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2630 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2632 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2634 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2636 (ENDFILE_LINUX_SPEC): Likewise.
2637 (LINK_EH_SPEC): Likewise.
2638 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2639 (LINK_OS_LINUX_SPEC32): Likewise.
2640 (LINK_OS_LINUX_SPEC64): Likewise.
2641 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2642 (LINK_OS_LINUX_SPEC): Likewise.
2644 2017-09-13 Martin Liska <mliska@suse.cz>
2647 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2648 CASE_HIGH is NULL_TREE.
2650 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2651 Alan Hayward <alan.hayward@arm.com>
2652 David Sherwood <david.sherwood@arm.com>
2654 * target.def (secondary_memory_needed): New hook.
2655 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2656 instead of SECONDARY_MEMORY_NEEDED.
2657 (secondary_memory_needed_mode): Likewise.
2658 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2659 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2660 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2661 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2662 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2663 * doc/tm.texi: Regenerate.
2664 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2665 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2666 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2667 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2668 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2669 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2670 mode argument first and change the reg_class arguments to reg_class_t.
2671 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2672 Make static. Update the call to inline_secondary_memory_needed.
2673 (ix86_register_move_cost): Update the call to
2674 inline_secondary_memory_needed.
2675 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2676 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2678 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2679 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2681 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2682 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2683 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2684 and match hook interface. Add comment from mips.h.
2685 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2686 * config/mmix/mmix.md (truncdfsf2): Refer to
2687 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2689 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2690 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2691 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2692 (pa_secondary_memory_needed): New function.
2693 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2694 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2695 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2696 (pdp11_secondary_memory_needed): Make static and match hook interface.
2697 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2698 * config/powerpcspe/powerpcspe-protos.h
2699 (rs6000_secondary_memory_needed_ptr): Delete.
2700 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2702 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2703 (rs6000_option_override_internal): Assign to
2704 targetm.secondary_memory_needed rather than
2705 rs6000_secondary_memory_needed_ptr.
2706 (rs6000_secondary_memory_needed): Match hook interface.
2707 (rs6000_debug_secondary_memory_needed): Likewise.
2708 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2709 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2710 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2711 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2712 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2713 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2715 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2716 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2717 (rs6000_option_override_internal): Assign to
2718 targetm.secondary_memory_needed rather than
2719 rs6000_secondary_memory_needed_ptr.
2720 (rs6000_secondary_memory_needed): Match hook interface.
2721 (rs6000_debug_secondary_memory_needed): Likewise.
2722 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2723 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2724 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2725 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2726 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2727 (sparc_secondary_memory_needed): New function.
2728 * lra-constraints.c (check_and_process_move): Refer to
2729 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2731 (curr_insn_transform): Likewise.
2732 (process_alt_operands): Use targetm.secondary_memory_needed
2733 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2734 (check_secondary_memory_needed_p): Likewise.
2735 (choose_split_class): Likewise.
2736 * reload.c: Unconditionally include code that was previously
2737 conditional on SECONDARY_MEMORY_NEEDED.
2738 (push_secondary_reload): Use targetm.secondary_memory_needed
2739 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2740 (push_reload): Likewise.
2741 * reload1.c: Unconditionally include code that was previously
2742 conditional on SECONDARY_MEMORY_NEEDED.
2743 (choose_reload_regs): Use targetm.secondary_memory_needed
2744 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2745 (gen_reload): Likewise.
2746 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2748 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2749 Alan Hayward <alan.hayward@arm.com>
2750 David Sherwood <david.sherwood@arm.com>
2752 * target.def (secondary_memory_needed_mode): New hook:
2753 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2754 * targhooks.h (default_secondary_memory_needed_mode): New function.
2755 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2756 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2757 * doc/tm.texi: Regenerate.
2758 * lra-constraints.c (check_and_process_move): Use
2759 targetm.secondary_memory_needed_mode instead of
2760 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2761 (curr_insn_transform): Likewise.
2762 * reload.c (get_secondary_mem): Likewise.
2763 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2764 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2766 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2767 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2768 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2769 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2770 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2772 * config/powerpcspe/powerpcspe-protos.h
2773 (rs6000_secondary_memory_needed_mode): Delete.
2774 * config/powerpcspe/powerpcspe.c
2775 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2776 (rs6000_secondary_memory_needed_mode): Make static.
2777 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2778 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2780 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2782 (rs6000_secondary_memory_needed_mode): Make static.
2783 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2784 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2785 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2786 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2787 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2789 (sparc_secondary_memory_needed_mode): New function.
2790 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2792 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2794 * config/aarch64/constraints.md (Umq): New constraint.
2795 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2797 (mov<mode>): Update condition.
2799 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2801 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2802 when bitposition is the same.
2804 2017-09-13 Richard Biener <rguenther@suse.de>
2806 * dwarf2out.c (output_die_symbol): Remove.
2807 (output_die): Do not output a DIEs symbol.
2809 2017-09-13 Richard Biener <rguenther@suse.de>
2812 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2813 default-def to avoid breaking iterator update with the weird
2814 interaction with cgraph_update_edges_for_call_stmt_node.
2816 2017-09-13 Richard Biener <rguenther@suse.de>
2818 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2819 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2820 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2821 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2823 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2825 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2826 Disable pc relative literal load irrespective of
2827 TARGET_FIX_ERR_A53_84341 for default.
2829 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2831 * config/sparc/sparc.c (output_return): Output the source location of
2832 the insn in the delay slot, if any.
2833 (output_sibcall): Likewise.
2835 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2838 * common.opt (-static-pie): New alias.
2839 (shared): Negate static-pie.
2840 (-no-pie): Update help text.
2842 (static-pie): New option.
2843 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2844 -static-pie support.
2845 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2846 (LINK_EH_SPEC): Likewise.
2847 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2848 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2849 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2850 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2851 (init_gcc_specs): Likewise.
2852 (init_spec): Likewise.
2853 (display_help): Update help message for -pie.
2854 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2857 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2859 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2860 (movdi_aarch64): Likewise.
2861 (movti_aarch64): Likewise.
2863 2017-09-12 Simon Wright <simon@pushface.org>
2866 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2867 calculation of the minor version, always output as 0.
2869 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2872 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2873 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2874 on it early, rather than manual conversion late. For
2875 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2876 instead of performing manual conversion.
2878 2017-09-12 Carl Love <cel@us.ibm.com>
2880 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2881 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2883 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2884 VMULOSW): Add definitions.
2885 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2886 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2887 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2888 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2889 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2891 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2893 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2895 (movti_aarch64): Likewise.
2896 (movdf_aarch64): Likewise.
2897 (movtf_aarch64): Likewise.
2898 (load_pairdi): Likewise.
2899 (store_pairdi): Likewise.
2900 (load_pairdf): Likewise.
2901 (store_pairdf): Likewise.
2902 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2903 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2904 (ldr_got_small_<mode>): Likewise.
2905 (ldr_got_small_28k_<mode>): Likewise.
2906 (ldr_got_tiny): Likewise.
2907 * config/aarch64/iterators.md (ldst_sz): New.
2908 (ldpstp_sz): Likewise.
2909 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2911 (thunderx_load): Split load_8 to load_16.
2912 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2914 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2915 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2916 (xgene1_store_pair): Split store_8 to store_16.
2917 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2918 (falkor_st_0_st_sd): Split store_8 to store_16.
2920 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2922 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2923 and store1/2/3/4 to store_4/8/12/16.
2924 * config/aarch64/aarch64.md: Update for rename.
2925 * config/arm/arm.md: Likewise.: Likewise.
2926 * config/arm/arm.c: Likewise.
2927 * config/arm/thumb1.md: Likewise.
2928 * config/arm/thumb2.md: Likewise.
2929 * config/arm/vfp.md: Likewise.
2930 * config/arm/arm-generic.md: Likewise.
2931 * config/arm/arm1020e.md: Likewise.
2932 * config/arm/arm1026ejs.md: Likewise.
2933 * config/arm/arm1136jfs.md: Likewise.
2934 * config/arm/arm926ejs.md: Likewise.
2935 * config/arm/cortex-a15.md: Likewise.
2936 * config/arm/cortex-a17.md: Likewise.
2937 * config/arm/cortex-a5.md: Likewise.
2938 * config/arm/cortex-a53.md: Likewise.
2939 * config/arm/cortex-a57.md: Likewise.
2940 * config/arm/cortex-a7.md: Likewise.
2941 * config/arm/cortex-a8.md: Likewise.
2942 * config/arm/cortex-a9.md: Likewise.
2943 * config/arm/cortex-m4.md: Likewise.
2944 * config/arm/cortex-m7.md: Likewise.
2945 * config/arm/cortex-r4.md: Likewise.
2946 * config/arm/exynos-m1.md: Likewise.
2947 * config/arm/fa526.md: Likewise.
2948 * config/arm/fa606te.md: Likewise.
2949 * config/arm/fa626te.md: Likewise.
2950 * config/arm/fa726te.md: Likewise.
2951 * config/arm/fmp626.md: Likewise.
2952 * config/arm/iwmmxt.md: Likewise.
2953 * config/arm/ldmstm.md: Likewise.
2954 * config/arm/marvell-pj4.md: Likewise.
2955 * config/arm/xgene1.md: Likewise.
2956 * config/aarch64/thunderx.md: Likewise.
2957 * config/aarch64/thunderx2t99.md: Likewise.
2958 * config/aarch64/falkor.md: Likewise.
2960 2017-09-12 Martin Liska <mliska@suse.cz>
2962 * attribs.c (private_lookup_attribute): New function.
2963 * attribs.h (private_lookup_attribute): Declared here.
2964 (lookup_attribute): Called from this place.
2966 2017-09-12 Richard Biener <rguenther@suse.de>
2968 PR tree-optimization/82157
2969 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2970 stmts with side-effects.
2972 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2973 Alan Hayward <alan.hayward@arm.com>
2974 David Sherwood <david.sherwood@arm.com>
2976 * target.def (hard_regno_nregs): New hook.
2977 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2978 * targhooks.h (default_hard_regno_nregs): Declare.
2979 * targhooks.c (default_hard_regno_nregs): New function.
2980 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2981 (TARGET_HARD_REGNO_NREGS): ...this hook.
2982 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2983 (CLASS_MAX_NREGS): Likewise.
2984 * doc/tm.texi: Regenerate.
2985 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2986 instead of HARD_REGNO_NREGS.
2987 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2988 HARD_REGNO_NREGS in the comment.
2989 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2990 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2991 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2992 Return an unsigned int.
2993 (TARGET_HARD_REGNO_NREGS): Redefine.
2994 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2995 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2996 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2997 (arc_hard_regno_nregs): New function.
2998 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2999 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3000 (arm_hard_regno_nregs): New function.
3001 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3002 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3003 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3004 (TARGET_HARD_REGNO_NREGS): Redefine.
3005 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3006 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3007 (HARD_REGNO_NREGS): Delete.
3008 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3009 (cr16_hard_regno_nregs): New function.
3010 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3011 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3012 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3013 (cris_hard_regno_nregs): New function.
3014 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3015 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3016 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3017 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3018 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3019 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3020 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3021 (frv_hard_regno_nregs): Make static. Take and return an
3023 (frv_class_max_nregs): Remove outdated copy of documentation.
3024 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3025 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3026 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3027 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3028 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3029 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3030 (TARGET_HARD_REGNO_NREGS): Redefine.
3031 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3032 (CLASS_MAX_NREGS): Update comment.
3033 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3034 (ia64_hard_regno_nregs): New function.
3035 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3036 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3037 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3038 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3039 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3041 (m32c_hard_regno_nregs): Likewise. Make static.
3042 (TARGET_HARD_REGNO_NREGS): Redefine.
3043 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3044 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3045 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3046 (m68k_hard_regno_nregs): New function.
3047 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3048 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3049 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3050 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3051 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3052 Take and return an unsigned int.
3053 (TARGET_HARD_REGNO_NREGS): Redefine.
3054 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3055 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3056 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3057 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3058 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3059 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3060 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3061 (msp430_hard_regno_nregs): Make static. Take and return an
3063 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3064 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3065 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3066 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3067 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3068 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3069 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3070 (TARGET_HARD_REGNO_NREGS): Redefine.
3071 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3072 (PA_HARD_REGNO_NREGS): ...this.
3073 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3074 (PA_HARD_REGNO_NREGS): ...this.
3075 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3076 (pa_hard_regno_nregs): New function.
3077 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3078 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3079 (pdp11_hard_regno_nregs): New function.
3080 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3081 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3082 (rs6000_hard_regno_nregs_hook): New function.
3083 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3084 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3085 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3086 Take and return an unsigned int. Move earlier in file.
3087 (TARGET_HARD_REGNO_NREGS): Redefine.
3088 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3089 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3090 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3091 (rl78_hard_regno_nregs): Make static. Take and return an
3093 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3094 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3095 (rs6000_hard_regno_nregs_hook): New function.
3096 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3097 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3098 (TARGET_HARD_REGNO_NREGS): Redefine.
3099 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3100 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3101 instead of HARD_REGNO_NREGS.
3102 (s390_hard_regno_nregs): New function.
3103 (s390_hard_regno_mode_ok): Add comment from s390.h.
3104 (TARGET_HARD_REGNO_NREGS): Redefine.
3105 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3106 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3107 (sh_hard_regno_nregs): New function.
3108 (sh_pass_in_reg_p): Use it.
3109 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3110 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3111 (sparc_hard_regno_nregs): New function.
3112 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3113 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3114 (spu_function_arg_advance): Use it, supplying a valid register number.
3115 (TARGET_HARD_REGNO_NREGS): Redefine.
3116 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3117 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3118 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3119 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3120 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3121 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3122 (CLASS_MAX_NREGS): Remove copy of old documentation.
3123 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3124 (visium_hard_regno_nregs): New function.
3125 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3126 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3127 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3128 xtensa_hard_regno_nregs): New function.
3129 * system.h (HARD_REGNO_NREGS): Poison.
3131 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3133 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3134 hard_regno_nregs instead of HARD_REGNO_NREGS.
3135 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3136 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3137 (c6x_expand_epilogue): Likewise.
3138 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3139 (frv_read_iacc_argument): Likewise.
3140 * config/sh/sh.c: Include regs.h.
3141 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3142 (regs_used): Likewise.
3143 (output_stack_adjust): Likewise.
3144 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3145 * expmed.c: Include regs.h.
3146 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3147 * ree.c: Include regs.h.
3148 (combine_reaching_defs): Use hard_regno_nregs instead of
3150 (add_removable_extension): Likewise.
3152 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3154 * regs.h (hard_regno_nregs): Turn into a function.
3155 (end_hard_regno): Update accordingly.
3156 * caller-save.c (setup_save_areas): Likewise.
3157 (save_call_clobbered_regs): Likewise.
3158 (replace_reg_with_saved_mem): Likewise.
3159 (insert_restore): Likewise.
3160 (insert_save): Likewise.
3161 * combine.c (can_change_dest_mode): Likewise.
3162 (move_deaths): Likewise.
3163 (distribute_notes): Likewise.
3164 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3165 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3166 (rs6000_split_multireg_move): Likewise.
3167 (rs6000_register_move_cost): Likewise.
3168 (rs6000_memory_move_cost): Likewise.
3169 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3170 (rs6000_split_multireg_move): Likewise.
3171 (rs6000_register_move_cost): Likewise.
3172 (rs6000_memory_move_cost): Likewise.
3173 * cselib.c (cselib_reset_table): Likewise.
3174 (cselib_lookup_1): Likewise.
3175 * emit-rtl.c (set_mode_and_regno): Likewise.
3176 * function.c (aggregate_value_p): Likewise.
3177 * ira-color.c (setup_profitable_hard_regs): Likewise.
3178 (check_hard_reg_p): Likewise.
3179 (calculate_saved_nregs): Likewise.
3180 (assign_hard_reg): Likewise.
3181 (improve_allocation): Likewise.
3182 (calculate_spill_cost): Likewise.
3183 * ira-emit.c (modify_move_list): Likewise.
3184 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3185 (ira_hard_reg_in_set_p): Likewise.
3186 * ira.c (setup_reg_mode_hard_regset): Likewise.
3187 (clarify_prohibited_class_mode_regs): Likewise.
3188 (check_allocation): Likewise.
3189 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3190 (lra_setup_reg_renumber): Likewise.
3191 (setup_try_hard_regno_pseudos): Likewise.
3192 (spill_for): Likewise.
3193 (assign_hard_regno): Likewise.
3194 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3195 * lra-constraints.c (in_class_p): Likewise.
3196 (lra_constraint_offset): Likewise.
3197 (simplify_operand_subreg): Likewise.
3198 (lra_constraints): Likewise.
3199 (split_reg): Likewise.
3200 (split_if_necessary): Likewise.
3201 (invariant_p): Likewise.
3202 (inherit_in_ebb): Likewise.
3203 * lra-lives.c (process_bb_lives): Likewise.
3204 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3205 (get_hard_regs): Likewise.
3206 (do_remat): Likewise.
3207 * lra-spills.c (assign_spill_hard_regs): Likewise.
3208 * mode-switching.c (create_pre_exit): Likewise.
3209 * postreload.c (reload_combine_recognize_pattern): Likewise.
3210 * recog.c (peep2_find_free_register): Likewise.
3211 * regcprop.c (kill_value_regno): Likewise.
3212 (set_value_regno): Likewise.
3213 (copy_value): Likewise.
3214 (maybe_mode_change): Likewise.
3215 (find_oldest_value_reg): Likewise.
3216 (copyprop_hardreg_forward_1): Likewise.
3217 * regrename.c (check_new_reg_p): Likewise.
3218 (regrename_do_replace): Likewise.
3219 * reload.c (push_reload): Likewise.
3220 (combine_reloads): Likewise.
3221 (find_dummy_reload): Likewise.
3222 (operands_match_p): Likewise.
3223 (find_reloads): Likewise.
3224 (find_equiv_reg): Likewise.
3225 (reload_adjust_reg_for_mode): Likewise.
3226 * reload1.c (count_pseudo): Likewise.
3227 (count_spilled_pseudo): Likewise.
3228 (find_reg): Likewise.
3229 (clear_reload_reg_in_use): Likewise.
3230 (free_for_value_p): Likewise.
3231 (allocate_reload_reg): Likewise.
3232 (choose_reload_regs): Likewise.
3233 (reload_adjust_reg_for_temp): Likewise.
3234 (emit_reload_insns): Likewise.
3235 (delete_output_reload): Likewise.
3236 * rtlanal.c (subreg_get_info): Likewise.
3237 * sched-deps.c (sched_analyze_reg): Likewise.
3238 * sel-sched.c (init_regs_for_mode): Likewise.
3239 (mark_unavailable_hard_regs): Likewise.
3240 (choose_best_reg_1): Likewise.
3241 (verify_target_availability): Likewise.
3242 * valtrack.c (dead_debug_insert_temp): Likewise.
3243 * var-tracking.c (track_loc_p): Likewise.
3244 (emit_note_insn_var_location): Likewise.
3245 * varasm.c (make_decl_rtl): Likewise.
3246 * reginfo.c (choose_hard_reg_mode): Likewise.
3247 (init_reg_modes_target): Refer directly to
3248 this_target_regs->x_hard_regno_nregs.
3250 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3252 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3253 instead of hard_regno_nregs.
3255 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3257 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3258 end_hard_regno instead of hard_regno_nregs.
3259 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3260 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3261 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3262 * ira-color.c (improve_allocation): Likewise.
3263 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3264 * lra-lives.c (mark_regno_live): Likewise.
3265 (mark_regno_dead): Likewise.
3266 * lra-remat.c (operand_to_remat): Likewise.
3267 * lra.c (collect_non_operand_hard_regs): Likewise.
3268 * postreload.c (reload_combine_note_store): Likewise.
3269 (move2add_valid_value_p): Likewise.
3270 * reload.c (regno_clobbered_p): Likewise.
3272 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3274 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3276 * config/v850/v850.c (v850_reorg): Likewise.
3277 * reload.c (refers_to_regno_for_reload_p): Likewise.
3278 (find_equiv_reg): Likewise.
3279 * reload1.c (reload_reg_reaches_end_p): Likewise.
3281 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3283 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3285 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3286 * config/arm/arm.c (output_move_neon): Likewise.
3287 (arm_attr_length_move_neon): Likewise.
3288 (neon_split_vcombine): Likewise.
3289 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3290 (c6x_mark_reg_written): Likewise.
3291 (c6x_dwarf_register_span): Likewise.
3292 * config/i386/i386.c (ix86_save_reg): Likewise.
3293 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3294 (rws_access_reg): Likewise.
3295 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3296 * mode-switching.c (create_pre_exit): Likewise.
3297 * ree.c (combine_reaching_defs): Likewise.
3298 (add_removable_extension): Likewise.
3299 * regcprop.c (find_oldest_value_reg): Likewise.
3300 (copyprop_hardreg_forward_1): Likewise.
3301 * reload.c (reload_inner_reg_of_subreg): Likewise.
3302 (push_reload): Likewise.
3303 (combine_reloads): Likewise.
3304 (find_dummy_reload): Likewise.
3305 (reload_adjust_reg_for_mode): Likewise.
3306 * reload1.c (find_reload_regs): Likewise.
3307 (forget_old_reloads_1): Likewise.
3308 (reload_reg_free_for_value_p): Likewise.
3309 (reload_adjust_reg_for_temp): Likewise.
3310 (emit_reload_insns): Likewise.
3311 (delete_output_reload): Likewise.
3312 * sel-sched.c (choose_best_reg_1): Likewise.
3313 (choose_best_pseudo_reg): Likewise.
3315 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3316 Alan Hayward <alan.hayward@arm.com>
3317 David Sherwood <david.sherwood@arm.com>
3319 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3320 * target.def (slow_unaligned_access): New hook.
3321 * targhooks.h (default_slow_unaligned_access): Declare.
3322 * targhooks.c (default_slow_unaligned_access): New function.
3323 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3324 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3325 * doc/tm.texi: Regenerate.
3326 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3327 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3328 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3330 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3331 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3333 (rs6000_slow_unaligned_access): New function.
3334 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3335 (expand_block_compare): Likewise.
3336 (expand_strn_compare): Likewise.
3337 (rs6000_rtx_costs): Likewise.
3338 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3339 (riscv_slow_unaligned_access): Likewise.
3340 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3341 (riscv_slow_unaligned_access_p): ...this and make static.
3342 (riscv_option_override): Update accordingly.
3343 (riscv_slow_unaligned_access): New function.
3344 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3345 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3346 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3347 (rs6000_slow_unaligned_access): New function.
3348 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3349 (rs6000_rtx_costs): Likewise.
3350 * config/rs6000/rs6000-string.c (expand_block_compare)
3351 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3352 of SLOW_UNALIGNED_ACCESS.
3353 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3354 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3355 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3356 of SLOW_UNALIGNED_ACCESS.
3357 * expmed.c (simple_mem_bitfield_p): Likewise.
3358 * expr.c (alignment_for_piecewise_move): Likewise.
3359 (emit_group_load_1): Likewise.
3360 (emit_group_store): Likewise.
3361 (copy_blkmode_from_reg): Likewise.
3362 (emit_push_insn): Likewise.
3363 (expand_assignment): Likewise.
3364 (store_field): Likewise.
3365 (expand_expr_real_1): Likewise.
3366 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3367 * lra-constraints.c (simplify_operand_subreg): Likewise.
3368 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3369 * gimple-ssa-store-merging.c: Likewise in block comment at start
3371 * tree-ssa-strlen.c: Include target.h.
3372 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3373 of SLOW_UNALIGNED_ACCESS.
3374 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3376 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3378 PR rtl-optimization/82185
3379 * expmed.c (emit_store_flag_int): Only test tem if it has been
3382 2017-09-12 Richard Biener <rguenther@suse.de>
3385 * match.pd ((FTYPE) N CMP CST): Fix typo.
3387 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3389 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3391 (mips_near_type_p): Add 'short_call' attribute as a synonym
3393 * doc/extend.texi (short_call): Document new function attribute.
3395 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3398 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3399 assertion check that in the condition.
3400 (get_atomic_generic_size): Likewise. Before testing if parameter
3401 has pointer type, if it has array type, call for C++
3402 default_conversion to perform array-to-pointer conversion.
3404 2017-09-12 Richard Biener <rguenther@suse.de>
3406 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3407 for operations we cannot scalarize.
3409 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3411 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3412 vectors heap vectors. Clean up comments.
3413 Make visited_bbs a reference.
3414 (profitable_jump_thread_path): Make GC
3415 vectors heap vectors. Clean up comments.
3417 (convert_and_register_jump_thread_path): Make GC vectors heap
3419 (check_subpath_and_update_thread_path): Same. Clean up comments.
3420 Make visited_bbs a reference.
3421 (handle_phi): Abstract common code to to
3422 register_jump_thread_path_if_profitable.
3423 Rename VAR_BB to DEF_BB.
3425 Make GC vectors heap vectors.
3426 Make visited_bbs a reference.
3427 (handle_assignment): Same.
3428 (register_jump_thread_path_if_profitable): New.
3429 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3431 Make GC vectors heap vectors. Clean up comments.
3432 Make visited_bbs a reference.
3433 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3434 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3437 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3440 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3441 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3443 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3445 Revert r251800 and r251799.
3447 2017-09-11 Martin Jambor <mjambor@suse.cz>
3450 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3451 arguments in advance.
3452 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3453 use it to find predecessor edges.
3454 (naive_outof_ssa): Collect vector of predecessors.
3456 2017-09-08 Jason Merrill <jason@redhat.com>
3458 PR c++/70029 - ICE with ref-qualifier and -flto
3459 * langhooks.h (struct lang_hooks_for_types): Add
3460 copy_lang_qualifiers.
3461 * attribs.c (build_type_attribute_qual_variant): Use it.
3462 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3464 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3465 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3467 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3470 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3471 (*mulsi3_sp64): New instruction.
3472 (mulsi3): New expander.
3474 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3476 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3478 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3480 * sancov.c: Include memmodel.h.
3482 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3485 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3488 2017-09-07 Carl Love <cel@us.ibm.com>
3490 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3491 the sldi instruction.
3493 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3495 * sancov.c: Include tm_p.h.
3497 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3500 * output.h (switch_to_other_text_partition): New declaration.
3501 * varasm.c (switch_to_other_text_partition): New function.
3502 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3503 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3504 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3505 to the other text partition before emitting LCL label and switch back
3506 after emitting the word after it.
3508 2017-09-07 Richard Biener <rguenther@suse.de>
3510 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3511 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3512 critical edges here, after loop init.
3513 (pass_data_pre): Remove PROP_no_crit_edges flags.
3514 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3515 for valueization of call args to avoid leaking VN_TOP.
3516 (visit_use): Assert we do not visit default defs.
3517 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3518 Use error_mark_node to more easily detect leaking VN_TOP.
3519 All default-defs are varying, not VN_TOP. Mark them visited.
3520 (run_scc_vn): Make code match comment.
3522 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3524 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3525 OPTION_MASK_FLOAT128_KEYWORD.
3526 (POWERPC_MASKS): Likewise.
3527 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3528 support for the -mfloat128-type option, and make -mfloat128
3529 default on PowerPC Linux systems. Define or undefine
3530 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3531 Define __float128 to be __ieee128 if IEEE 128-bit support is
3532 enabled, or undefine it.
3533 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3534 Delete defining __FLOAT128_TYPE__.
3535 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3536 -mfloat128-type option and make -mfloat128 default on PowerPC
3538 (TARGET_FLOAT128_TYPE): Likewise.
3539 (-mfloat128-type): Likewise.
3540 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3541 Delete the -mfloat128-type option and make -mfloat128 default on
3542 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3543 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3544 128-bit floating point is enabled. Change tests from using
3545 -mfloat128-type to -mfloat128.
3546 (rs6000_mangle_type): Use the correct mangling for the __float128
3547 type even if normal long double is restricted to 64-bits.
3548 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3550 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3551 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3552 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3554 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3555 documentation for -mfloat128.
3557 2017-09-06 Olivier Hainque <hainque@adacore.com>
3559 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3561 2017-09-06 Wish Wu <wishwu007@gmail.com>
3562 Jakub Jelinek <jakub@redhat.com>
3564 * asan.c (initialize_sanitizer_builtins): Add
3565 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3566 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3567 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3568 BT_FN_VOID_UINT64_PTR variables.
3569 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3570 (BT_FN_VOID_UINT16_UINT16): Likewise.
3571 (BT_FN_VOID_UINT32_UINT32): Likewise.
3572 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3573 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3574 (BT_FN_VOID_UINT64_PTR): Likewise.
3575 * common.opt (flag_sanitize_coverage): New variable.
3576 (fsanitize-coverage=trace-pc): Remove.
3577 (fsanitize-coverage=): Add.
3578 * flag-types.h (enum sanitize_coverage_code): New enum.
3579 * fold-const.c (fold_range_test): Disable non-short-circuit
3580 optimization if flag_sanitize_coverage.
3581 (fold_truth_andor): Likewise.
3582 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3583 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3584 (coverage_sanitizer_opts): New array.
3585 (get_closest_sanitizer_option): Add OPTS argument, handle also
3586 OPT_fsanitize_coverage_.
3587 (parse_sanitizer_options): Adjusted to also handle
3588 OPT_fsanitize_coverage_.
3589 (common_handle_option): Add OPT_fsanitize_coverage_.
3590 * sancov.c (instrument_comparison, instrument_switch): New function.
3591 (sancov_pass): Add trace-cmp support.
3592 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3593 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3594 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3595 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3596 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3597 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3598 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3599 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3600 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3601 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3603 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3605 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3606 error. Only quit immediately if parsing is complete.
3607 (BEGIN): Initialize fatal_err and parse_done.
3608 (begin fpu, end fpu): Check number of arguments.
3609 (begin arch, end arch): Likewise.
3610 (begin cpu, end cpu): Likewise.
3611 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3612 (optalias): Likewise.
3614 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3616 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3617 * config/arm/arm-isa.h: Delete. Move definitions to ...
3618 * arm-cpus.in: ... here. Use new feature and fgroup values.
3619 * config/arm/arm.c (arm_option_override): Use lower case for feature
3621 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3622 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3623 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3625 (print_isa_bits_for): New function.
3626 (gen_isa): New function.
3627 (gen_comm_data): Use print_isa_bits_for.
3628 (define feature): New keyword.
3629 (define fgroup): New keyword.
3630 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3631 (arm-isa.h): Add rule to generate file.
3632 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3633 case for feature bit names.
3635 2017-09-06 Richard Biener <rguenther@suse.de>
3637 * tree-ssa-pre.c (NECESSARY): Remove.
3638 (create_expression_by_pieces): Do not touch pass-local flags.
3639 (insert_into_preds_of_block): Likewise.
3640 (do_pre_regular_insertion): Likewise.
3641 (eliminate_insert): Likewise.
3642 (eliminate_dom_walker::before_dom_children): Likewise.
3643 (fini_eliminate): Do not look at inserted_exprs.
3644 (mark_operand_necessary): Remove.
3645 (remove_dead_inserted_code): Replace with simple work-list
3646 algorithm based on inserted_exprs and SSA uses.
3647 (pass_pre::execute): Re-order fini_eliminate and
3648 remove_dead_inserted_code.
3650 2017-09-06 Olivier Hainque <hainque@adacore.com>
3652 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3653 for VxWorks 7. Adjust surrounding comments.
3655 2017-09-06 Richard Biener <rguenther@suse.de>
3657 * gimple-ssa-strength-reduction.c
3658 (find_candidates_dom_walker::before_dom_children): Also allow
3661 2017-09-06 Richard Biener <rguenther@suse.de>
3663 PR tree-optimization/82108
3664 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3665 for gap in the non-permutation SLP case.
3667 2017-09-06 Martin Jambor <mjambor@suse.cz>
3669 PR tree-optimization/82078
3670 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3671 add_access_to_work_queue...
3672 (build_accesses_from_assign): ...here.
3673 (propagate_all_subaccesses): Make sure racc is the group
3674 representative, if there is one.
3676 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3679 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3682 2017-09-06 Richard Biener <rguenther@suse.de>
3684 * gimple-ssa-strength-reduction.c
3685 (find_candidates_dom_walker::before_doom_children): Use a
3686 type and not a mode check.
3688 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3691 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3693 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3695 2017-09-05 Jeff Law <law@redhat.com>
3697 PR tree-optimization/64910
3698 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3699 cases where we have 3 or more operands.
3701 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3704 * omp-low.c (lower_omp_for): Recompute tree invariant if
3705 gimple_omp_for_initial/final is ADDR_EXPR.
3708 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3709 into gimple val before gimplification fo the COND_EXPR.
3711 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3713 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3714 REGION_COPY argument.
3715 (thread_through_all_blocks): Remove unused argument to
3716 duplicate_thread_path.
3718 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3719 Alan Hayward <alan.hayward@arm.com>
3720 David Sherwood <david.sherwood@arm.com>
3722 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3723 Take a scalar_mode rather than a machine_mode.
3724 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3725 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3726 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3727 (aarch64_gen_adjusted_ldpstp): Likewise.
3728 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3730 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3731 Alan Hayward <alan.hayward@arm.com>
3732 David Sherwood <david.sherwood@arm.com>
3734 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3735 Take a scalar_int_mode instead of a machine_mode.
3736 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3737 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3738 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3739 (aarch64_simd_attr_length_rglist): Delete.
3740 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3741 a scalar_int_mode instead of a machine_mode.
3742 (aarch64_add_offset): Likewise.
3743 (aarch64_internal_mov_immediate): Likewise
3744 (aarch64_add_constant_internal): Likewise.
3745 (aarch64_add_constant): Likewise.
3746 (aarch64_movw_imm): Likewise.
3747 (aarch64_rtx_arith_op_extract_p): Likewise.
3748 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3749 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3750 Remove assert that the mode isn't a vector.
3751 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3752 (aarch64_expand_mov_immediate): Update calls after above changes.
3753 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3754 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3755 (aarch64_move_imm): Likewise.
3756 (aarch64_can_const_movi_rtx_p): Likewise.
3757 (aarch64_strip_extend): Likewise.
3758 (aarch64_extr_rtx_p): Likewise.
3759 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3760 a CONST_INT when the mode parameter is VOIDmode.
3761 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3763 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3765 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3766 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3767 (bitwise_type_for_mode): Update accordingly.
3769 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3771 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3772 * stor-layout.c (mode_for_size_tree): Likewise.
3773 (mode_for_array): Update accordingly.
3774 (layout_decl): Likewise.
3775 (compute_record_mode): Likewise. Only set the mode once.
3777 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3779 * target.def (get_mask_mode): Change return type to opt_mode.
3781 * doc/tm.texi: Regenerate.
3782 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3783 * targhooks.c (default_get_mask_mode): Likewise.
3784 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3785 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3786 targetm.get_mask_mode.
3787 * tree.c (build_truth_vector_type): Likewise.
3789 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3791 * machmode.h (mode_for_vector): Return an opt_mode.
3792 * stor-layout.c (mode_for_vector): Likewise.
3793 (mode_for_int_vector): Update accordingly.
3794 (layout_type): Likewise.
3795 * config/i386/i386.c (emit_memmov): Likewise.
3796 (ix86_expand_set_or_movmem): Likewise.
3797 (ix86_expand_vector_init): Likewise.
3798 (ix86_get_mask_mode): Likewise.
3799 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3801 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3802 * expmed.c (extract_bit_field_1): Likewise.
3803 * expr.c (expand_expr_real_2): Likewise.
3804 * optabs-query.c (can_vec_perm_p): Likewise.
3805 (can_vec_mask_load_store_p): Likewise.
3806 * optabs.c (expand_vec_perm): Likewise.
3807 * targhooks.c (default_get_mask_mode): Likewise.
3808 * tree-vect-stmts.c (vectorizable_store): Likewise.
3809 (vectorizable_load): Likewise.
3810 (get_vectype_for_scalar_type_and_size): Likewise.
3812 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3814 * machmode.h (mode_for_int_vector): New function.
3815 * stor-layout.c (mode_for_int_vector): Likewise.
3816 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3817 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3818 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3819 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3820 (s390_expand_vcond): Likewise.
3822 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3824 * machmode.h (opt_machine_mode): New type.
3825 (opt_mode<T>): Allow construction from anything that can be
3827 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3828 (mode_for_size): Return an opt_machine_mode.
3829 * stor-layout.c (mode_for_size): Likewise.
3830 (mode_for_size_tree): Update call accordingly.
3831 (bitwise_mode_for_mode): Likewise.
3832 (make_fract_type): Likewise.
3833 (make_accum_type): Likewise.
3834 * caller-save.c (replace_reg_with_saved_mem): Update call
3836 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3837 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3838 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3839 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3840 * expmed.c (extract_bit_field_1): Likewise.
3841 * reload.c (get_secondary_mem): Likewise.
3842 * varasm.c (assemble_integer): Likewise.
3843 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3846 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3848 * machmode.h (decimal_float_mode_for_size): New function.
3849 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3850 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3851 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3852 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3853 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3854 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3856 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3858 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3859 (get_builtin_sync_mode): Likewise.
3860 (expand_ifn_atomic_compare_exchange): Likewise.
3861 (expand_builtin_atomic_clear): Likewise.
3862 (expand_builtin_atomic_test_and_set): Likewise.
3863 (fold_builtin_atomic_always_lock_free): Likewise.
3864 * calls.c (compute_argument_addresses): Likewise.
3865 (emit_library_call_value_1): Likewise.
3866 (store_one_arg): Likewise.
3867 * combine.c (combine_instructions): Likewise.
3868 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3869 * config/arm/arm.c (arm_function_value): Likewise.
3870 (aapcs_allocate_return_reg): Likewise.
3871 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3872 * config/i386/i386.c (construct_container): Likewise.
3873 (ix86_gimplify_va_arg): Likewise.
3874 (ix86_expand_sse_cmp): Likewise.
3875 (emit_memmov): Likewise.
3876 (emit_memset): Likewise.
3877 (expand_small_movmem_or_setmem): Likewise.
3878 (ix86_expand_pextr): Likewise.
3879 (ix86_expand_pinsr): Likewise.
3880 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3881 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3883 * config/mips/mips.c (mips_function_value_1) Likewise.
3884 (mips_block_move_straight): Likewise.
3885 (mips_expand_ins_as_unaligned_store): Likewise.
3886 * config/powerpcspe/powerpcspe.c
3887 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3888 (rs6000_darwin64_record_arg_flush): Likewise.
3889 * config/rs6000/rs6000.c
3890 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3891 (rs6000_darwin64_record_arg_flush): Likewise.
3892 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3893 (sparc_function_value_1): Likewise.
3894 * config/spu/spu.c (adjust_operand): Likewise.
3895 (spu_emit_branch_or_set): Likewise.
3896 (arith_immediate_p): Likewise.
3897 * emit-rtl.c (gen_lowpart_common): Likewise.
3898 * expr.c (expand_expr_real_1): Likewise.
3899 * function.c (assign_parm_setup_block): Likewise.
3900 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3901 * reload1.c (alter_reg): Likewise.
3902 * stor-layout.c (mode_for_vector): Likewise.
3903 (layout_type): Likewise.
3905 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3907 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3908 (spu_convert_move): Likewise.
3909 * lower-subreg.c (resolve_simple_move): Likewise.
3911 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3914 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3915 define_insn to a define_expand.
3916 (altivec_vsum2sws_direct): New define_insn.
3917 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3919 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3921 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3924 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3929 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3930 realign stack if stack alignment needed is less than incoming
3933 2017-09-05 Marek Polacek <polacek@redhat.com>
3936 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3939 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3941 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3943 2017-09-05 Richard Biener <rguenther@suse.de>
3945 PR tree-optimization/82084
3946 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3948 2017-09-05 Richard Biener <rguenther@suse.de>
3950 PR tree-optimization/82102
3951 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3953 2017-09-05 Martin Liska <mliska@suse.cz>
3955 PR tree-optimization/82032
3956 * tree-cfg.c (generate_range_test): New function.
3957 * tree-cfg.h (generate_range_test): Declared here.
3958 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3959 (cleanup_control_expr_graph): Use it.
3960 * tree-switch-conversion.c (try_switch_expansion): Remove
3962 (emit_case_nodes): Use generate_range_test.
3964 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3967 * config/i386/i386.md (*<btsc><mode>_mask): Add
3968 TARGET_USE_BT to insn constraint.
3969 (*btr<mode>_mask): Ditto.
3971 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3973 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3974 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3976 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3979 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3980 TARGET_NEON and TARGET_IWMMXT.
3981 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3982 TARGET_NEON and TARGET_IWMMXT.
3983 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3985 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3987 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3988 (ix86_rewrite_tls_address): Ditto.
3989 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3990 (ix86_rewrite_tls_address_1): Ditto.
3991 (ix86_rewrite_tls_address): Ditto.
3992 * config/i386/predicates.md (tls_address_pattern): New predicate.
3993 * config/i386/i386.md (TLS address splitter): New splitter.
3995 2017-09-04 Richard Biener <rguenther@suse.de>
3997 PR tree-optimization/82084
3998 * fold-const.h (can_native_encode_string_p): Declare.
3999 * fold-const.c (can_native_encode_string_p): Factor out from ...
4000 (native_encode_string): ... here.
4001 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4002 vectorizing stores from constants we later cannot handle.
4004 2017-09-04 Marek Polacek <polacek@redhat.com>
4007 * doc/invoke.texi: Update -Wtautological-compare documentation.
4009 2017-09-04 Jeff Law <law@redhat.com>
4011 PR tree-optimization/64910
4012 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4013 swap the first and last operand if the last is a constant.
4015 2017-09-04 Marek Polacek <polacek@redhat.com>
4018 * convert.c (do_narrow): When sanitizing signed integer overflows,
4019 bail out for signed types.
4020 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4022 2017-09-04 Richard Biener <rguenther@suse.de>
4024 PR tree-optimization/82060
4025 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4026 Move devirtualization after stmt folding and before EH/AB/noreturn
4027 cleanup to get the stmt refs canonicalized. Use a bool instead
4028 of gimple_modified_p since that doesn't work for NOPs. Schedule
4029 NOPs generated by folding for removal.
4031 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4032 Alan Hayward <alan.hayward@arm.com>
4033 David Sherwood <david.sherwood@arm.com>
4035 * coretypes.h (pad_direction): New enum.
4036 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4037 (FUNCTION_ARG_PADDING): Likewise.
4038 * target.def (function_arg_padding): New hook.
4039 * targhooks.h (default_function_arg_padding): Declare.
4040 * targhooks.c (default_function_arg_padding): New function.
4041 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4042 (TARGET_FUNCTION_ARG_PADDING): ...this.
4043 * doc/tm.texi: Regenerate.
4044 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4045 instead of direction.
4046 (compute_argument_addresses): Likewise.
4047 (load_register_parameters): Likewise.
4048 (emit_library_call_value_1): Likewise.
4049 (store_one_arg): Use targetm.calls.function_arg_padding instead
4050 of FUNCTION_ARG_PADDING.
4051 (must_pass_in_stack_var_size_or_pad): Likewise.
4052 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4053 (emit_group_store): Likewise.
4054 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4055 instead of FUNCTION_ARG_PADDING.
4056 (emit_push_insn): Likewise, and propagate enum change throughout
4058 * function.h (direction): Delete.
4059 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4061 * function.c (assign_parm_find_stack_rtl): Likewise.
4062 (assign_parm_setup_block_p): Likewise.
4063 (assign_parm_setup_block): Likewise.
4064 (gimplify_parameters): Likewise.
4065 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4066 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4068 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4069 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4070 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4071 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4072 (aarch64_function_arg_padding): ...this new function.
4073 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4074 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4075 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4076 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4077 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4078 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4079 (arm_pad_arg_upward): Replace with...
4080 (arm_function_arg_padding): ...this new function.
4081 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4083 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4084 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4085 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4086 (ia64_hpux_function_arg_padding): Replace with...
4087 (ia64_function_arg_padding): ...this new function. Use pad_direction
4088 instead of direction. Check for TARGET_HPUX.
4089 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4090 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4091 (iq2000_function_arg_padding): New function.
4092 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4093 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4094 (mips_function_arg_padding): ...this new function.
4095 (mips_pad_reg_upward): Update accordingly.
4096 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4097 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4098 targetm.calls.function_arg_padding.
4099 (FUNCTION_ARG_PADDING): Delete.
4100 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4101 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4102 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4103 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4104 (nios2_block_reg_padding): Return pad_direction instead of direction.
4105 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4106 instead of direction.
4107 (nios2_function_arg_padding): Likewise. Make static.
4108 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4109 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4110 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4111 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4112 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4113 (pa_function_arg_padding): Make static. Return pad_direction instead
4115 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4116 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4117 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4118 instead of direction. Use targetm.calls.function_arg_padding.
4119 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4120 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4121 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4122 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4123 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4125 (function_arg_padding): Rename to...
4126 (rs6000_function_arg_padding): ...this. Make static. Return
4127 pad_direction instead of direction.
4128 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4129 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4130 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4131 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4132 instead of direction. Use targetm.calls.function_arg_padding.
4133 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4134 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4135 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4136 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4137 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4138 (function_arg_padding): Rename to...
4139 (rs6000_function_arg_padding): ...this. Make static. Return
4140 pad_direction instead of direction.
4141 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4142 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4143 * config/s390/s390.c (s390_function_arg_padding): New function.
4144 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4145 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4146 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4147 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4148 (function_arg_padding): Rename to...
4149 (sparc_function_arg_padding): ...this. Make static. Return
4150 pad_direction instead of direction.
4151 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4152 * config/spu/spu.c (spu_function_arg_padding): New function.
4153 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4154 * system.h (FUNCTION_ARG_PADDING): Poison.
4156 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4157 Alan Hayward <alan.hayward@arm.com>
4158 David Sherwood <david.sherwood@arm.com>
4160 * target.def (modes_tieable_p): New hook.
4161 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4162 (TARGET_MODES_TIEABLE_P): ...this.
4163 * doc/tm.texi.in: Regenerate.
4164 * hooks.h (hook_bool_mode_mode_true): Declare.
4165 * hooks.c (hook_bool_mode_mode_true): New function.
4166 * combine.c (subst): Use targetm.modes_tieable_p instead of
4168 * dse.c (find_shift_sequence): Likewise.
4169 * expmed.c (extract_low_bits): Likewise.
4170 * lower-subreg.c: Include target.h.
4171 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4173 * rtlanal.c (rtx_cost): Likewise.
4174 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4175 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4176 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4177 (TARGET_MODES_TIEABLE_P): Redefine.
4178 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4179 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4180 (TARGET_MODES_TIEABLE_P): Redefine.
4181 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4182 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4183 (arc_modes_tieable_p): New function.
4184 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4185 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4186 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4187 (arm_modes_tieable_p): Make static.
4188 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4189 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4190 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4191 (TARGET_MODES_TIEABLE_P): Redefine.
4192 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4193 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4194 (TARGET_MODES_TIEABLE_P): Redefine.
4195 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4196 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4197 (cr16_modes_tieable_p): New function.
4198 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4199 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4200 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4201 (TRULY_NOOP_TRUNCATION): Update comment.
4202 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4203 (TRULY_NOOP_TRUNCATION): Update comment.
4204 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4205 (frv_modes_tieable_p): New function.
4206 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4207 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4208 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4209 (TARGET_MODES_TIEABLE_P): Redefine.
4210 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4211 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4212 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4213 (TARGET_MODES_TIEABLE_P): Redefine.
4214 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4215 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4216 (ia64_modes_tieable_p): New function.
4217 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4218 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4219 (iq2000_modes_tieable_p): New function.
4220 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4221 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4222 (lm32_modes_tieable_p): New function.
4223 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4224 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4225 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4226 (TARGET_MODES_TIEABLE_P): Redefine.
4227 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4228 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4229 (m32r_modes_tieable_p): New function.
4230 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4231 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4232 (m68k_modes_tieable_p): New function.
4233 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4234 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4235 (mcore_modes_tieable_p): New function.
4236 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4237 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4239 (TARGET_MODES_TIEABLE_P): Redefine.
4240 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4241 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4242 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4243 (TARGET_MODES_TIEABLE_P): Redefine.
4244 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4245 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4246 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4247 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4248 (mn10300_modes_tieable_p): ...this and make static.
4249 (TARGET_MODES_TIEABLE_P): Redefine.
4250 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4251 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4252 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4253 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4254 (msp430_modes_tieable_p): Make static.
4255 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4256 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4257 (TARGET_MODES_TIEABLE_P): Redefine.
4258 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4259 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4260 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4261 (TARGET_MODES_TIEABLE_P): Redefine.
4262 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4263 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4264 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4265 (TARGET_MODES_TIEABLE_P): Redefine.
4266 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4267 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4268 (pdp11_modes_tieable_p): New function.
4269 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4270 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4271 (rs6000_modes_tieable_p): New function.
4272 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4273 * config/powerpcspe/powerpcspe.md: Update comment.
4274 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4275 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4276 (TARGET_MODES_TIEABLE_P): Redefine.
4277 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4278 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4279 (rl78_modes_tieable_p): New function.
4280 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4281 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4282 (rs6000_modes_tieable_p): New function.
4283 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4284 * config/rs6000/rs6000.md: Update comment.
4285 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4286 * config/rx/rx.c (rx_modes_tieable_p): New function.
4287 (TARGET_MODES_TIEABLE_P): Redefine.
4288 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4289 * config/s390/s390.c (s390_modes_tieable_p): New function.
4290 (TARGET_MODES_TIEABLE_P): Redefine.
4291 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4292 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4293 (sh_modes_tieable_p): New function.
4294 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4295 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4296 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4297 (sparc_modes_tieable_p): Make static.
4298 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4299 * config/spu/spu.c (spu_modes_tieable_p): New function.
4300 (TARGET_MODES_TIEABLE_P): Redefine.
4301 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4302 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4303 (TARGET_MODES_TIEABLE_P): Redefine.
4304 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4305 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4306 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4307 * config/v850/v850.c (v850_modes_tieable_p): New function.
4308 (TARGET_MODES_TIEABLE_P): Redefine.
4309 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4310 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4311 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4312 (visium_modes_tieable_p): New function.
4313 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4314 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4315 (xtensa_modes_tieable_p): New function.
4316 * system.h (MODES_TIEABLE_P): Poison.
4318 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4319 Alan Hayward <alan.hayward@arm.com>
4320 David Sherwood <david.sherwood@arm.com>
4322 * target.def (hard_regno_mode_ok): New hook.
4323 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4324 (TARGET_HARD_REGNO_MODE_OK): ...this.
4325 * doc/tm.texi.in: Regenerate.
4326 * hooks.h (hook_bool_uint_mode_true): Declare.
4327 * hooks.c (hook_bool_uint_mode_true): New function.
4328 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4330 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4331 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4332 instead of HARD_REGNO_MODE_OK.
4333 * caller-save.c: Include target.h.
4334 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4336 * combine.c (can_combine_p): Likewise.
4337 (combinable_i3pat): Likewise.
4338 (can_change_dest_mode): Likewise.
4339 * expr.c (init_expr_target): Likewise.
4340 (convert_move): Likewise.
4341 (convert_modes): Likewise.
4342 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4343 (setup_prohibited_mode_move_regs): Likewise.
4344 * ira.h (target_ira): Likewise.
4345 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4346 * lra-constraints.c (process_alt_operands): Likewise.
4347 (split_reg): Likewise.
4348 * recog.c (peep2_find_free_register): Likewise.
4349 * ree.c (combine_reaching_defs): Likewise.
4350 * regcprop.c (maybe_mode_change): Likewise.
4351 * reginfo.c (init_reg_sets_1): Likewise.
4352 (choose_hard_reg_mode): Likewise.
4353 (simplifiable_subregs): Likewise.
4354 * regrename.c (check_new_reg_p): Likewise.
4355 * reload.c (find_valid_class): Likewise.
4356 (find_valid_class_1): Likewise.
4357 (reload_inner_reg_of_subreg): Likewise.
4358 (push_reload): Likewise.
4359 (combine_reloads): Likewise.
4360 (find_dummy_reload): Likewise.
4361 (find_reloads): Likewise.
4362 * reload1.c (find_reg): Likewise.
4363 (set_reload_reg): Likewise.
4364 (allocate_reload_reg): Likewise.
4365 (choose_reload_regs): Likewise.
4366 (reload_adjust_reg_for_temp): Likewise.
4367 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4368 (simplify_subreg_regno): Likewise.
4369 * sel-sched.c (init_regs_for_mode): Likewise.
4370 * varasm.c (make_decl_rtl): Likewise.
4371 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4372 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4374 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4375 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4376 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4377 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4378 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4379 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4380 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4381 (arc_mode_class): Delete.
4382 (HARD_REGNO_MODE_OK): Delete.
4383 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4384 (arc_hard_regno_mode_ok): Rename old array to...
4385 (arc_hard_regno_mode_ok_modes): ...this.
4386 (arc_conditional_register_usage): Update accordingly.
4387 (arc_mode_class): Make static.
4388 (arc_hard_regno_mode_ok): New function.
4389 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4390 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4391 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4392 (arm_hard_regno_mode_ok): Make static.
4393 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4395 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4396 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4397 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4399 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4400 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4401 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4402 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4403 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4405 * config/bfin/predicates.md (valid_reg_operand): Use
4406 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4407 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4408 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4409 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4410 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4411 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4412 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4413 (cr16_hard_regno_mode_ok): Make static and return a bool.
4414 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4415 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4416 (cris_hard_regno_mode_ok): New function.
4417 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4418 (epiphany_mode_class): Delete.
4419 (HARD_REGNO_MODE_OK): Delete.
4420 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4421 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4422 (hard_regno_mode_ok): Rename to...
4423 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4424 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4425 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4427 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4428 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4429 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4430 (frv_hard_regno_mode_ok): Make static and return a bool.
4431 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4433 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4434 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4435 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4436 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4438 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4439 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4440 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4441 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4444 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4445 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4446 (ia64_hard_regno_mode_ok): New function.
4447 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4448 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4449 (iq2000_hard_regno_mode_ok): New function.
4450 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4451 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4452 (lm32_hard_regno_mode_ok): New function.
4453 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4454 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4455 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4456 instead of HARD_REGNO_MODE_OK.
4457 (m32c_hard_regno_ok): Rename to...
4458 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4459 (m32c_cannot_change_mode_class): Update accordingly.
4460 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4461 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4462 (m32r_mode_class): Delete.
4463 (HARD_REGNO_MODE_OK): Delete.
4464 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4465 (m32r_hard_regno_mode_ok): Rename to...
4466 (m32r_hard_regno_modes): ...this.
4467 (m32r_mode_class): Make static.
4468 (m32r_hard_regno_mode_ok): New function.
4469 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4470 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4471 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4472 (m68k_hard_regno_mode_ok): Make static.
4473 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4474 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4475 (mcore_hard_regno_mode_ok): New function.
4476 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4477 (HARD_REGNO_MODE_OK): Delete.
4478 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4480 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4481 (microblaze_hard_regno_mode_ok): New function.
4482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4483 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4484 (mips_hard_regno_mode_ok): Delete.
4485 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4486 (mips_hard_regno_mode_ok_p): ...this and make static.
4487 (mips_hard_regno_mode_ok_p): Rename to...
4488 (mips_hard_regno_mode_ok_uncached): ...this.
4489 (mips_hard_regno_mode_ok): New function.
4490 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4491 of HARD_REGNO_MODE_OK.
4492 (mips_option_override): Update after above name changes.
4493 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4494 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4495 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4496 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4497 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4498 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4499 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4500 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4501 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4502 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4503 (msp430_hard_regno_mode_ok): Make static and return a bool.
4504 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4505 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4506 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4508 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4509 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4510 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4511 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4512 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4513 (PA_HARD_REGNO_MODE_OK): ...this
4514 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4515 (PA_HARD_REGNO_MODE_OK): ...this.
4516 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4517 (pa_hard_regno_mode_ok): New function.
4518 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4519 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4520 (pdp11_hard_regno_mode_ok): New function.
4521 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4522 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4524 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4526 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4527 (rs6000_hard_regno_mode_ok): Rename to...
4528 (rs6000_hard_regno_mode_ok_uncached): ...this.
4529 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4530 (rs6000_hard_regno_mode_ok): New function.
4531 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4532 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4533 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4534 (riscv_hard_regno_mode_ok): ...this and make static.
4535 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4536 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4537 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4538 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4539 (rl78_hard_regno_mode_ok): Make static and return bool.
4540 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4541 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4543 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4544 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4545 (rs6000_hard_regno_mode_ok): Rename to...
4546 (rs6000_hard_regno_mode_ok_uncached): ...this.
4547 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4548 (rs6000_hard_regno_mode_ok): New function.
4549 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4550 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4551 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4552 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4553 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4554 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4555 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4556 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4557 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4558 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4559 (sh_hard_regno_mode_ok): Make static.
4560 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4561 instead of HARD_REGNO_MODE_OK.
4562 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4563 (sparc_mode_class): Delete.
4564 (HARD_REGNO_MODE_OK): Delete.
4565 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4566 (hard_regno_mode_classes): Make static.
4567 (sparc_mode_class): Likewise.
4568 (sparc_hard_regno_mode_ok): New function.
4569 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4570 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4571 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4574 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4575 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4576 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4577 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4578 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4579 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4580 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4581 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4582 (visium_hard_regno_mode_ok): New function.
4583 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4584 instead of HARD_REGNO_MODE_OK.
4585 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4586 (HARD_REGNO_MODE_OK): Delete.
4587 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4588 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4589 (xtensa_option_override): Update accordingly.
4590 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4591 (xtensa_hard_regno_mode_ok): New function.
4592 * system.h (HARD_REGNO_MODE_OK): Poison.
4594 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4595 Alan Hayward <alan.hayward@arm.com>
4596 David Sherwood <david.sherwood@arm.com>
4598 * target.def (hard_regno_call_part_clobbered): New hook.
4599 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4600 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4601 * doc/tm.texi: Regenerate.
4602 * hooks.h (hook_bool_uint_mode_false): Declare.
4603 * hooks.c (hook_bool_uint_mode_false): New function.
4604 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4605 * cselib.c (cselib_process_insn): Use
4606 targetm.hard_regno_call_part_clobbered instead of
4607 HARD_REGNO_CALL_PART_CLOBBERED.
4608 * ira-conflicts.c (ira_build_conflicts): Likewise.
4609 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4610 * lra-constraints.c (need_for_call_save_p): Likewise.
4611 * lra-lives.c: Include target.h.
4612 (check_pseudos_live_through_calls): Use
4613 targetm.hard_regno_call_part_clobbered instead of
4614 HARD_REGNO_CALL_PART_CLOBBERED.
4615 * regcprop.c: Include target.h.
4616 (copyprop_hardreg_forward_1): Use
4617 targetm.hard_regno_call_part_clobbered instead of
4618 HARD_REGNO_CALL_PART_CLOBBERED.
4619 * reginfo.c (choose_hard_reg_mode): Likewise.
4620 * regrename.c (check_new_reg_p): Likewise.
4621 * reload.c (find_equiv_reg): Likewise.
4622 * reload1.c (emit_reload_insns): Likewise.
4623 * sched-deps.c (deps_analyze_insn): Likewise.
4624 * sel-sched.c (init_regs_for_mode): Likewise.
4625 (mark_unavailable_hard_regs): Likewise.
4626 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4627 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4628 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4630 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4631 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4632 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4634 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4636 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4637 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4638 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4640 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4641 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4642 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4644 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4645 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4647 * config/powerpcspe/powerpcspe.c
4648 (rs6000_hard_regno_call_part_clobbered): New function.
4649 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4650 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4651 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4653 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4654 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4655 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4657 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4658 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4659 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4661 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4662 Alan Hayward <alan.hayward@arm.com>
4663 David Sherwood <david.sherwood@arm.com>
4665 * rtl.h (subreg_memory_offset): Declare.
4666 * emit-rtl.c (subreg_memory_offset): New function.
4667 * expmed.c (store_bit_field_1): Use it.
4668 * expr.c (undefined_operand_subword_p): Likewise.
4669 * simplify-rtx.c (simplify_subreg): Likewise.
4671 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4673 PR rtl-optimization/57448
4676 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4677 using atomic_load pattern.
4678 (expand_atomic_store): Likewise.
4680 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4683 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4684 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4685 handling. Use replace_call_with_value with NULL instead of
4686 gsi_replace, unlink_stmt_vdef and release_defs.
4688 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4691 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4693 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4696 * rtl.h (emit_library_call_value_1): Declare.
4697 (emit_library_call): Replace declaration with a series of overloads.
4698 Remove the parameter count argument.
4699 (emit_library_call_value): Likewise.
4700 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4701 with an "rtx_mode_t *".
4702 (emit_library_call_value): Delete.
4703 (emit_library_call): Likewise.
4704 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4705 (asan_emit_allocas_unpoison): Likewise.
4706 * builtins.c (expand_builtin_powi): Likewise.
4707 (expand_asan_emit_allocas_unpoison): Likewise.
4708 * cfgexpand.c (expand_main_function): Likewise.
4709 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4710 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4711 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4712 * config/arm/arm.c (arm_trampoline_init): Likewise.
4713 (arm_call_tls_get_addr): Likewise.
4714 (arm_expand_divmod_libfunc): Likewise.
4715 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4716 (smulsi3_highpart): Likewise.
4717 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4718 (c6x_expand_compare): Likewise.
4719 (c6x_expand_movmem): Likewise.
4720 * config/frv/frv.c (frv_trampoline_init): Likewise.
4721 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4722 (ix86_expand_divmod_libfunc): Likewise.
4723 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4724 (ia64_expand_compare): Likewise.
4725 (ia64_profile_hook): Likewise.
4726 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4727 (nonlocal_goto): Likewise.
4728 (restore_stack_nonlocal): Likewise.
4729 * config/m32r/m32r.c (block_move_call): Likewise.
4730 (m32r_trampoline_init): Likewise.
4731 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4732 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4733 (m68k_call_m68k_read_tp): Likewise.
4734 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4735 (microblaze_expand_divide): Likewise.
4736 * config/mips/mips.h (mips_args): Likewise.
4737 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4738 (MIPS_ICACHE_SYNC): Likewise.
4739 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4740 (nios2_trampoline_init): Likewise.
4741 * config/pa/pa.c (hppa_tls_call): Likewise.
4742 (pa_trampoline_init): Likewise.
4743 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4744 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4745 (expand_strn_compare): Likewise.
4746 (rs6000_generate_compare): Likewise.
4747 (rs6000_expand_float128_convert): Likewise.
4748 (output_profile_hook): Likewise.
4749 (rs6000_trampoline_init): Likewise.
4750 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4751 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4752 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4753 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4754 (rs6000_generate_compare): Likewise.
4755 (rs6000_expand_float128_convert): Likewise.
4756 (output_profile_hook): Likewise.
4757 (rs6000_trampoline_init): Likewise.
4758 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4759 * config/sh/sh.c (sh_trampoline_init): Likewise.
4760 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4761 (sparc_emit_float_lib_cmp): Likewise.
4762 (sparc32_initialize_trampoline): Likewise.
4763 (sparc64_initialize_trampoline): Likewise.
4764 (sparc_profile_hook): Likewise.
4765 * config/spu/spu.c (ea_load_store): Likewise.
4766 * config/spu/spu.md (floatunssidf2): Likewise.
4767 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4768 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4769 * config/visium/visium.c (expand_block_move_4): Likewise.
4770 (expand_block_move_2): Likewise.
4771 (expand_block_move_1): Likewise.
4772 (expand_block_set_4): Likewise.
4773 (expand_block_set_2): Likewise.
4774 (expand_block_set_1): Likewise.
4775 (visium_trampoline_init): Likewise.
4776 (visium_profile_hook): Likewise.
4777 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4778 (xtensa_setup_frame_addresses): Likewise.
4779 (xtensa_trampoline_init): Likewise.
4780 * except.c (sjlj_emit_function_enter): Likewise.
4781 (sjlj_emit_function_exit): Likewise.
4782 * explow.c (allocate_dynamic_stack_space): Likewise.
4783 (probe_stack_range): Likewise.
4784 * expr.c (convert_mode_scalar): Likewise.
4785 * optabs.c (expand_binop): Likewise.
4786 (expand_twoval_binop_libfunc): Likewise.
4787 (expand_unop): Likewise.
4788 (prepare_cmp_insn): Likewise.
4789 (prepare_float_lib_cmp): Likewise.
4790 (expand_float): Likewise.
4791 (expand_fix): Likewise.
4792 (expand_fixed_convert): Likewise.
4793 (maybe_emit_sync_lock_test_and_set): Likewise.
4794 (expand_atomic_compare_and_swap): Likewise.
4795 (expand_mem_thread_fence): Likewise.
4796 (expand_atomic_fetch_op): Likewise.
4798 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4800 * doc/generic.texi (OpenACC): Adjust URL.
4801 * doc/invoke.texi (C Dialect Options): Ditto.
4803 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4805 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4806 predicate for operand 1. Add (m,<S>) constraint.
4807 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4808 Prevent memory operand 1 with register operand 2.
4810 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4812 PR rtl-optimization/82024
4813 * combine.c (try_combine): If the combination result is a PARALLEL,
4814 and we only need to retain the SET in there that would be placed
4815 at I2, check that we can place that at I3 instead, before doing so.
4817 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4820 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4822 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4823 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4826 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4827 Jeff Law <law@redhat.com>
4829 * varasm.c (bss_initializer_p): Do not put constants into .bss
4830 (categorize_decl_for_section): Handle bss_initializer_p returning
4831 false when DECL_INITIAL is NULL.
4833 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4836 * config/s390/s390.c (s390_can_inline_p): New function.
4838 2017-09-01 Jeff Law <law@redhat.com>
4840 PR tree-optimization/82052
4841 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4842 Always initialize the returned slot after a hash table miss
4843 when INSERT is true.
4845 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4847 * config/s390/s390.md (mem_signal_fence): Remove.
4848 * doc/md.texi (mem_signal_fence): Remove.
4849 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4851 * target-insns.def (mem_signal_fence): Remove.
4853 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4856 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4859 * asan.c (create_odr_indicator): Strip name encoding from assembler
4860 name before appending it after __odr_asan_.
4862 2017-09-01 Martin Liska <mliska@suse.cz>
4864 PR tree-optimization/82059
4865 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4866 frequency only when an edge is redirected.
4868 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4870 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4871 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4872 (arc_conditional_register_usage): Remove ARC600 lp_count
4874 (arc_file_start): Emit Tag_ARC_CPU_variation.
4875 (arc_can_use_doloop_p): New conditions to use ZOLs.
4876 (hwloop_fail): New function.
4877 (hwloop_optimize): Likewise.
4878 (hwloop_pattern_reg): Likewise.
4879 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4880 (arc_reorg_loops): New function, calls reorg_loops.
4881 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4882 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4884 (arc_loop_hazard): Remove function, functionality moved into
4886 (arc_hazard): Remove arc_loop_hazard call.
4887 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4888 into hwloop_optimize.
4889 (arc_label_align): Remove ZOL handling.
4890 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4891 * config/arc/arc.md (doloop_begin): Remove pattern.
4892 (doloop_begin_i): Likewise.
4893 (doloop_end_i): Likewise.
4894 (doloop_fallback): Likewise.
4895 (doloop_fallback_m): Likewise.
4896 (doloop_end): Reimplement expand.
4897 (arc_lp): New pattern for LP instruction.
4898 (loop_end): New pattern.
4899 (loop_fail): Likewise.
4900 (decrement_and_branch_until_zero): Likewise.
4901 * config/arc/arc.opt (mlpc-width): New option.
4902 * doc/invoke.texi (mlpc-width): Document option.
4904 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4906 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4907 (arc_ccfsm_advance): Fix checking for delay slots.
4908 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4910 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4912 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4914 (movhi_insn): Update store instruction constraint which are saving
4915 6-bit short immediates.
4916 (movsi_insn): Consider also short scaled load operations.
4917 (zero_extendhisi2_i): Use Usd constraint instead of T.
4918 (extendhisi2_i): Add q constraint.
4919 (arc_clzsi2): Add type and length attributes.
4920 (arc_ctzsi2): Likewise.
4921 * config/arc/constraints.md (Usc): Update constraint, the
4922 assembler can parse two relocations for a single instruction.
4924 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4926 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4927 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4929 2017-08-31 Olivier Hainque <hainque@adacore.com>
4931 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4932 match as powerpc-wrs-vxworks*.
4934 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4936 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4937 register constraint for by-element operand.
4938 (aarch64_mls_elt_merge<mode>): Likewise.
4940 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4942 * config/arc/arc.c (arc_can_follow_jump): Check for short
4945 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4947 * config.gcc: Use g.opt for arc.
4948 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4949 functionality moved to ...
4950 (legitimate_scaled_address_p): New function, ...here.
4951 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4952 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4953 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4955 (arc_override_options): Handle G option.
4956 (arc_output_pic_addr_const): Correct function definition.
4957 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4958 (arc_decl_anon_ns_mem_p): Delete.
4959 (arc_in_small_data_p): Overhaul this function to take into
4960 consideration the value given via G option.
4961 (arc_rewrite_small_data_1): Renamed and corrected old
4962 arc_rewrite_small_data function.
4963 (arc_rewrite_small_data): New function.
4964 (small_data_pattern): Don't use pic_offset_table_rtx.
4965 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4966 * config/arc/simdext.md (movmisalignv2hi): Use
4967 prepare_move_operands function.
4969 (movmisalign*): Likewise.
4970 * doc/invoke.texi (ARC options): Document -G option.
4972 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4974 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4976 * config/arc/arc.c (arc_print_operand): Output scalled address for
4977 sdata whenever is possible.
4978 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4979 load/stores are available.
4980 (compact_sda_memory_operand): Check for the alignment required by
4981 code density instructions.
4982 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4984 * config/arc/constraints.md (Usd): Update constraint.
4985 (Us0): New constraint.
4986 (Usc): Update constraint.
4988 2017-08-31 Richard Biener <rguenther@suse.de>
4991 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4994 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4996 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4997 Resize type_signature.
4999 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5000 Alan Hayward <alan.hayward@arm.com>
5001 David Sherwood <david.sherwood@arm.com>
5003 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5004 subregs whose inner modes can be stored in GPRs.
5005 (aarch64_classify_index): Likewise.
5007 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5008 Alan Hayward <alan.hayward@arm.com>
5009 David Sherwood <david.sherwood@arm.com>
5011 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5012 (V_INT_EQUIV): ...this.
5013 (v_cmp_result): Rename to...
5014 (v_int_equiv): ...this.
5015 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5016 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5017 (copysign<mode>3): Likewise.
5018 (aarch64_simd_bsl<mode>_internal): Likewise.
5019 (aarch64_simd_bsl<mode>): Likewise.
5020 (vec_cmp<mode><mode>): Likewise.
5021 (vcond<mode><mode>): Likewise.
5022 (vcond<v_cmp_mixed><mode>): Likewise.
5023 (vcondu<mode><v_cmp_mixed>): Likewise.
5024 (aarch64_cm<optab><mode>): Likewise.
5025 (aarch64_cmtst<mode>): Likewise.
5026 (aarch64_fac<optab><mode>): Likewise.
5027 (vec_perm_const<mode>): Likewise.
5028 (vcond_mask_<mode><v_cmp_result>): Rename to...
5029 (vcond_mask_<mode><v_int_equiv>): ...this.
5030 (vec_cmp<mode><v_cmp_result>): Rename to...
5031 (vec_cmp<mode><v_int_equiv>): ...this.
5033 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5034 Alan Hayward <alan.hayward@arm.com>
5035 David Sherwood <david.sherwood@arm.com>
5037 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5039 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5040 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5041 (UNSPEC_LD4_DREG): New unspecs.
5042 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5043 (aarch64_ld2<mode>_dreg_be): Replace with...
5044 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5046 (aarch64_ld3<mode>_dreg_le)
5047 (aarch64_ld3<mode>_dreg_be): Replace with...
5048 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5050 (aarch64_ld4<mode>_dreg_le)
5051 (aarch64_ld4<mode>_dreg_be): Replace with...
5052 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5055 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5057 PR tree-optimization/81987
5058 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5059 insert an initializer in a location not dominated by the stride
5062 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5064 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5065 on the entire header of the finally block in the fallthru case.
5067 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5069 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5071 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5073 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5074 rs6000_emit_move_from_cr and call renamed function.
5075 (rs6000_emit_prologue): Call renamed functions.
5076 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5077 movesi_from_cr, remove volatile CRs.
5079 2017-08-30 Jon Beniston <jon@beniston.com>
5080 Richard Biener <rguenther@suse.de>
5082 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5083 of VECTOR_MODE_P check.
5084 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5085 element vector types.
5087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5089 * df.h (df_read_modify_subreg_p): Remove in favor of...
5090 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5091 const_rtx instead of an rtx.
5092 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5093 * df-problems.c (df_word_lr_mark_ref): Likewise.
5094 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5095 (mark_pseudo_reg_dead): Likewise.
5096 (mark_ref_dead): Likewise.
5097 * reginfo.c (init_subregs_of_mode): Likewise.
5098 * sched-deps.c (sched_analyze_1): Likewise.
5099 * df-scan.c (df_def_record_1): Likewise.
5100 (df_uses_record): Likewise.
5101 (df_read_modify_subreg_p): Remove in favor of...
5102 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5103 const_rtx instead of an rtx.
5105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5106 Alan Hayward <alan.hayward@arm.com>
5107 David Sherwood <david.sherwood@arm.com>
5109 * rtl.h (partial_subreg_p): New function.
5110 * caller-save.c (save_call_clobbered_regs): Use it.
5111 * calls.c (expand_call): Likewise.
5112 * combine.c (combinable_i3pat): Likewise.
5113 (simplify_set): Likewise.
5114 (make_extraction): Likewise.
5115 (make_compound_operation_int): Likewise.
5116 (gen_lowpart_or_truncate): Likewise.
5117 (force_to_mode): Likewise.
5118 (make_field_assignment): Likewise.
5119 (reg_truncated_to_mode): Likewise.
5120 (record_truncated_value): Likewise.
5121 (move_deaths): Likewise.
5122 * cse.c (record_jump_cond): Likewise.
5123 (cse_insn): Likewise.
5124 * cselib.c (cselib_lookup_1): Likewise.
5125 * expmed.c (extract_bit_field_using_extv): Likewise.
5126 * function.c (assign_parm_setup_reg): Likewise.
5127 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5128 * ira-build.c (create_insn_allocnos): Likewise.
5129 * lra-coalesce.c (merge_pseudos): Likewise.
5130 * lra-constraints.c (match_reload): Likewise.
5131 (simplify_operand_subreg): Likewise.
5132 (curr_insn_transform): Likewise.
5133 * lra-lives.c (process_bb_lives): Likewise.
5134 * lra.c (new_insn_reg): Likewise.
5135 (lra_substitute_pseudo): Likewise.
5136 * regcprop.c (mode_change_ok): Likewise.
5137 (maybe_mode_change): Likewise.
5138 (copyprop_hardreg_forward_1): Likewise.
5139 * reload.c (push_reload): Likewise.
5140 (find_reloads): Likewise.
5141 (find_reloads_subreg_address): Likewise.
5142 * reload1.c (alter_reg): Likewise.
5143 (eliminate_regs_1): Likewise.
5144 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5146 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5148 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5149 back to if statements, including unpack.
5151 2017-08-30 Martin Liska <mliska@suse.cz>
5154 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5156 (func_checker::compare_asm_inputs_outputs): ... this function.
5157 (func_checker::compare_gimple_asm): Use the function to compare
5158 also ASM constrains.
5159 * ipa-icf-gimple.h: Rename the function.
5161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5162 Alan Hayward <alan.hayward@arm.com>
5163 David Sherwood <david.sherwood@arm.com>
5165 * coretypes.h (complex_mode): New type.
5166 * gdbhooks.py (build_pretty_printer): Handle it.
5167 * machmode.h (complex_mode): New class.
5168 (complex_mode::includes_p): New function.
5169 (is_complex_int_mode): Likewise.
5170 (is_complex_float_mode): Likewise.
5171 * genmodes.c (get_mode_class): Handle complex mode classes.
5172 * function.c (expand_function_end): Use is_complex_int_mode.
5174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5175 Alan Hayward <alan.hayward@arm.com>
5176 David Sherwood <david.sherwood@arm.com>
5178 * coretypes.h (scalar_mode_pod): New typedef.
5179 * gdbhooks.py (build_pretty_printer): Handle it.
5180 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5181 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5182 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5183 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5187 Alan Hayward <alan.hayward@arm.com>
5188 David Sherwood <david.sherwood@arm.com>
5190 * machmode.h (mode_for_vector): Take a scalar_mode instead
5192 * stor-layout.c (mode_for_vector): Likewise.
5193 * explow.c (promote_mode): Use as_a <scalar_mode>.
5194 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5197 Alan Hayward <alan.hayward@arm.com>
5198 David Sherwood <david.sherwood@arm.com>
5200 * target.def (preferred_simd_mode): Take a scalar_mode
5201 instead of a machine_mode.
5202 * targhooks.h (default_preferred_simd_mode): Likewise.
5203 * targhooks.c (default_preferred_simd_mode): Likewise.
5204 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5205 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5206 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5207 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5208 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5209 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5210 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5211 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5213 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5214 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5215 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5216 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5217 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5218 * doc/tm.texi: Regenerate.
5219 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5223 Alan Hayward <alan.hayward@arm.com>
5224 David Sherwood <david.sherwood@arm.com>
5226 * target.def (scalar_mode_supported_p): Take a scalar_mode
5227 instead of a machine_mode.
5228 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5229 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5230 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5231 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5232 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5233 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5234 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5235 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5236 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5237 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5238 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5239 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5240 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5241 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5243 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5244 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5245 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5246 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5247 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5249 * doc/tm.texi: Regenerate.
5251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5252 Alan Hayward <alan.hayward@arm.com>
5253 David Sherwood <david.sherwood@arm.com>
5255 * coretypes.h (opt_scalar_mode): New typedef.
5256 * gdbhooks.py (build_pretty_printers): Handle it.
5257 * machmode.h (mode_iterator::get_2xwider): Add overload for
5259 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5261 * expr.c (convert_mode_scalar): Likewise.
5262 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5263 * optabs.c (expand_float): Likewise.
5264 (expand_fix): Likewise.
5265 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5268 Alan Hayward <alan.hayward@arm.com>
5269 David Sherwood <david.sherwood@arm.com>
5271 * optabs.c (expand_float): Explicitly check for scalars before
5272 using a branching expansion.
5273 (expand_fix): Likewise.
5275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5276 Alan Hayward <alan.hayward@arm.com>
5277 David Sherwood <david.sherwood@arm.com>
5279 * expr.c (convert_mode): Split scalar handling out into...
5280 (convert_mode_scalar): ...this new function. Treat the modes
5283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayward <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5289 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5296 rather than a machine_mode.
5297 (fixed_from_string): Likewise.
5298 (fixed_convert): Likewise.
5299 (fixed_convert_from_int): Likewise.
5300 (fixed_convert_from_real): Likewise.
5301 (real_convert_from_fixed): Likewise.
5302 * fixed-value.c (fixed_from_double_int): Likewise.
5303 (fixed_from_string): Likewise.
5304 (fixed_convert): Likewise.
5305 (fixed_convert_from_int): Likewise.
5306 (fixed_convert_from_real): Likewise.
5307 (real_convert_from_fixed): Likewise.
5308 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5311 Alan Hayward <alan.hayward@arm.com>
5312 David Sherwood <david.sherwood@arm.com>
5314 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5315 of separate mode class checks. Do not allow vector modes here.
5316 (immed_wide_int_const): Use as_a <scalar_mode>.
5317 * explow.c (trunc_int_for_mode): Likewise.
5318 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5319 (wi::shwi): Likewise.
5320 (wi::min_value): Likewise.
5321 (wi::max_value): Likewise.
5322 * dwarf2out.c (loc_descriptor): Likewise.
5323 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5327 Alan Hayward <alan.hayward@arm.com>
5328 David Sherwood <david.sherwood@arm.com>
5330 * tree.h (SCALAR_TYPE_MODE): New macro.
5331 * expr.c (expand_expr_addr_expr_1): Use it.
5332 (expand_expr_real_2): Likewise.
5333 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5334 (fold_convert_const_fixed_from_int): Likewise.
5335 (fold_convert_const_fixed_from_real): Likewise.
5336 (native_encode_fixed): Likewise
5337 (native_encode_complex): Likewise
5338 (native_encode_vector): Likewise.
5339 (native_interpret_fixed): Likewise.
5340 (native_interpret_real): Likewise.
5341 (native_interpret_complex): Likewise.
5342 (native_interpret_vector): Likewise.
5343 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5344 (simd_clone_adjust_argument_types): Likewise.
5345 (simd_clone_init_simd_arrays): Likewise.
5346 (simd_clone_adjust): Likewise.
5347 * stor-layout.c (layout_type): Likewise.
5348 * tree.c (build_minus_one_cst): Likewise.
5349 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5350 * tree-inline.c (estimate_move_cost): Likewise.
5351 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5352 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5353 (vectorizable_reduction): Likewise.
5354 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5355 (vect_recog_mixed_size_cond_pattern): Likewise.
5356 (check_bool_pattern): Likewise.
5357 (adjust_bool_pattern): Likewise.
5358 (search_type_for_mask_1): Likewise.
5359 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5360 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5361 (vectorizable_load): Likewise.
5362 (vectorizable_store): Likewise.
5363 * ubsan.c (ubsan_encode_value): Likewise.
5364 * varasm.c (output_constant): Likewise.
5366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5367 Alan Hayward <alan.hayward@arm.com>
5368 David Sherwood <david.sherwood@arm.com>
5370 * coretypes.h (scalar_mode): New class.
5371 * machmode.h (scalar_mode): Likewise.
5372 (scalar_mode::includes_p): New function.
5373 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5374 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5375 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5376 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5377 * expmed.c (store_bit_field_1): Likewise.
5378 (extract_bit_field_1): Likewise.
5379 * expr.c (write_complex_part): Likewise.
5380 (read_complex_part): Likewise.
5381 (emit_move_complex_push): Likewise.
5382 (expand_expr_real_2): Likewise.
5383 * function.c (assign_parm_setup_reg): Likewise.
5384 (assign_parms_unsplit_complex): Likewise.
5385 * optabs.c (expand_binop): Likewise.
5386 * rtlanal.c (subreg_get_info): Likewise.
5387 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5388 * varasm.c (output_constant_pool_2): Likewise.
5390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5391 Alan Hayward <alan.hayward@arm.com>
5392 David Sherwood <david.sherwood@arm.com>
5394 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5396 (expmed_mult_highpart_optab): Likewise.
5397 (expmed_mult_highpart): Likewise.
5399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5400 Alan Hayward <alan.hayward@arm.com>
5401 David Sherwood <david.sherwood@arm.com>
5403 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5404 instead of a machine_mode.
5405 (builtin_memset_read_str): Likewise.
5406 * builtins.c (c_readstr): Likewise.
5407 (builtin_memcpy_read_str): Likewise.
5408 (builtin_strncpy_read_str): Likewise.
5409 (builtin_memset_read_str): Likewise.
5410 (builtin_memset_gen_str): Likewise.
5411 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5412 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5413 instead of a machine_mode.
5414 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5416 (make_extraction): Likewise.
5417 (try_widen_shift_mode): Take and return scalar_int_modes instead
5419 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5420 a scalar_int_mode instead of a machine_mode.
5421 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5422 (avr_addr_space_pointer_mode): Likewise.
5423 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5424 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5425 (msp430_unwind_word_mode): Likewise.
5426 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5427 (spu_addr_space_pointer_mode): Likewise.
5428 (spu_addr_space_address_mode): Likewise.
5429 (spu_libgcc_cmp_return_mode): Likewise.
5430 (spu_libgcc_shift_count_mode): Likewise.
5431 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5432 (rl78_addr_space_pointer_mode): Likewise.
5433 (fl78_unwind_word_mode): Likewise.
5434 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5436 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5437 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5438 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5439 (mips_valid_pointer_mode): Likewise.
5440 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5441 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5442 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5444 (ft32_addr_space_address_mode): Likewise.
5445 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5446 scalar_int_mode instead of a machine_mode.
5447 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5449 (m32c_addr_space_address_mode): Likewise.
5450 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5451 (rs6000_eh_return_filter_mode): Likewise.
5452 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5453 (rs6000_eh_return_filter_mode): Likewise.
5454 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5455 (s390_libgcc_shift_count_mode): Likewise.
5456 (s390_unwind_word_mode): Likewise.
5457 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5459 * target.def (mode_rep_extended): Likewise.
5460 (valid_pointer_mode): Likewise.
5461 (addr_space.valid_pointer_mode): Likewise.
5462 (eh_return_filter_mode): Return a scalar_int_mode rather than
5464 (libgcc_cmp_return_mode): Likewise.
5465 (libgcc_shift_count_mode): Likewise.
5466 (unwind_word_mode): Likewise.
5467 (addr_space.pointer_mode): Likewise.
5468 (addr_space.address_mode): Likewise.
5469 * doc/tm.texi: Regenerate.
5470 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5472 (do_jump): Use scalar_int_mode for local variables.
5473 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5474 rather than a machine_mode.
5475 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5476 (scompare_loc_descriptor_wide): Likewise.
5477 (scompare_loc_descriptor_narrow): Likewise.
5478 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5480 * except.c (sjlj_emit_dispatch_table): Likewise.
5481 (expand_builtin_eh_copy_values): Likewise.
5482 * explow.c (convert_memory_address_addr_space_1): Likewise.
5483 Take a scalar_int_mode rather than a machine_mode.
5484 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5485 than a machine_mode.
5486 (memory_address_addr_space): Use scalar_int_mode for local variables.
5487 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5488 rather than a machine_mode.
5489 * expmed.c (mask_rtx): Likewise.
5490 (init_expmed_one_conv): Likewise.
5491 (expand_mult_highpart_adjust): Likewise.
5492 (extract_high_half): Likewise.
5493 (expmed_mult_highpart_optab): Likewise.
5494 (expmed_mult_highpart): Likewise.
5495 (expand_smod_pow2): Likewise.
5496 (expand_sdiv_pow2): Likewise.
5497 (emit_store_flag_int): Likewise.
5498 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5500 (extract_low_bits): Likewise.
5501 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5503 * expr.c (pieces_addr::adjust): Likewise.
5504 (can_store_by_pieces): Likewise.
5505 (store_by_pieces): Likewise.
5506 (clear_by_pieces_1): Likewise.
5507 (expand_expr_addr_expr_1): Likewise.
5508 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5509 (expand_expr_real_1): Likewise.
5510 (try_casesi): Likewise.
5511 * final.c (shorten_branches): Likewise.
5512 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5513 type of "mode" to machine_mode.
5514 * internal-fn.c (expand_arith_overflow_result_store): Take a
5515 scalar_int_mode rather than a machine_mode.
5516 (expand_mul_overflow): Use scalar_int_mode for local variables.
5517 * loop-doloop.c (doloop_modify): Likewise.
5518 (doloop_optimize): Likewise.
5519 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5520 than a machine_mode.
5521 (expand_doubleword_shift_condmove): Likewise.
5522 (expand_doubleword_shift): Likewise.
5523 (expand_doubleword_clz): Likewise.
5524 (expand_doubleword_popcount): Likewise.
5525 (expand_doubleword_parity): Likewise.
5526 (expand_absneg_bit): Use scalar_int_mode for local variables.
5527 (prepare_float_lib_cmp): Likewise.
5528 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5529 rather than a machine_mode.
5530 (convert_memory_address_addr_space): Likewise.
5531 (get_mode_bounds): Likewise.
5532 (get_address_mode): Return a scalar_int_mode rather than a
5534 * rtlanal.c (get_address_mode): Likewise.
5535 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5536 than a machine_mode.
5537 * targhooks.c (default_mode_rep_extended): Likewise.
5538 (default_valid_pointer_mode): Likewise.
5539 (default_addr_space_valid_pointer_mode): Likewise.
5540 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5541 than a machine_mode.
5542 (default_libgcc_cmp_return_mode): Likewise.
5543 (default_libgcc_shift_count_mode): Likewise.
5544 (default_unwind_word_mode): Likewise.
5545 (default_addr_space_pointer_mode): Likewise.
5546 (default_addr_space_address_mode): Likewise.
5547 * targhooks.h (default_eh_return_filter_mode): Likewise.
5548 (default_libgcc_cmp_return_mode): Likewise.
5549 (default_libgcc_shift_count_mode): Likewise.
5550 (default_unwind_word_mode): Likewise.
5551 (default_addr_space_pointer_mode): Likewise.
5552 (default_addr_space_address_mode): Likewise.
5553 (default_mode_rep_extended): Take a scalar_int_mode rather than
5555 (default_valid_pointer_mode): Likewise.
5556 (default_addr_space_valid_pointer_mode): Likewise.
5557 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5559 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5560 rather than a machine_mode.
5561 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5562 for local variables.
5563 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5564 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5565 than a machine_mode.
5567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5568 Alan Hayward <alan.hayward@arm.com>
5569 David Sherwood <david.sherwood@arm.com>
5571 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5572 the mode argument to scalar_int_mode.
5573 (do_jump_by_parts_zero_rtx): Likewise.
5574 (do_jump_by_parts_equality_rtx): Likewise.
5575 (do_jump_by_parts_greater): Take a mode argument.
5576 (do_jump_by_parts_equality): Likewise.
5577 (do_jump_1): Update calls accordingly.
5579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5580 Alan Hayward <alan.hayward@arm.com>
5581 David Sherwood <david.sherwood@arm.com>
5583 * is-a.h (safe_dyn_cast): New function.
5584 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5585 (jump_table_for_label): Likewise.
5586 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5587 instead of an rtx_insn *.
5588 (shorten_branches): Use dyn_cast instead of LABEL_P and
5589 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5590 rtx_jump_table_data::get_data_mode.
5591 (final_scan_insn): Likewise.
5593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5594 Alan Hayward <alan.hayward@arm.com>
5595 David Sherwood <david.sherwood@arm.com>
5597 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5598 trying to combine a full-register integer set with a subreg
5601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5602 Alan Hayward <alan.hayward@arm.com>
5603 David Sherwood <david.sherwood@arm.com>
5605 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5606 that is always either address_mode or pointer_mode.
5608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5609 Alan Hayward <alan.hayward@arm.com>
5610 David Sherwood <david.sherwood@arm.com>
5612 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5613 when the two are known to be equal.
5615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5616 Alan Hayward <alan.hayward@arm.com>
5617 David Sherwood <david.sherwood@arm.com>
5619 * simplify-rtx.c (simplify_const_unary_operation): Use
5620 is_a <scalar_int_mode> instead of checking for a nonzero
5621 precision. Forcibly convert op_mode to a scalar_int_mode
5622 in that case. More clearly differentiate the operand and
5623 result modes and use the former when deciding what the value
5624 of a count-bits operation should be. Use is_int_mode instead
5625 of checking for a MODE_INT. Remove redundant check for whether
5626 this mode has a zero precision.
5628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5629 Alan Hayward <alan.hayward@arm.com>
5630 David Sherwood <david.sherwood@arm.com>
5632 * optabs.c (widen_leading): Change the type of the mode argument
5633 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5634 (widen_bswap): Likewise.
5635 (expand_parity): Likewise.
5636 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5637 (expand_ffs): Likewise.
5638 (epand_unop): Check for scalar integer modes before calling the
5641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5642 Alan Hayward <alan.hayward@arm.com>
5643 David Sherwood <david.sherwood@arm.com>
5645 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5647 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5654 value_mode arguments. Use scalar_int_mode internally.
5655 (store_bit_field_1): Rename the new integer mode from imode
5656 to op0_mode and use it instead of GET_MODE (op0). Update calls
5657 to store_split_bit_field, store_bit_field_using_insv and
5658 store_fixed_bit_field.
5659 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5660 Use scalar_int_mode internally. Use a bit count rather than a mode
5661 when calculating the largest bit size for get_best_mode.
5662 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5663 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5664 Remove assertion that OP0 has a scalar integer mode.
5665 (store_split_bit_field): Add op0_mode and value_mode arguments.
5666 Update calls to extract_fixed_bit_field.
5667 (extract_bit_field_using_extv): Add an op0_mode argument.
5668 Use scalar_int_mode internally.
5669 (extract_bit_field_1): Rename the new integer mode from imode to
5670 op0_mode and use it instead of GET_MODE (op0). Update calls to
5671 extract_split_bit_field, extract_bit_field_using_extv and
5672 extract_fixed_bit_field.
5673 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5674 to extract_split_bit_field and extract_fixed_bit_field_1.
5675 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5676 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5678 (extract_split_bit_field): Add an op0_mode argument. Update call
5679 to extract_fixed_bit_field.
5681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5682 Alan Hayward <alan.hayward@arm.com>
5683 David Sherwood <david.sherwood@arm.com>
5685 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5686 * explow.c (hard_function_value): Likewise.
5687 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5688 convert_to_mode call outside the loop.
5689 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5690 for the mode iterator. Require the mode specified by max_pieces
5692 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5694 (copy_blkmode_to_reg): Likewise.
5695 (set_storage_via_setmem): Likewise.
5696 * optabs.c (prepare_cmp_insn): Likewise.
5697 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5698 * stor-layout.c (finish_bitfield_representative): Likewise.
5700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5701 Alan Hayward <alan.hayward@arm.com>
5702 David Sherwood <david.sherwood@arm.com>
5704 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5705 * expr.c (convert_move): Use them.
5706 (convert_modes): Likewise.
5707 (store_expr_with_bounds): Likewise.
5709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5710 Alan Hayward <alan.hayward@arm.com>
5711 David Sherwood <david.sherwood@arm.com>
5713 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5714 parameter for the mode of "x". Remove the "known_x", "known_mode"
5715 and "known_ret" arguments. Change the type of the mode argument
5717 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5718 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5719 (reg_num_sign_bit_copies_for_combine): Likewise.
5720 * rtlanal.c (nonzero_bits1): Likewise.
5721 (num_sign_bit_copies1): Likewise.
5722 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5723 (reg_num_sign_bit_copies_general): Likewise.
5724 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5725 (reg_nonzero_bits_general): Likewise.
5727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5728 Alan Hayward <alan.hayward@arm.com>
5729 David Sherwood <david.sherwood@arm.com>
5731 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5732 than in subroutines. Return 1 for non-integer modes.
5733 (cached_num_sign_bit_copies): Change the type of the mode parameter
5735 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5736 classes. Handle CONST_INT_P first and then check whether X also
5737 has a scalar integer mode. Check the same thing for inner registers
5738 of a SUBREG and for values that are being extended or truncated.
5740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5741 Alan Hayward <alan.hayward@arm.com>
5742 David Sherwood <david.sherwood@arm.com>
5744 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5745 in subroutines. Return the mode mask for non-integer modes.
5746 (cached_nonzero_bits): Change the type of the mode parameter
5748 (nonzero_bits1): Likewise. Remove early exit for other mode
5749 classes. Handle CONST_INT_P first and then check whether X
5750 also has a scalar integer mode.
5752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5753 Alan Hayward <alan.hayward@arm.com>
5754 David Sherwood <david.sherwood@arm.com>
5756 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5757 Return a scalar_int_mode and assert that the size is greater than
5759 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5760 (op_by_pieces_d::op_by_pieces_d): Likewise.
5761 (op_by_pieces_d::run): Likewise.
5762 (can_store_by_pieces): Likewise.
5764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5765 Alan Hayward <alan.hayward@arm.com>
5766 David Sherwood <david.sherwood@arm.com>
5768 * combine.c (extract_left_shift): Add a mode argument and update
5770 (make_compound_operation_int): Change the type of the mode parameter
5771 to scalar_int_mode and update the call to extract_left_shift.
5773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5774 Alan Hayward <alan.hayward@arm.com>
5775 David Sherwood <david.sherwood@arm.com>
5777 * combine.c (simplify_and_const_int): Change the type of the mode
5778 parameter to scalar_int_mode.
5779 (simplify_and_const_int_1): Likewise. Update recursive call.
5781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5782 Alan Hayward <alan.hayward@arm.com>
5783 David Sherwood <david.sherwood@arm.com>
5785 * combine.c (simplify_compare_const): Check that the mode is a
5786 scalar_int_mode (rather than VOIDmode) before testing its
5788 (simplify_comparison): Move COMPARISON_P handling out of the
5789 loop and restrict the latter part of the loop to scalar_int_modes.
5790 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5791 and when considering SUBREG_REGs. Use is_int_mode instead of
5792 checking GET_MODE_CLASS against MODE_INT.
5794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5795 Alan Hayward <alan.hayward@arm.com>
5796 David Sherwood <david.sherwood@arm.com>
5798 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5799 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5800 shift_unit_mode and for modes involved in scalar shifts.
5802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5803 Alan Hayward <alan.hayward@arm.com>
5804 David Sherwood <david.sherwood@arm.com>
5806 * combine.c (force_int_to_mode): New function, split out from...
5807 (force_to_mode): ...here. Keep xmode up-to-date and use it
5808 instead of GET_MODE (x).
5810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5811 Alan Hayward <alan.hayward@arm.com>
5812 David Sherwood <david.sherwood@arm.com>
5814 * optabs-query.h (extraction_insn::struct_mode): Change type to
5815 opt_scalar_int_mode and update comment.
5816 (extraction_insn::field_mode): Change type to scalar_int_mode.
5817 (extraction_insn::pos_mode): Likewise.
5818 * combine.c (make_extraction): Update accordingly.
5819 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5820 (get_optab_extraction_insn): Likewise.
5821 * recog.c (simplify_while_replacing): Likewise.
5822 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5823 parameter to opt_scalar_int_mode.
5825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5826 Alan Hayward <alan.hayward@arm.com>
5827 David Sherwood <david.sherwood@arm.com>
5829 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5830 to a scalar_int_mode instead of a machine_mode.
5831 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5832 (get_best_mode): Return a boolean and use a pointer argument to store
5833 the selected mode. Replace the limit mode parameter with a bit limit.
5834 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5835 for the values returned by bit_field_mode_iterator::next_mode.
5836 (store_bit_field): Update call to get_best_mode.
5837 (store_fixed_bit_field): Likewise.
5838 (extract_fixed_bit_field): Likewise.
5839 * expr.c (optimize_bitfield_assignment_op): Likewise.
5840 * fold-const.c (optimize_bit_field_compare): Likewise.
5841 (fold_truth_andor_1): Likewise.
5842 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5843 Update for new type of m_mode.
5844 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5851 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5852 (store_bit_field): Check is_a <scalar_int_mode> before calling
5853 strict_volatile_bitfield_p.
5854 (extract_bit_field): Likewise.
5856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5857 Alan Hayward <alan.hayward@arm.com>
5858 David Sherwood <david.sherwood@arm.com>
5860 * target.def (cstore_mode): Return a scalar_int_mode.
5861 * doc/tm.texi: Regenerate.
5862 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5863 * targhooks.h (default_cstore_mode): Likewise.
5864 * targhooks.c (default_cstore_mode): Likewise, using a forced
5866 * expmed.c (emit_cstore): Expect the target of the cstore to be
5869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5870 Alan Hayward <alan.hayward@arm.com>
5871 David Sherwood <david.sherwood@arm.com>
5873 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5875 (niter_desc): Likewise mode.
5876 (iv_analyze): Add a mode parameter.
5878 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5879 and change its type to scalar_int_mode.
5880 * loop-iv.c: Update commentary at head of file.
5881 (iv_constant): Pass the mode paraeter before the rtx it describes
5882 and change its type to scalar_int_mode. Remove VOIDmode handling.
5883 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5884 (iv_extend): Likewise.
5885 (shorten_into_mode): Likewise.
5886 (iv_add): Use scalar_int_mode.
5887 (iv_mult): Likewise.
5888 (iv_shift): Likewise.
5889 (canonicalize_iv_subregs): Likewise.
5890 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5891 it describes and change its mode to scalar_int_mode. Also change
5892 the type of the returned inner_mode to scalar_int_mode.
5893 (get_biv_step): Likewise, turning outer_mode from a pointer
5894 into a direct parameter. Update call to get_biv_step_1.
5895 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5896 iv_constant and get_biv_step.
5897 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5898 and change its type to scalar_int_mode. Don't initialise iv->mode
5899 to VOIDmode and remove later checks for its still being VOIDmode.
5900 Update calls to iv_analyze_op and iv_analyze_expr. Check
5901 is_a <scalar_int_mode> when changing the mode under consideration.
5902 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5903 Update call to iv_analyze_expr.
5904 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5905 inner register is not also a scalar_int_mode. Update call to
5907 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5908 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5909 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5910 separate mode class checks. Update calls to iv_analyze. Remove
5911 fix-up of VOIDmodes after iv_analyze_biv.
5912 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5913 don't have a scalar_int_mode. Update call to biv_p.
5915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5916 Alan Hayward <alan.hayward@arm.com>
5917 David Sherwood <david.sherwood@arm.com>
5919 * cfgexpand.c (convert_debug_memory_address): Use
5920 as_a <scalar_int_mode>.
5921 * combine.c (expand_compound_operation): Likewise.
5922 (make_extraction): Likewise.
5923 (change_zero_ext): Likewise.
5924 (simplify_comparison): Likewise.
5925 * cse.c (cse_insn): Likewise.
5926 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5927 (mem_loc_descriptor): Likewise.
5928 (loc_descriptor): Likewise.
5929 * expmed.c (init_expmed_one_mode): Likewise.
5930 (synth_mult): Likewise.
5931 (emit_store_flag_1): Likewise.
5932 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5933 of a comparison with size.
5934 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5935 (reduce_to_bit_field_precision): Likewise.
5936 * function.c (expand_function_end): Likewise.
5937 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5938 * loop-doloop.c (doloop_modify): Likewise.
5939 * optabs.c (expand_binop): Likewise.
5940 (expand_unop): Likewise.
5941 (expand_copysign_absneg): Likewise.
5942 (prepare_cmp_insn): Likewise.
5943 (maybe_legitimize_operand): Likewise.
5944 * recog.c (const_scalar_int_operand): Likewise.
5945 * rtlanal.c (get_address_mode): Likewise.
5946 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5947 (simplify_cond_clz_ctz): Likewise.
5948 * tree-nested.c (get_nl_goto_field): Likewise.
5949 * tree.c (build_vector_type_for_mode): Likewise.
5950 * var-tracking.c (use_narrower_mode): Likewise.
5952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5953 Alan Hayward <alan.hayward@arm.com>
5954 David Sherwood <david.sherwood@arm.com>
5956 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5957 * builtins.c (expand_builtin_signbit): Use it.
5958 * cfgexpand.c (expand_debug_expr): Likewise.
5959 * dojump.c (do_jump): Likewise.
5960 (do_compare_and_jump): Likewise.
5961 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5962 * expmed.c (make_tree): Likewise.
5963 * expr.c (expand_expr_real_2): Likewise.
5964 (expand_expr_real_1): Likewise.
5965 (try_casesi): Likewise.
5966 * fold-const-call.c (fold_const_call_ss): Likewise.
5967 * fold-const.c (unextend): Likewise.
5968 (extract_muldiv_1): Likewise.
5969 (fold_single_bit_test): Likewise.
5970 (native_encode_int): Likewise.
5971 (native_encode_string): Likewise.
5972 (native_interpret_int): Likewise.
5973 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5974 * internal-fn.c (expand_addsub_overflow): Likewise.
5975 (expand_neg_overflow): Likewise.
5976 (expand_mul_overflow): Likewise.
5977 (expand_arith_overflow): Likewise.
5978 * match.pd: Likewise.
5979 * stor-layout.c (layout_type): Likewise.
5980 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5981 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5982 * tree-ssanames.c (get_range_info): Likewise.
5983 * tree-switch-conversion.c (array_value_type) Likewise.
5984 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5985 (vect_recog_divmod_pattern): Likewise.
5986 (vect_recog_mixed_size_cond_pattern): Likewise.
5987 * tree-vrp.c (extract_range_basic): Likewise.
5988 (simplify_float_conversion_using_ranges): Likewise.
5989 * tree.c (int_fits_type_p): Likewise.
5990 * ubsan.c (instrument_bool_enum_load): Likewise.
5991 * varasm.c (mergeable_string_section): Likewise.
5992 (narrowing_initializer_constant_valid_p): Likewise.
5993 (output_constant): Likewise.
5995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5996 Alan Hayward <alan.hayward@arm.com>
5997 David Sherwood <david.sherwood@arm.com>
5999 * machmode.h (NARROWEST_INT_MODE): New macro.
6000 * expr.c (alignment_for_piecewise_move): Use it instead of
6001 GET_CLASS_NARROWEST_MODE (MODE_INT).
6002 (push_block): Likewise.
6003 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6005 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6008 Alan Hayward <alan.hayward@arm.com>
6009 David Sherwood <david.sherwood@arm.com>
6011 * postreload.c (move2add_valid_value_p): Change the type of the
6012 mode parameter to scalar_int_mode.
6013 (move2add_use_add2_insn): Add a mode parameter and use it instead
6015 (move2add_use_add3_insn): Likewise.
6016 (reload_cse_move2add): Update accordingly.
6018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6019 Alan Hayward <alan.hayward@arm.com>
6020 David Sherwood <david.sherwood@arm.com>
6022 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6024 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6025 * optabs.c (expand_unop): Likewise.
6027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6028 Alan Hayward <alan.hayward@arm.com>
6029 David Sherwood <david.sherwood@arm.com>
6031 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6032 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6033 (popcount_loc_descriptor): Likewise.
6034 (bswap_loc_descriptor): Likewise.
6035 (rotate_loc_descriptor): Likewise.
6036 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6037 calling the functions above.
6039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6040 Alan Hayward <alan.hayward@arm.com>
6041 David Sherwood <david.sherwood@arm.com>
6043 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6045 (try_combine): Likewise.
6046 (simplify_if_then_else): Likewise.
6047 * cse.c (cse_insn): Likewise.
6048 * dwarf2out.c (mem_loc_descriptor): Likewise.
6049 * emit-rtl.c (gen_lowpart_common): Likewise.
6050 * simplify-rtx.c (simplify_truncation): Likewise.
6051 (simplify_binary_operation_1): Likewise.
6052 (simplify_const_relational_operation): Likewise.
6053 (simplify_ternary_operation): Likewise.
6054 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6057 Alan Hayward <alan.hayward@arm.com>
6058 David Sherwood <david.sherwood@arm.com>
6060 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6061 * reload.c (push_reload): Likewise.
6062 (find_reloads): Likewise.
6064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6065 Alan Hayward <alan.hayward@arm.com>
6066 David Sherwood <david.sherwood@arm.com>
6068 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6069 (make_compound_operation_int): Likewise.
6070 (change_zero_ext): Likewise.
6071 * expr.c (convert_move): Likewise.
6072 (convert_modes): Likewise.
6073 * fwprop.c (forward_propagate_subreg): Likewise.
6074 * loop-iv.c (get_biv_step_1): Likewise.
6075 * optabs.c (widen_operand): Likewise.
6076 * postreload.c (move2add_valid_value_p): Likewise.
6077 * recog.c (simplify_while_replacing): Likewise.
6078 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6079 (simplify_binary_operation_1): Likewise. Remove redundant
6080 mode equality check.
6082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6083 Alan Hayward <alan.hayward@arm.com>
6084 David Sherwood <david.sherwood@arm.com>
6086 * combine.c (combine_simplify_rtx): Add checks for
6087 is_a <scalar_int_mode>.
6088 (simplify_if_then_else): Likewise.
6089 (make_field_assignment): Likewise.
6090 (simplify_comparison): Likewise.
6091 * ifcvt.c (noce_try_bitop): Likewise.
6092 * loop-invariant.c (canonicalize_address_mult): Likewise.
6093 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6096 Alan Hayward <alan.hayward@arm.com>
6097 David Sherwood <david.sherwood@arm.com>
6099 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6100 is_a <scalar_int_mode> instead of != BLKmode.
6102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6103 Alan Hayward <alan.hayward@arm.com>
6104 David Sherwood <david.sherwood@arm.com>
6106 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6107 instead of != VOIDmode.
6108 * combine.c (if_then_else_cond): Likewise.
6109 (change_zero_ext): Likewise.
6110 * dwarf2out.c (mem_loc_descriptor): Likewise.
6111 (loc_descriptor): Likewise.
6112 * rtlanal.c (canonicalize_condition): Likewise.
6113 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6116 Alan Hayward <alan.hayward@arm.com>
6117 David Sherwood <david.sherwood@arm.com>
6119 * simplify-rtx.c (simplify_binary_operation_1): Use
6120 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6123 Alan Hayward <alan.hayward@arm.com>
6124 David Sherwood <david.sherwood@arm.com>
6126 * wide-int.h (int_traits<unsigned char>) New class.
6127 (int_traits<unsigned short>) Likewise.
6128 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6129 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6131 * combine.c (set_nonzero_bits_and_sign_copies): Use
6132 is_a <scalar_int_mode>.
6133 (find_split_point): Likewise.
6134 (combine_simplify_rtx): Likewise.
6135 (simplify_logical): Likewise.
6136 (expand_compound_operation): Likewise.
6137 (expand_field_assignment): Likewise.
6138 (make_compound_operation): Likewise.
6139 (extended_count): Likewise.
6140 (change_zero_ext): Likewise.
6141 (simplify_comparison): Likewise.
6142 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6143 (ucompare_loc_descriptor): Likewise.
6144 (minmax_loc_descriptor): Likewise.
6145 (mem_loc_descriptor): Likewise.
6146 (loc_descriptor): Likewise.
6147 * expmed.c (init_expmed_one_mode): Likewise.
6148 * lra-constraints.c (lra_constraint_offset): Likewise.
6149 * optabs.c (prepare_libcall_arg): Likewise.
6150 * postreload.c (move2add_note_store): Likewise.
6151 * reload.c (operands_match_p): Likewise.
6152 * rtl.h (load_extend_op): Likewise.
6153 * rtlhooks.c (gen_lowpart_general): Likewise.
6154 * simplify-rtx.c (simplify_truncation): Likewise.
6155 (simplify_unary_operation_1): Likewise.
6156 (simplify_binary_operation_1): Likewise.
6157 (simplify_const_binary_operation): Likewise.
6158 (simplify_const_relational_operation): Likewise.
6159 (simplify_subreg): Likewise.
6160 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6161 * var-tracking.c (adjust_mems): Likewise.
6162 (prepare_call_arguments): Likewise.
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 * machmode.h (is_int_mode): New fuction.
6169 * combine.c (find_split_point): Use it.
6170 (combine_simplify_rtx): Likewise.
6171 (simplify_if_then_else): Likewise.
6172 (simplify_set): Likewise.
6173 (simplify_shift_const_1): Likewise.
6174 (simplify_comparison): Likewise.
6175 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6176 * cse.c (notreg_cost): Likewise.
6177 (cse_insn): Likewise.
6178 * cselib.c (cselib_lookup_1): Likewise.
6179 * dojump.c (do_jump_1): Likewise.
6180 (do_compare_rtx_and_jump): Likewise.
6181 * dse.c (get_call_args): Likewise.
6182 * dwarf2out.c (rtl_for_decl_init): Likewise.
6183 (native_encode_initializer): Likewise.
6184 * expmed.c (emit_store_flag_1): Likewise.
6185 (emit_store_flag): Likewise.
6186 * expr.c (convert_modes): Likewise.
6187 (store_field): Likewise.
6188 (expand_expr_real_1): Likewise.
6189 * fold-const.c (fold_read_from_constant_string): Likewise.
6190 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6191 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6192 * optabs.c (expand_binop): Likewise.
6193 (expand_unop): Likewise.
6194 (expand_abs_nojump): Likewise.
6195 (expand_one_cmpl_abs_nojump): Likewise.
6196 * simplify-rtx.c (mode_signbit_p): Likewise.
6197 (val_signbit_p): Likewise.
6198 (val_signbit_known_set_p): Likewise.
6199 (val_signbit_known_clear_p): Likewise.
6200 (simplify_relational_operation_1): Likewise.
6201 * tree.c (vector_type_mode): Likewise.
6203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6204 Alan Hayward <alan.hayward@arm.com>
6205 David Sherwood <david.sherwood@arm.com>
6207 * machmode.h (smallest_mode_for_size): Fix formatting.
6208 (smallest_int_mode_for_size): New function.
6209 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6210 instead of smallest_mode_for_size.
6211 * combine.c (make_extraction): Likewise.
6212 * config/arc/arc.c (arc_expand_movmem): Likewise.
6213 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6214 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6215 * config/s390/s390.c (s390_expand_insv): Likewise.
6216 * config/sparc/sparc.c (assign_int_registers): Likewise.
6217 * config/spu/spu.c (spu_function_value): Likewise.
6218 (spu_function_arg): Likewise.
6219 * coverage.c (get_gcov_type): Likewise.
6220 (get_gcov_unsigned_t): Likewise.
6221 * dse.c (find_shift_sequence): Likewise.
6222 * expmed.c (store_bit_field_1): Likewise.
6223 * expr.c (convert_move): Likewise.
6224 (store_field): Likewise.
6225 * internal-fn.c (expand_arith_overflow): Likewise.
6226 * optabs-query.c (get_best_extraction_insn): Likewise.
6227 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6228 * stor-layout.c (layout_type): Likewise.
6229 (initialize_sizetypes): Likewise.
6230 * targhooks.c (default_get_mask_mode): Likewise.
6231 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6234 Alan Hayward <alan.hayward@arm.com>
6235 David Sherwood <david.sherwood@arm.com>
6237 * machmode.h (opt_mode::else_blk): New function.
6238 (int_mode_for_mode): Declare.
6239 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6240 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6242 * cfgexpand.c (expand_debug_expr): Likewise.
6243 * combine.c (gen_lowpart_or_truncate): Likewise.
6244 (gen_lowpart_for_combine): Likewise.
6245 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6246 * config/avr/avr.c (avr_to_int_mode): Likewise.
6247 (avr_out_plus_1): Likewise.
6248 (avr_out_plus): Likewise.
6249 (avr_out_round): Likewise.
6250 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6251 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6252 (s390_expand_vcond): Likewise.
6253 * config/spu/spu.c (spu_split_immediate): Likewise.
6254 (spu_expand_mov): Likewise.
6255 * dse.c (get_stored_val): Likewise.
6256 * expmed.c (store_bit_field_1): Likewise.
6257 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6259 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6260 (extract_low_bits): Likewise.
6261 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6262 handling rather than repeating the check.
6263 (emit_group_store): Likewise.
6264 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6265 * optabs.c (expand_absneg_bit): Likewise.
6266 (expand_copysign_absneg): Likewise.
6267 (expand_copysign_bit): Likewise.
6268 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6269 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6270 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6271 * var-tracking.c (prepare_call_arguments): Likewise.
6272 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6273 int_mode_for_mode instead of mode_for_size.
6274 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6277 Alan Hayward <alan.hayward@arm.com>
6278 David Sherwood <david.sherwood@arm.com>
6280 * machmode.h (int_mode_for_size): New function.
6281 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6282 instead of mode_for_size.
6283 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6285 * combine.c (expand_field_assignment): Use int_mode_for_size
6286 instead of mode_for_size.
6287 (make_extraction): Likewise.
6288 (simplify_shift_const_1): Likewise.
6289 (simplify_comparison): Likewise.
6290 * dojump.c (do_jump): Likewise.
6291 * dwarf2out.c (mem_loc_descriptor): Likewise.
6292 * emit-rtl.c (init_derived_machine_modes): Likewise.
6293 * expmed.c (flip_storage_order): Likewise.
6294 (convert_extracted_bit_field): Likewise.
6295 * expr.c (copy_blkmode_from_reg): Likewise.
6296 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6297 * internal-fn.c (expand_mul_overflow): Likewise.
6298 * lower-subreg.c (simple_move): Likewise.
6299 * optabs-libfuncs.c (init_optabs): Likewise.
6300 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6301 * tree.c (vector_type_mode): Likewise.
6302 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6303 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6304 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6305 * tree-vect-stmts.c (vectorizable_load): Likewise.
6306 (vectorizable_store): Likewise.
6308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6309 Alan Hayward <alan.hayward@arm.com>
6310 David Sherwood <david.sherwood@arm.com>
6312 * coretypes.h (pod_mode): New type.
6313 (scalar_int_mode_pod): New typedef.
6314 * machmode.h (pod_mode): New class.
6315 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6316 * genmodes.c (emit_mode_int_n): Update accordingly.
6317 * lower-subreg.h (target_lower_subreg): Change type to
6318 scalar_int_mode_pod.
6319 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6320 scalar_int_mode_pod.
6322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6323 Alan Hayward <alan.hayward@arm.com>
6324 David Sherwood <david.sherwood@arm.com>
6326 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6327 machine_mode to scalar_int_mode.
6328 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6329 (rs6000_option_override_internal): Remove cast to int.
6330 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6331 machine_mode to scalar_int_mode.
6332 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6333 (rs6000_option_override_internal): Remove cast to int.
6334 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6335 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6337 * config/s390/s390.c (s390_expand_builtin): Likewise.
6338 * coretypes.h (scalar_int_mode): New type.
6339 (opt_scalar_int_mode): New typedef.
6340 * machmode.h (scalar_int_mode): New class.
6341 (scalar_int_mode::includes_p): New function.
6342 (byte_mode): Change type to scalar_int_mode.
6343 (word_mode): Likewise.
6344 (ptr_mode): Likewise.
6345 * emit-rtl.c (byte_mode): Likewise.
6346 (word_mode): Likewise.
6347 (ptr_mode): Likewise.
6348 (init_derived_machine_modes): Update accordingly.
6349 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6350 and MODE_PARTIAL_INT.
6351 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6352 opt_scalar_int_mode.
6354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6355 Alan Hayward <alan.hayward@arm.com>
6356 David Sherwood <david.sherwood@arm.com>
6358 * target.def (libgcc_floating_mode_supported_p): Take a
6360 * doc/tm.texi: Regenerate.
6361 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6363 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6364 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6368 Alan Hayward <alan.hayward@arm.com>
6369 David Sherwood <david.sherwood@arm.com>
6371 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6372 * doc/tm.texi: Regenerate.
6373 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6374 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6375 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6376 * targhooks.h (default_floatn_mode): Likewise.
6377 * targhooks.c (default_floatn_mode): Likewise.
6378 * tree.c (build_common_tree_nodes): Update accordingly.
6380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6381 Alan Hayward <alan.hayward@arm.com>
6382 David Sherwood <david.sherwood@arm.com>
6384 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6385 (mode_iterator::iterate_p): Likewise.
6386 (mode_iterator::get_wider): Likewise.
6387 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6390 Alan Hayward <alan.hayward@arm.com>
6391 David Sherwood <david.sherwood@arm.com>
6393 * coretypes.h (opt_scalar_float_mode): New typedef.
6394 * machmode.h (float_mode_for_size): New function.
6395 * emit-rtl.c (double_mode): Delete.
6396 (init_emit_once): Use float_mode_for_size.
6397 * stor-layout.c (layout_type): Likewise.
6398 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6401 Alan Hayward <alan.hayward@arm.com>
6402 David Sherwood <david.sherwood@arm.com>
6404 * output.h (assemble_real): Take a scalar_float_mode.
6405 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6406 * config/arm/arm.md (consttable_4): Likewise.
6407 (consttable_8): Likewise.
6408 (consttable_16): Likewise.
6409 * config/mips/mips.md (consttable_float): Likewise.
6410 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6411 * varasm.c (assemble_real): Take a scalar_float_mode.
6412 (output_constant_pool_2): Update accordingly.
6413 (output_constant): Likewise.
6415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6416 Alan Hayward <alan.hayward@arm.com>
6417 David Sherwood <david.sherwood@arm.com>
6419 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6420 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6421 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6422 (native_encode_real): Likewise.
6423 (native_interpret_real): Likewise.
6424 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6425 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6428 Alan Hayward <alan.hayward@arm.com>
6429 David Sherwood <david.sherwood@arm.com>
6431 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6432 <scalar_float_mode>. Simplify.
6433 (gen_extend_conv_libfunc): Likewise.
6435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6436 Alan Hayward <alan.hayward@arm.com>
6437 David Sherwood <david.sherwood@arm.com>
6439 * coretypes.h (scalar_float_mode): New type.
6440 * machmode.h (mode_traits::from_int): Use machine_mode if
6441 USE_ENUM_MODES is defined.
6442 (is_a): New function.
6444 (dyn_cast): Likewise.
6445 (scalar_float_mode): New class.
6446 (scalar_float_mode::includes_p): New function.
6447 (is_float_mode): Likewise.
6448 * gdbhooks.py (MachineModePrinter): New class.
6449 (build_pretty_printer): Use it for scalar_float_mode.
6450 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6451 (format_helper::format_helper): Turn into a template.
6452 * genmodes.c (get_mode_class): New function.
6453 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6454 or machine_mode if none.
6455 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6456 as_a <scalar_float_mode>.
6457 * dwarf2out.c (mem_loc_descriptor): Likewise.
6458 (insert_float): Likewise.
6459 (add_const_value_attribute): Likewise.
6460 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6461 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6462 (expand_unop): Update accordingly.
6463 (expand_abs_nojump): Likewise.
6464 (expand_copysign_absneg): Take a scalar_float_mode.
6465 (expand_copysign_bit): Likewise.
6466 (expand_copysign): Update accordingly.
6468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6469 Alan Hayward <alan.hayward@arm.com>
6470 David Sherwood <david.sherwood@arm.com>
6472 * coretypes.h (opt_mode): New class.
6473 * machmode.h (opt_mode): Likewise.
6474 (opt_mode::else_void): New function.
6475 (opt_mode::require): Likewise.
6476 (opt_mode::exists): Likewise.
6477 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6478 (GET_MODE_2XWIDER_MODE): Likewise.
6479 (mode_iterator::get_wider): Update accordingly.
6480 (mode_iterator::get_2xwider): Likewise.
6481 (mode_iterator::get_known_wider): Likewise, turning into a template.
6482 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6483 forcing a wider mode to exist.
6484 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6485 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6486 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6487 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6488 * lower-subreg.c (init_lower_subreg): Likewise.
6489 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6490 on the final iteration.
6491 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6492 a wider mode exists before asking for a move pattern.
6493 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6494 forcing a wider mode to exist.
6495 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6496 returning false if no such mode exists.
6497 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6498 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6499 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6500 Avoid checking for a MODE_INT if we already know the mode is not a
6502 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6503 forcing a wider mode to exist.
6504 (expmed_mult_highpart_optab): Likewise.
6505 (expmed_mult_highpart): Likewise.
6506 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6508 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6509 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6510 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6511 * internal-fn.c (expand_mul_overflow): Update use of
6512 GET_MODE_2XWIDER_MODE.
6513 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6514 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6515 GET_MODE_WIDER_MODE.
6516 (convert_plusminus_to_widen): Likewise.
6517 * tree-switch-conversion.c (array_value_type): Likewise.
6518 * var-tracking.c (emit_note_insn_var_location): Likewise.
6519 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6520 Return false inside rather than outside the loop if no wider mode
6522 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6523 and GET_MODE_2XWIDER_MODE
6524 (can_compare_p): Use else_void.
6525 * gdbhooks.py (OptMachineModePrinter): New class.
6526 (build_pretty_printer): Use it for opt_mode.
6528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6529 Alan Hayward <alan.hayward@arm.com>
6530 David Sherwood <david.sherwood@arm.com>
6532 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6533 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6536 Alan Hayward <alan.hayward@arm.com>
6537 David Sherwood <david.sherwood@arm.com>
6539 * machmode.h (mode_traits): New structure.
6540 (get_narrowest_mode): New function.
6541 (mode_iterator::start): Likewise.
6542 (mode_iterator::iterate_p): Likewise.
6543 (mode_iterator::get_wider): Likewise.
6544 (mode_iterator::get_known_wider): Likewise.
6545 (mode_iterator::get_2xwider): Likewise.
6546 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6547 (FOR_EACH_MODE): Likewise.
6548 (FOR_EACH_MODE_FROM): Likewise.
6549 (FOR_EACH_MODE_UNTIL): Likewise.
6550 (FOR_EACH_WIDER_MODE): Likewise.
6551 (FOR_EACH_2XWIDER_MODE): Likewise.
6552 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6553 * combine.c (simplify_comparison): Likewise
6554 * config/i386/i386.c (type_natural_mode): Likewise.
6555 * cse.c (cse_insn): Likewise.
6556 * dse.c (find_shift_sequence): Likewise.
6557 * emit-rtl.c (init_derived_machine_modes): Likewise.
6558 (init_emit_once): Likewise.
6559 * explow.c (hard_function_value): Likewise.
6560 * expmed.c (extract_fixed_bit_field_1): Likewise.
6561 (extract_bit_field_1): Likewise.
6562 (expand_divmod): Likewise.
6563 (emit_store_flag_1): Likewise.
6564 * expr.c (init_expr_target): Likewise.
6565 (convert_move): Likewise.
6566 (alignment_for_piecewise_move): Likewise.
6567 (widest_int_mode_for_size): Likewise.
6568 (emit_block_move_via_movmem): Likewise.
6569 (copy_blkmode_to_reg): Likewise.
6570 (set_storage_via_setmem): Likewise.
6571 (compress_float_constant): Likewise.
6572 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6573 * optabs-query.c (get_best_extraction_insn): Likewise.
6574 * optabs.c (expand_binop): Likewise.
6575 (expand_twoval_unop): Likewise.
6576 (expand_twoval_binop): Likewise.
6577 (widen_leading): Likewise.
6578 (widen_bswap): Likewise.
6579 (expand_parity): Likewise.
6580 (expand_unop): Likewise.
6581 (prepare_cmp_insn): Likewise.
6582 (prepare_float_lib_cmp): Likewise.
6583 (expand_float): Likewise.
6584 (expand_fix): Likewise.
6585 (expand_sfix_optab): Likewise.
6586 * postreload.c (move2add_use_add2_insn): Likewise.
6587 * reg-stack.c (reg_to_stack): Likewise.
6588 * reginfo.c (choose_hard_reg_mode): Likewise.
6589 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6590 * stor-layout.c (mode_for_size): Likewise.
6591 (smallest_mode_for_size): Likewise.
6592 (mode_for_vector): Likewise.
6593 (finish_bitfield_representative): Likewise.
6594 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6595 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6596 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6597 * var-tracking.c (prepare_call_arguments): Likewise.
6599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6600 Alan Hayward <alan.hayward@arm.com>
6601 David Sherwood <david.sherwood@arm.com>
6603 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6604 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6605 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6606 * machmode.h (mode_size): Move earlier in file.
6607 (mode_precision): Likewise.
6608 (mode_inner): Likewise.
6609 (mode_nunits): Likewise.
6610 (mode_unit_size): Likewise.
6611 (unit_unit_precision): Likewise.
6612 (mode_wider): Likewise.
6613 (mode_2xwider): Likewise.
6614 (machine_mode): New class.
6615 (mode_to_bytes): New function.
6616 (mode_to_bits): Likewise.
6617 (mode_to_precision): Likewise.
6618 (mode_to_inner): Likewise.
6619 (mode_to_unit_size): Likewise.
6620 (mode_to_unit_precision): Likewise.
6621 (mode_to_nunits): Likewise.
6622 (GET_MODE_SIZE): Use mode_to_bytes.
6623 (GET_MODE_BITSIZE): Use mode_to_bits.
6624 (GET_MODE_PRECISION): Use mode_to_precision.
6625 (GET_MODE_INNER): Use mode_to_inner.
6626 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6627 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6628 (GET_MODE_NUNITS): Use mode_to_nunits.
6629 * system.h (ALWAYS_INLINE): New macro.
6630 * config/powerpcspe/powerpcspe-c.c
6631 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6632 int for arg1_mode and arg2_mode.
6634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6635 Alan Hayward <alan.hayward@arm.com>
6636 David Sherwood <david.sherwood@arm.com>
6638 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6639 Prefix mode names with E_ in case statements.
6640 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6641 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6642 (aarch64_split_simd_move): Likewise.
6643 (aarch64_gen_storewb_pair): Likewise.
6644 (aarch64_gen_loadwb_pair): Likewise.
6645 (aarch64_gen_store_pair): Likewise.
6646 (aarch64_gen_load_pair): Likewise.
6647 (aarch64_get_condition_code_1): Likewise.
6648 (aarch64_constant_pool_reload_icode): Likewise.
6649 (get_rsqrte_type): Likewise.
6650 (get_rsqrts_type): Likewise.
6651 (get_recpe_type): Likewise.
6652 (get_recps_type): Likewise.
6653 (aarch64_gimplify_va_arg_expr): Likewise.
6654 (aarch64_simd_container_mode): Likewise.
6655 (aarch64_emit_load_exclusive): Likewise.
6656 (aarch64_emit_store_exclusive): Likewise.
6657 (aarch64_expand_compare_and_swap): Likewise.
6658 (aarch64_gen_atomic_cas): Likewise.
6659 (aarch64_emit_bic): Likewise.
6660 (aarch64_emit_atomic_swap): Likewise.
6661 (aarch64_emit_atomic_load_op): Likewise.
6662 (aarch64_evpc_trn): Likewise.
6663 (aarch64_evpc_uzp): Likewise.
6664 (aarch64_evpc_zip): Likewise.
6665 (aarch64_evpc_ext): Likewise.
6666 (aarch64_evpc_rev): Likewise.
6667 (aarch64_evpc_dup): Likewise.
6668 (aarch64_gen_ccmp_first): Likewise.
6669 (aarch64_gen_ccmp_next): Likewise.
6670 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6671 (alpha_emit_xfloating_libcall): Likewise.
6672 (emit_insxl): Likewise.
6673 (alpha_arg_type): Likewise.
6674 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6675 (arc_preferred_simd_mode): Likewise.
6676 (arc_secondary_reload): Likewise.
6677 (get_arc_condition_code): Likewise.
6678 (arc_print_operand): Likewise.
6679 (arc_legitimate_constant_p): Likewise.
6680 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6681 * config/arc/arc.md (casesi_load): Likewise.
6682 (casesi_compact_jump): Likewise.
6683 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6684 (cc_use_register): Likewise.
6685 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6686 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6687 (arm_init_iwmmxt_builtins): Likewise.
6688 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6689 (neon_expand_vector_init): Likewise.
6690 (arm_attr_length_move_neon): Likewise.
6691 (maybe_get_arm_condition_code): Likewise.
6692 (arm_emit_vector_const): Likewise.
6693 (arm_preferred_simd_mode): Likewise.
6694 (arm_output_iwmmxt_tinsr): Likewise.
6695 (thumb1_output_casesi): Likewise.
6696 (thumb2_output_casesi): Likewise.
6697 (arm_emit_load_exclusive): Likewise.
6698 (arm_emit_store_exclusive): Likewise.
6699 (arm_expand_compare_and_swap): Likewise.
6700 (arm_evpc_neon_vuzp): Likewise.
6701 (arm_evpc_neon_vzip): Likewise.
6702 (arm_evpc_neon_vrev): Likewise.
6703 (arm_evpc_neon_vtrn): Likewise.
6704 (arm_evpc_neon_vext): Likewise.
6705 (arm_validize_comparison): Likewise.
6706 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6707 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6708 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6709 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6710 (c6x_preferred_simd_mode): Likewise.
6711 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6712 (epiphany_rtx_costs): Likewise.
6713 * config/epiphany/predicates.md (proper_comparison_operator):
6715 * config/frv/frv.c (condexec_memory_operand): Likewise.
6716 (frv_emit_move): Likewise.
6717 (output_move_single): Likewise.
6718 (output_condmove_single): Likewise.
6719 (frv_hard_regno_mode_ok): Likewise.
6720 (frv_matching_accg_mode): Likewise.
6721 * config/h8300/h8300.c (split_adds_subs): Likewise.
6722 (h8300_rtx_costs): Likewise.
6723 (h8300_print_operand): Likewise.
6724 (compute_mov_length): Likewise.
6725 (output_logical_op): Likewise.
6726 (compute_logical_op_length): Likewise.
6727 (compute_logical_op_cc): Likewise.
6728 (h8300_shift_needs_scratch_p): Likewise.
6729 (output_a_shift): Likewise.
6730 (compute_a_shift_length): Likewise.
6731 (compute_a_shift_cc): Likewise.
6732 (expand_a_rotate): Likewise.
6733 (output_a_rotate): Likewise.
6734 * config/i386/i386.c (classify_argument): Likewise.
6735 (function_arg_advance_32): Likewise.
6736 (function_arg_32): Likewise.
6737 (function_arg_64): Likewise.
6738 (function_value_64): Likewise.
6739 (ix86_gimplify_va_arg): Likewise.
6740 (ix86_legitimate_constant_p): Likewise.
6741 (put_condition_code): Likewise.
6742 (split_double_mode): Likewise.
6743 (ix86_avx256_split_vector_move_misalign): Likewise.
6744 (ix86_expand_vector_logical_operator): Likewise.
6745 (ix86_split_idivmod): Likewise.
6746 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6747 (ix86_build_const_vector): Likewise.
6748 (ix86_build_signbit_mask): Likewise.
6749 (ix86_match_ccmode): Likewise.
6750 (ix86_cc_modes_compatible): Likewise.
6751 (ix86_expand_branch): Likewise.
6752 (ix86_expand_sse_cmp): Likewise.
6753 (ix86_expand_sse_movcc): Likewise.
6754 (ix86_expand_int_sse_cmp): Likewise.
6755 (ix86_expand_vec_perm_vpermi2): Likewise.
6756 (ix86_expand_vec_perm): Likewise.
6757 (ix86_expand_sse_unpack): Likewise.
6758 (ix86_expand_int_addcc): Likewise.
6759 (ix86_split_to_parts): Likewise.
6760 (ix86_vectorize_builtin_gather): Likewise.
6761 (ix86_vectorize_builtin_scatter): Likewise.
6762 (avx_vpermilp_parallel): Likewise.
6763 (inline_memory_move_cost): Likewise.
6764 (ix86_tieable_integer_mode_p): Likewise.
6765 (x86_maybe_negate_const_int): Likewise.
6766 (ix86_expand_vector_init_duplicate): Likewise.
6767 (ix86_expand_vector_init_one_nonzero): Likewise.
6768 (ix86_expand_vector_init_one_var): Likewise.
6769 (ix86_expand_vector_init_concat): Likewise.
6770 (ix86_expand_vector_init_interleave): Likewise.
6771 (ix86_expand_vector_init_general): Likewise.
6772 (ix86_expand_vector_set): Likewise.
6773 (ix86_expand_vector_extract): Likewise.
6774 (emit_reduc_half): Likewise.
6775 (ix86_emit_i387_round): Likewise.
6776 (ix86_mangle_type): Likewise.
6777 (ix86_expand_round_sse4): Likewise.
6778 (expand_vec_perm_blend): Likewise.
6779 (canonicalize_vector_int_perm): Likewise.
6780 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6781 (expand_vec_perm_1): Likewise.
6782 (expand_vec_perm_interleave3): Likewise.
6783 (expand_vec_perm_even_odd_pack): Likewise.
6784 (expand_vec_perm_even_odd_1): Likewise.
6785 (expand_vec_perm_broadcast_1): Likewise.
6786 (ix86_vectorize_vec_perm_const_ok): Likewise.
6787 (ix86_expand_vecop_qihi): Likewise.
6788 (ix86_expand_mul_widen_hilo): Likewise.
6789 (ix86_expand_sse2_abs): Likewise.
6790 (ix86_expand_pextr): Likewise.
6791 (ix86_expand_pinsr): Likewise.
6792 (ix86_preferred_simd_mode): Likewise.
6793 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6794 * config/i386/sse.md (*andnot<mode>3): Likewise.
6795 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6796 (*<code><mode>3): Likewise.
6797 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6798 (ia64_expand_atomic_op): Likewise.
6799 (ia64_arg_type): Likewise.
6800 (ia64_mode_to_int): Likewise.
6801 (ia64_scalar_mode_supported_p): Likewise.
6802 (ia64_vector_mode_supported_p): Likewise.
6803 (expand_vec_perm_broadcast): Likewise.
6804 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6805 (iq2000_function_arg_advance): Likewise.
6806 (iq2000_function_arg): Likewise.
6807 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6808 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6809 (m68k_libcall_value): Likewise.
6810 (m68k_function_value): Likewise.
6811 (sched_attr_op_type): Likewise.
6812 * config/mcore/mcore.c (mcore_output_move): Likewise.
6813 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6815 (microblaze_function_arg): Likewise.
6816 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6817 (mips_print_operand): Likewise.
6818 (mips_mode_ok_for_mov_fmt_p): Likewise.
6819 (mips_vector_mode_supported_p): Likewise.
6820 (mips_preferred_simd_mode): Likewise.
6821 (mips_expand_vpc_loongson_even_odd): Likewise.
6822 (mips_expand_vec_unpack): Likewise.
6823 (mips_expand_vi_broadcast): Likewise.
6824 (mips_expand_vector_init): Likewise.
6825 (mips_expand_vec_reduc): Likewise.
6826 (mips_expand_msa_cmp): Likewise.
6827 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6828 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6829 (cc_flags_for_mode): Likewise.
6830 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6831 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6832 (nds32_output_casesi_pc_relative): Likewise.
6833 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6834 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6835 (nvptx_gen_unpack): Likewise.
6836 (nvptx_gen_pack): Likewise.
6837 (nvptx_gen_shuffle): Likewise.
6838 (nvptx_gen_wcast): Likewise.
6839 (nvptx_preferred_simd_mode): Likewise.
6840 * config/pa/pa.c (pa_secondary_reload): Likewise.
6841 * config/pa/predicates.md (base14_operand): Likewise.
6842 * config/powerpcspe/powerpcspe-c.c
6843 (altivec_resolve_overloaded_builtin): Likewise.
6844 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6846 (rs6000_preferred_simd_mode): Likewise.
6847 (output_vec_const_move): Likewise.
6848 (rs6000_expand_vector_extract): Likewise.
6849 (rs6000_split_vec_extract_var): Likewise.
6850 (reg_offset_addressing_ok_p): Likewise.
6851 (rs6000_legitimate_offset_address_p): Likewise.
6852 (rs6000_legitimize_address): Likewise.
6853 (rs6000_emit_set_const): Likewise.
6854 (rs6000_const_vec): Likewise.
6855 (rs6000_emit_move): Likewise.
6856 (spe_build_register_parallel): Likewise.
6857 (rs6000_darwin64_record_arg_recurse): Likewise.
6858 (swap_selector_for_mode): Likewise.
6859 (spe_init_builtins): Likewise.
6860 (paired_init_builtins): Likewise.
6861 (altivec_init_builtins): Likewise.
6862 (do_load_for_compare): Likewise.
6863 (rs6000_generate_compare): Likewise.
6864 (rs6000_expand_float128_convert): Likewise.
6865 (emit_load_locked): Likewise.
6866 (emit_store_conditional): Likewise.
6867 (rs6000_output_function_epilogue): Likewise.
6868 (rs6000_handle_altivec_attribute): Likewise.
6869 (rs6000_function_value): Likewise.
6870 (emit_fusion_gpr_load): Likewise.
6871 (emit_fusion_p9_load): Likewise.
6872 (emit_fusion_p9_store): Likewise.
6873 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6874 (fusion_gpr_mem_load): Likewise.
6875 (fusion_addis_mem_combo_load): Likewise.
6876 (fusion_addis_mem_combo_store): Likewise.
6877 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6878 (fusion_gpr_mem_load): Likewise.
6879 (fusion_addis_mem_combo_load): Likewise.
6880 (fusion_addis_mem_combo_store): Likewise.
6881 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6883 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6884 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6885 (rs6000_preferred_simd_mode): Likewise.
6886 (output_vec_const_move): Likewise.
6887 (rs6000_expand_vector_extract): Likewise.
6888 (rs6000_split_vec_extract_var): Likewise.
6889 (reg_offset_addressing_ok_p): Likewise.
6890 (rs6000_legitimate_offset_address_p): Likewise.
6891 (rs6000_legitimize_address): Likewise.
6892 (rs6000_emit_set_const): Likewise.
6893 (rs6000_const_vec): Likewise.
6894 (rs6000_emit_move): Likewise.
6895 (rs6000_darwin64_record_arg_recurse): Likewise.
6896 (swap_selector_for_mode): Likewise.
6897 (paired_init_builtins): Likewise.
6898 (altivec_init_builtins): Likewise.
6899 (rs6000_expand_float128_convert): Likewise.
6900 (emit_load_locked): Likewise.
6901 (emit_store_conditional): Likewise.
6902 (rs6000_output_function_epilogue): Likewise.
6903 (rs6000_handle_altivec_attribute): Likewise.
6904 (rs6000_function_value): Likewise.
6905 (emit_fusion_gpr_load): Likewise.
6906 (emit_fusion_p9_load): Likewise.
6907 (emit_fusion_p9_store): Likewise.
6908 * config/rx/rx.c (rx_gen_move_template): Likewise.
6909 (flags_from_mode): Likewise.
6910 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6911 (s390_slb_comparison): Likewise.
6912 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6913 (s390_vector_mode_supported_p): Likewise.
6914 (s390_cc_modes_compatible): Likewise.
6915 (s390_match_ccmode_set): Likewise.
6916 (s390_canonicalize_comparison): Likewise.
6917 (s390_emit_compare_and_swap): Likewise.
6918 (s390_branch_condition_mask): Likewise.
6919 (s390_rtx_costs): Likewise.
6920 (s390_secondary_reload): Likewise.
6921 (__SECONDARY_RELOAD_CASE): Likewise.
6922 (s390_expand_cs): Likewise.
6923 (s390_preferred_simd_mode): Likewise.
6924 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6925 * config/sh/sh.c (sh_print_operand): Likewise.
6926 (dump_table): Likewise.
6927 (sh_secondary_reload): Likewise.
6928 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6929 * config/sh/sh.md (casesi_worker_1): Likewise.
6930 (casesi_worker_2): Likewise.
6931 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6932 (fcc_comparison_operator): Likewise.
6933 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6934 (emit_soft_tfmode_cvt): Likewise.
6935 (sparc_preferred_simd_mode): Likewise.
6936 (output_cbranch): Likewise.
6937 (sparc_print_operand): Likewise.
6938 (sparc_expand_vec_perm_bmask): Likewise.
6939 (vector_init_bshuffle): Likewise.
6940 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6941 (spu_vector_mode_supported_p): Likewise.
6942 (spu_expand_insv): Likewise.
6943 (spu_emit_branch_or_set): Likewise.
6944 (spu_handle_vector_attribute): Likewise.
6945 (spu_builtin_splats): Likewise.
6946 (spu_builtin_extract): Likewise.
6947 (spu_builtin_promote): Likewise.
6948 (spu_expand_sign_extend): Likewise.
6949 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6950 (tilegx_simd_int): Likewise.
6951 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6952 (tilepro_simd_int): Likewise.
6953 * config/v850/v850.c (const_double_split): Likewise.
6954 (v850_print_operand): Likewise.
6955 (ep_memory_offset): Likewise.
6956 * config/vax/vax.c (vax_rtx_costs): Likewise.
6957 (vax_output_int_move): Likewise.
6958 (vax_output_int_add): Likewise.
6959 (vax_output_int_subtract): Likewise.
6960 * config/visium/predicates.md (visium_branch_operator): Likewise.
6961 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6962 (visium_print_operand_address): Likewise.
6963 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6964 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6965 (xtensa_expand_conditional_branch): Likewise.
6966 (xtensa_copy_incoming_a7): Likewise.
6967 (xtensa_output_literal): Likewise.
6968 * dfp.c (decimal_real_maxval): Likewise.
6969 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6972 Alan Hayward <alan.hayward@arm.com>
6973 David Sherwood <david.sherwood@arm.com>
6975 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6976 (mode_nunits_inline): Likewise.
6977 (mode_inner_inline): Likewise.
6978 (mode_unit_size_inline): Likewise.
6979 (mode_unit_precision_inline): Likewise.
6980 (emit_insn_modes_h): Likewise. Also emit a #define of the
6982 (emit_mode_wider): Add an E_ prefix to mode names.
6983 (emit_mode_complex): Likewise.
6984 (emit_mode_inner): Likewise.
6985 (emit_mode_adjustments): Likewise.
6986 (emit_mode_int_n): Likewise.
6987 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6988 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6989 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6990 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6991 (CRC32_BUILTIN, ENTRY): Likewise.
6992 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6993 (aarch64_pop_regs): Likewise.
6994 (aarch64_process_components): Likewise.
6995 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6996 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6997 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6998 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6999 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7000 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7002 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7003 (pcmpistr): Likewise.
7004 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7005 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7006 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7008 * config/rl78/rl78.c (mduc_regs): Likewise.
7009 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7010 (htm_expand_builtin): Likewise.
7011 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7012 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7013 * config/xtensa/xtensa.c (print_operand): Likewise.
7014 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7015 (NUM_MODE_VECTOR_INT): Likewise.
7016 * genoutput.c (null_operand): Likewise.
7017 (output_operand_data): Likewise.
7018 * genrecog.c (print_parameter_value): Likewise.
7019 * lra.c (debug_operand_data): Likewise.
7021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7023 * dwarf2out.c (scompare_loc_descriptor_wide)
7024 (scompare_loc_descriptor_narrow): New functions, split out from...
7025 (scompare_loc_descriptor): ...here.
7026 * expmed.c (emit_store_flag_int): New function, split out from...
7027 (emit_store_flag): ...here.
7029 2017-08-30 Richard Biener <rguenther@suse.de>
7031 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7032 (dwarf2out_early_finish): Move setting of AT_pubnames from
7033 early debug output to early finish.
7035 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7037 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7038 and -mdata-region to the assembler.
7040 2017-08-30 Richard Biener <rguenther@suse.de>
7042 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7044 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7045 (dwarf2out_early_global_decl): Only generate a DIE for the
7046 abstract origin if it doesn't already exist or is a declaration DIE.
7047 (resolve_addr): Do not add the linkage name twice when
7048 generating a stub DIE for the DW_TAG_GNU_call_site target.
7050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7052 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7053 Use machine_mode rather than int for arg1_mode.
7055 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7058 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7059 that the second argument of the built-in functions to unpack
7060 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7061 switch statement instead a lot of if statements.
7062 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7063 Allow 64-bit values to be in Altivec registers as well as
7064 traditional floating point registers.
7065 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7067 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7069 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7070 MAX_REGS_PER_ADDRESS == 1.
7072 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7074 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7075 * config/i386/i386.c (ix86_option_override_internal): Simplify
7076 setting of opts->x_flag_entry.
7078 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7079 Jakub Jelinek <jakub@redhat.com>
7080 Richard Biener <rguenther@suse.de>
7082 PR tree-optimization/81503
7083 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7084 folded constant fits in the target type; reorder tests for clarity.
7086 2017-08-29 Martin Liska <mliska@suse.cz>
7088 * passes.def: Include pass_lower_switch.
7089 * stmt.c (dump_case_nodes): Remove and move to
7090 tree-switch-conversion.
7091 (case_values_threshold): Likewise.
7092 (expand_switch_as_decision_tree_p): Likewise.
7093 (emit_case_decision_tree): Likewise.
7094 (expand_case): Likewise.
7095 (balance_case_nodes): Likewise.
7096 (node_has_low_bound): Likewise.
7097 (node_has_high_bound): Likewise.
7098 (node_is_bounded): Likewise.
7099 (emit_case_nodes): Likewise.
7100 (struct simple_case_node): New struct.
7101 (add_case_node): Remove.
7102 (emit_case_dispatch_table): Use vector instead of case_list.
7103 (reset_out_edges_aux): Remove.
7104 (compute_cases_per_edge): Likewise.
7105 (expand_case): Build list of simple_case_node.
7106 (expand_sjlj_dispatch_table): Use it.
7107 * tree-switch-conversion.c (struct case_node): Moved from
7108 stmt.c and adjusted.
7109 (emit_case_nodes): Likewise.
7110 (node_has_low_bound): Likewise.
7111 (node_has_high_bound): Likewise.
7112 (node_is_bounded): Likewise.
7113 (case_values_threshold): Likewise.
7114 (reset_out_edges_aux): Likewise.
7115 (compute_cases_per_edge): Likewise.
7116 (add_case_node): Likewise.
7117 (dump_case_nodes): Likewise.
7118 (balance_case_nodes): Likewise.
7119 (expand_switch_as_decision_tree_p): Likewise.
7120 (emit_jump): Likewise.
7121 (emit_case_decision_tree): Likewise.
7122 (try_switch_expansion): Likewise.
7123 (do_jump_if_equal): Likewise.
7124 (emit_cmp_and_jump_insns): Likewise.
7125 (fix_phi_operands_for_edge): New function.
7126 (record_phi_operand_mapping): Likewise.
7127 (class pass_lower_switch): New pass.
7128 (pass_lower_switch::execute): New function.
7129 (make_pass_lower_switch): Likewise.
7130 (conditional_probability):
7131 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7132 * tree-pass.h: Add make_pass_lower_switch.
7134 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7137 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7140 2017-08-29 Richard Biener <rguenther@suse.de>
7142 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7143 we do not add a DW_AT_inline attribute twice.
7144 (gen_subprogram_die): Remove code setting DW_AT_inline on
7145 DECL_ABSTRACT_P nodes.
7147 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7149 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7150 calls to internal functions.
7151 (gimplify_modify_expr): Likewise.
7152 * tree-call-cdce.c (use_internal_fn): Likewise.
7153 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7154 (convert_to_divmod): Set the nothrow flag.
7155 * tree-if-conv.c (predicate_mem_writes): Likewise.
7156 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7157 (vectorizable_call): Likewise.
7158 (vectorizable_store): Likewise.
7159 (vectorizable_load): Likewise.
7160 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7161 (vect_recog_mask_conversion_pattern): Likewise.
7163 2017-08-29 Martin Liska <mliska@suse.cz>
7166 * gcc.c (driver_handle_option): Add new argument.
7167 * opts-common.c (handle_option): Pass
7168 target_option_override_hook.
7169 * opts-global.c (lang_handle_option): Add new option.
7170 (set_default_handlers): Add new argument.
7171 (decode_options): Likewise.
7172 * opts.c (target_handle_option): Likewise.
7173 (common_handle_option): Call target_option_override_hook.
7174 * opts.h (struct cl_option_handler_func): Add hook for
7175 target option override.
7176 (struct cl_option_handlers): Likewise.
7177 (set_default_handlers): Add new argument.
7178 (decode_options): Likewise.
7179 (common_handle_option): Likewise.
7180 (target_handle_option): Likewise.
7181 * toplev.c (toplev::main): Pass targetm.target_option.override
7184 2017-08-29 Richard Biener <rguenther@suse.de>
7185 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7187 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7188 life to the active subtree.
7190 2017-08-28 Jeff Law <law@redhat.com>
7192 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7193 derive_equivalences.
7194 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7196 (edge_info::derive_equivalences): New private member function
7198 * tree-ssa-dom.c (class edge_info): Changed from a struct
7199 to a class. Add ctor/dtor, methods and data members.
7200 (edge_info::edge_info): Renamed from allocate_edge_info.
7201 Initialize additional members.
7202 (edge_info::~edge_info): New.
7203 (free_dom_edge_info): Delete the edge info.
7204 (record_edge_info): Use new class & associated member functions.
7205 Tighten forms for testing for edge equivalences.
7206 (record_temporary_equivalences): Iterate over the simple
7207 equivalences rather than assuming there's only one per edge.
7208 (cprop_into_successor_phis): Iterate over the simple
7209 equivalences rather than assuming there's only one per edge.
7210 (optimize_stmt): Use operand_equal_p rather than pointer
7211 equality for mini-DSE code.
7213 2017-08-28 Nathan Sidwell <nathan@acm.org>
7215 * gcc.c (execute): Fold SIGPIPE handling into switch
7216 statement. Adjust internal error message.
7218 2017-08-28 Richard Biener <rguenther@suse.de>
7221 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7222 Do nothing for removed DIEs.
7224 2017-08-28 Richard Biener <rguenther@suse.de>
7226 PR tree-optimization/81977
7227 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7230 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7233 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7234 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7235 using targetm.gen_mem_thread_fence.
7237 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7240 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7241 predicate to register_operand. Reorder operands.
7242 (*btr<mode>): Ditto.
7243 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7244 (*btr<mode>_mask): Ditto.
7246 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7248 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7249 * config/rs6000/xmmintrin.h: New file.
7250 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7252 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7255 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7256 parameter and_insn and return it.
7257 (recombine_lvx_pattern): Insert a copy to ensure availability of
7258 the base register of the copied masking operation at the point of
7259 the instruction replacement.
7260 (recombine_stvx_pattern): Likewise.
7262 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7264 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7265 undocumented switches.
7266 (-mpower9-dform-vector): Likewise.
7267 (-mpower9-dform): Likewise.
7268 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7269 comments to delete references to -mpower9-dform* switches.
7270 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7271 Delete reference to -mpower9-dform* switches, test for
7272 -mpower9-vector instead.
7273 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7274 (OTHER_P9_VECTOR_MASKS): Likewise.
7275 (POWERPC_MASKS): Likewise.
7276 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7277 tests against -mpower9-dform* to -mpower9-vector. Delete code
7278 that checked for -mpower9-dform* consistancy with other options.
7279 Add test for -mpower9-misc to enable other power9 switches.
7280 (rs6000_init_hard_regno_mode_ok): Likewise.
7281 (rs6000_option_override_internal): Likewise.
7282 (rs6000_emit_prologue): Likewise.
7283 (rs6000_emit_epilogue): Likewise.
7284 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7285 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7286 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7288 (emit_fusion_p9_store): Likewise.
7289 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7290 resetting these macros if the assembler does not support ISA 3.0
7292 (TARGET_P9_DFORM_VECTOR): Likewise.
7293 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7294 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7296 2017-08-25 Alan Modra <amodra@gmail.com>
7298 PR rtl-optimization/81747
7299 * cse.c (cse_extended_basic_block): Don't attempt to record
7300 equivalences for degenerate conditional branches that branch
7301 to their fall-through.
7303 2017-08-24 Martin Sebor <msebor@redhat.com>
7306 * gimple-fold.c (size_must_be_zero_p): New function.
7307 (gimple_fold_builtin_memory_op): Call it.
7309 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7311 * config/rs6000/mm_malloc.h: New file.
7313 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7315 PR tree-optimization/81913
7316 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7317 analysis when either IVs in condition can wrap.
7319 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7321 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7322 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7324 2017-08-24 Richard Biener <rguenther@suse.de>
7327 * targhooks.c (default_target_can_inline_p): Properly
7328 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7329 is present and always compare.
7330 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7331 infer -mfpmath=sse from TARGET_SSE_P.
7332 (ix86_can_inline_p): Properly use target_option_default_node when
7333 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7335 2017-08-24 Richard Biener <rguenther@suse.de>
7338 * dwarf2out.c (output_die): Handle flag_generate_offload like
7340 (output_comp_unit): Likewise.
7341 (gen_array_type_die): Likewise.
7342 (dwarf2out_early_finish): Likewise.
7343 (note_variable_value_in_expr): Likewise.
7344 (dwarf2out_finish): Likewise. Adjust assert.
7345 * cgraphunit.c (symbol_table::compile): Move setting of
7346 flag_generate_offload earlier ...
7347 (symbol_table::finalize_compilation_unit): ... here, before
7348 early debug finalization.
7350 2017-08-24 Richard Biener <rguenther@suse.de>
7352 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7353 and ipa-fnsummary.h.
7354 (ix86_can_inline_p): When ix86_fpmath flags do not match
7355 check whether the callee uses FP math at all.
7357 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7360 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7361 instead of TYPE_PRECISION.
7363 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7364 Alan Hayward <alan.hayward@arm.com>
7365 David Sherwood <david.sherwood@arm.com>
7367 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7369 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7371 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7373 (ix86_frame::stack_realign_allocate): New field.
7374 (struct machine_frame_state): Modify comments.
7375 (machine_frame_state::sp_realigned_fp_end): New field.
7376 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7378 (sp_valid_at): Add assertion to assure no attempt to access invalid
7379 offset of a realigned stack.
7380 (fp_valid_at): Likewise.
7381 (choose_baseaddr): Modify comments.
7382 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7383 ix86_expand_prologue.
7384 (ix86_expand_prologue): Modify stack realignment and allocation.
7385 (ix86_expand_epilogue): Modify comments.
7386 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7387 avx2_runtime, avx512f, and avx512f_runtime.
7389 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7391 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7392 (mstackrealign): Do not init to -1.
7393 * config/i386/i386.c (ix86_option_override_internal):
7394 Check opts_set, not opts when setting default value of
7395 opts->x_ix86_force_align_arg_pointer.
7397 2017-08-23 Richard Biener <rguenther@suse.de>
7399 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7400 lang_hooks.decl_printable_name.
7401 * print-rtl-function.c (print_rtx_function): Likewise.
7402 * tree-pretty-print.c (dump_function_header): Likewise.
7404 2017-08-23 Richard Biener <rguenther@suse.de>
7407 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7408 -g0 at compile-time.
7410 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7413 * doc/sourcebuild.texi (Other hardware attributes):
7416 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7419 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7420 Fix single-use check.
7422 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7424 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7426 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7428 * doc/install.texi: Modify to add more details on running selected
7431 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7433 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7434 is combined with -mabi=ms.
7435 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7438 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7440 PR tree-optimization/81488
7441 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7442 and cached_basis fields.
7443 (MAX_SPREAD): New constant.
7444 (alloc_cand_and_find_basis): Initialize new fields.
7445 (clear_visited): New function.
7446 (create_phi_basis_1): Rename from create_phi_basis, set visited
7447 and cached_basis fields.
7448 (create_phi_basis): New wrapper function.
7449 (phi_add_costs_1): Rename from phi_add_costs, add spread
7450 parameter, set visited field, short-circuit when limits reached.
7451 (phi_add_costs): New wrapper function.
7452 (record_phi_increments_1): Rename from record_phi_increments, set
7454 (record_phi_increments): New wrapper function.
7455 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7456 (phi_incr_cost): New wrapper function.
7457 (all_phi_incrs_profitable_1): Rename from
7458 all_phi_incrs_profitable, set visited field.
7459 (all_phi_incrs_profitable): New wrapper function.
7461 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7462 Alan Hayward <alan.hayward@arm.com>
7463 David Sherwood <david.sherwood@arm.com>
7465 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7466 that takes the outer and inner modes.
7467 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7468 comparison as the canonical test for a paradoxical subreg.
7469 * combine.c (simplify_set): Use paradoxical_subreg_p.
7470 (make_extraction): Likewise.
7471 (force_to_mode): Likewise.
7472 (rtx_equal_for_field_assignment_p): Likewise.
7473 (gen_lowpart_for_combine): Likewise.
7474 (simplify_comparison): Likewise.
7475 * cse.c (equiv_constant): Likewise.
7476 * expmed.c (store_bit_field_1): Likewise.
7477 * final.c (alter_subreg): Likewise.
7478 * fwprop.c (propagate_rtx): Likewise.
7479 (forward_propagate_subreg): Likewise.
7480 * ira-conflicts.c (ira_build_conflicts): Likewise.
7481 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7482 * lra-constraints.c (curr_insn_transform): Likewise.
7483 (split_reg): Likewise.
7484 * lra-eliminations.c (move_plus_up): Likewise.
7485 (lra_eliminate_regs_1): Likewise.
7486 * recog.c (general_operand): Likewise.
7487 * ree.c (combine_reaching_defs): Likewise.
7488 * reload.c (push_reload): Likewise.
7489 (find_reloads): Likewise.
7490 * reload1.c (elimination_effects): Likewise.
7491 (compute_reload_subreg_offset): Likewise.
7492 (choose_reload_regs): Likewise.
7493 * rtlanal.c (subreg_lsb_1): Likewise.
7494 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7495 (simplify_subreg): Likewise.
7496 * var-tracking.c (track_loc_p): Likewise.
7497 * emit-rtl.c (byte_lowpart_offset): Likewise.
7498 (paradoxical_subreg_p): Delete out-of-line definition.
7500 2017-08-22 Jeff Law <law@redhat.com>
7502 PR tree-optimization/81741
7503 PR tree-optimization/71947
7504 * tree-ssa-dom.c: Include tree-inline.h.
7505 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7506 equivalences if one is more expensive to compute than the other.
7507 * tree-ssa-scopedtables.h (class const_or_copies): Make
7508 record_const_or_copy_raw method private.
7509 (class avail_exprs_stack): New method simplify_binary_operation.
7510 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7511 avail_exprs_stack::simplify_binary_operation as needed.
7512 (avail_exprs_stack::simplify_binary_operation): New function.
7514 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7516 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7517 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7518 (DOT_SYMBOLS): Likewise.
7519 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7520 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7521 (RS6000_ABI_NAME): Likewise.
7522 (TARGET_CMODEL): Likewise.
7523 (TOC_SECTION_ASM_OP): Likewise.
7524 (SET_CMODEL): New macro.
7525 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7527 2017-08-22 Richard Biener <rguenther@suse.de>
7529 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7530 to allow for free-lang-data replacements similar to verify_type_variant.
7532 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7533 Alan Hayward <alan.hayward@arm.com>
7534 David Sherwood <david.sherwood@arm.com>
7536 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7537 VOIDmode for the LABEL_REF.
7539 2017-08-22 Richard Biener <rguenther@suse.de>
7541 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7543 2017-08-22 Richard Biener <rguenther@suse.de>
7545 * common.opt (feliminate-dwarf2-dups): Ignore.
7546 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7547 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7548 same_die_p_wrap, compute_section_prefix,
7549 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7550 (comdat_symbol_id, comdat_symbol_number): Likewise.
7551 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7553 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7554 (output_die): Mark unreachable path unreachable.
7555 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7556 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7557 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7558 (dwarf2out_early_finish): Likewise.
7560 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7562 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7564 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7567 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7568 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7569 (avr_attribute_table) <io, io_low, address>: Initialize
7570 .decl_required with true.
7572 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7574 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7575 undocumented debugging options.
7576 (-mvsx-scalar-double): Likewise.
7577 (-mallow-df-permute): Likewise.
7578 (-mvectorize-builtins): Likewise.
7579 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7580 (rs6000_builtin_vectorized_function): Likewise.
7581 (rs6000_builtin_md_vectorized_function): Likewise.
7582 (rs6000_opt_vars): Likewise.
7584 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7587 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7588 (*btrq_imm): Rename from *btrq.
7589 (*btcq_imm): Rename from *btcq.
7590 (btsc): New code attribute.
7591 (*<btsc><mode>): New insn pattern.
7592 (*btr<mode>): Ditto.
7593 (*<btsc><mode>_mask): New insn_and_split pattern.
7594 (*btr<mode>_mask): Ditto.
7596 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7597 Alan Hayward <alan.hayward@arm.com>
7598 David Sherwood <david.sherwood@arm.com>
7600 * function.c (pad_below): Simplify padding calculation.
7602 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7603 Alan Hayward <alan.hayward@arm.com>
7604 David Sherwood <david.sherwood@arm.com>
7606 * target.def (function_prologue): Remove frame size argument.
7607 (function_epilogue): Likewise.
7608 * doc/tm.texi: Regenerate.
7609 * final.c (final_start_function): Update call to function_prologue.
7610 (final_end_function): Update call to function_epilogue.
7611 (default_function_pro_epilogue): Remove frame size argument.
7612 * output.h (default_function_pro_epilogue): Likewise.
7613 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7614 (arm_output_function_prologue): Likewise.
7615 * config/frv/frv.c (frv_function_prologue): Likewise.
7616 (frv_function_epilogue): Likewise.
7617 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7618 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7619 (ia64_output_function_epilogue): Likewise.
7620 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7621 (m32r_output_function_epilogue): Likewise.
7622 * config/microblaze/microblaze.c (microblaze_function_prologue)
7623 (microblaze_function_epilogue): Likewise.
7624 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7625 (mips_output_function_epilogue): Likewise.
7626 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7627 (mmix_target_asm_function_epilogue): Likewise.
7628 * config/msp430/msp430.c (msp430_start_function): Likewise.
7629 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7630 (nds32_asm_function_epilogue): Likewise.
7631 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7632 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7633 (pa_output_function_epilogue): Likewise.
7634 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7635 (rs6000_output_function_epilogue): Likewise.
7636 * config/rl78/rl78.c (rl78_start_function): Likewise.
7637 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7638 (rs6000_output_function_epilogue): Likewise.
7639 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7640 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7641 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7642 (sparc_asm_function_epilogue): Likewise.
7644 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7646 * tree.h (type_has_mode_precision_p): New function.
7647 * convert.c (convert_to_integer_1): Use it.
7648 * expr.c (expand_expr_real_2): Likewise.
7649 (expand_expr_real_1): Likewise.
7650 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7651 * match.pd: Likewise.
7652 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7653 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7654 * tree-tailcall.c (process_assignment): Likewise.
7655 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7656 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7657 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7658 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7659 (vectorizable_assignment): Likewise.
7660 (vectorizable_shift): Likewise.
7661 (vectorizable_operation): Likewise.
7662 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7664 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7666 * match.pd: Add pow (C, x) simplification.
7668 2017-08-21 Richard Biener <rguenther@suse.de>
7670 PR tree-optimization/81900
7671 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7672 for blocks with abnormal predecessors.
7673 (compute_antic): Do not set visited flag prematurely.
7675 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7678 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7680 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7682 * stor-layout.h (vector_type_mode): Move to...
7683 * tree.h (vector_type_mode): ...here.
7684 * stor-layout.c (vector_type_mode): Move to...
7685 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7687 2017-08-21 Richard Biener <rguenther@suse.de>
7689 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7690 register_external_die hooks.
7691 (debug_false_tree_charstarstar_uhwistar): Declare.
7692 (debug_nothing_tree_charstar_uhwi): Likewise.
7693 * debug.c (do_nothing_debug_hooks): Adjust.
7694 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7695 (debug_nothing_tree_charstar_uhwi): Likewise.
7696 * dbxout.c (dbx_debug_hooks): Adjust.
7697 (xcoff_debug_hooks): Likewise.
7698 * sdbout.c (sdb_debug_hooks): Likewise.
7699 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7700 * dwarf2out.c (macinfo_label_base): New global.
7701 (dwarf2out_register_external_die): New function for the
7702 register_external_die hook.
7703 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7704 (dwarf2_debug_hooks): Use them.
7705 (dwarf2_lineno_debug_hooks): Adjust.
7706 (struct die_struct): Add with_offset flag.
7707 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7708 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7709 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7710 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7711 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7712 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7713 defining section names for the early LTO debug variants.
7714 (reset_indirect_string): New helper.
7715 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7716 (print_dw_val): Add support for offsetted symbol references.
7717 (get_ultimate_context): Split out from is_cxx.
7718 (is_cxx): Use get_ultimate_context.
7719 (is_fortran): Add decl overload.
7720 (compute_comp_unit_symbol): Split out worker from
7721 compute_section_prefix.
7722 (compute_section_prefix): Call compute_comp_unit_symbol and
7723 set comdat_type_p here.
7724 (output_die): Skip DIE symbol output for the LTO added one.
7725 Handle DIE symbol references with offset.
7726 (output_comp_unit): Guard section name mangling properly.
7727 For LTO debug sections emit a symbol at the section beginning
7728 which we use to refer to its DIEs.
7729 (add_abstract_origin_attribute): For DIEs registered via
7730 dwarf2out_register_external_die directly refer to the early
7731 DIE rather than indirectly through the shadow one we created.
7732 Remove obsolete call to dwarf2out_abstract_function for
7733 non-function/block origins.
7734 (gen_array_type_die): When generating early LTO debug do
7735 not emit DW_AT_string_length.
7736 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7737 late when in LTO. As suggested place a gcc_unreachable for
7738 the DECL_ABSTRACT_P case.
7739 (gen_subprogram_die): Avoid another specification DIE
7740 for early built declarations/definitions for the late LTO case.
7741 (gen_variable_die): Add type references for late duplicated VLA dies
7743 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7744 we have the abstract instance already.
7745 (process_scope_var): Adjust decl DIE contexts in LTO which
7746 first puts them in limbo.
7747 (gen_decl_die): Do not generate type DIEs late apart from
7748 types for VLAs or for decls we do not yet have a DIE. Do not
7749 call dwarf2out_abstract_function late.
7750 (dwarf2out_early_global_decl): Make sure to create DIEs
7751 for abstract instances of a decl first.
7752 (dwarf2out_late_global_decl): Adjust comment.
7753 (output_macinfo_op): With multiple macro sections use
7754 macinfo_label_base to distinguish labels.
7755 (output_macinfo): Likewise. Update macinfo_label_base.
7756 Pass in the line info label.
7757 (note_variable_value_in_expr): When generating LTO resolve
7758 all variable values here by generating DIEs as needed.
7759 (init_sections_and_labels): Add early LTO debug flag parameter
7760 and generate different sections and names if set. Add generation
7761 counter for the labels so we can have multiple of them.
7762 (reset_dies): Helper to allow DIEs to be output multiple times.
7763 (dwarf2out_finish): When outputting DIEs to the fat part of an
7764 LTO object first reset DIEs.
7765 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7766 (modified_type_die): Check for decl_ultimate_origin being self
7768 (gen_type_die_with_usage): Likewise.
7769 (gen_typedef_die): Allow decl_ultimate_origin being self.
7770 (set_decl_abstract_flags): Remove.
7771 (set_block_abstract_flags): Likewise.
7772 (dwarf2out_abstract_function): Treat the early generated DIEs
7773 as the abstract copy and only add DW_AT_inline and
7774 DW_AT_artificial here and call set_decl_origin_self.
7775 If the DIE has an abstract origin don't do anything.
7776 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7777 if we have none yet (Go fails to build one, PR78628).
7778 (variably_modified_type_p): Prevent endless recursion for Ada
7779 cyclic pointer types.
7780 * lto-streamer-in.c: Include debug.h.
7781 (dref_queue): New global.
7782 (lto_read_tree_1): Stream in DIE references.
7783 (lto_input_tree): Register DIE references.
7784 (input_function): Stream DECL_DEBUG_ARGS.
7785 * lto-streamer-out.c: Include debug.h.
7786 (lto_write_tree_1): Output DIE references.
7787 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7788 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7789 (output_function): Stream DECL_DEBUG_ARGS.
7790 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7791 Stream DECL_ABSTRACT_ORIGIN.
7792 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7793 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7794 DECL_CONTEXT for file-scope decls.
7795 * lto-streamer.h (struct dref_entry): Declare.
7796 (dref_queue): Likewise.
7797 * cfgexpand.c (pass_expand::execute): Do not call the
7798 outlining_inline_function hook here.
7799 * lto-wrapper.c (debug_obj): New global.
7800 (tool_cleanup): Unlink it if required.
7801 (debug_objcopy): New function.
7802 (run_gcc): Handle early debug sections in the IL files by
7803 extracting them to separate files, partially linkin them and
7804 feeding the result back as result to the linker.
7805 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7806 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7807 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7808 sections into a separate segment.
7809 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7811 (darwin_asm_dwarf_section): Likewise.
7812 (darwin_asm_output_dwarf_offset): Likewise.
7813 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7815 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7816 Alan Hayward <alan.hayward@arm.com>
7817 David Sherwood <david.sherwood@arm.com>
7819 * read-md.h (md_reader::record_potential_iterator_use): Replace
7820 pointer argument with an rtx and an index.
7821 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7822 (apply_mode_iterator): Likewise.
7823 (apply_code_iterator): Likewise.
7824 (apply_int_iterator): Likewise.
7825 (apply_subst_iterator): Likewise.
7826 (record_iterator_use): Likewise.
7827 (record_attribute_use): Likewise.
7828 (md_reader::record_potential_iterator_use): Likewise. Update calls
7829 to record_iterator_use and apply_iterator.
7830 (iterator_use): Replace ptr with x and index.
7831 (attribute_use): Likewise.
7832 (apply_attribute_uses): Update calls to apply_iterator.
7833 (apply_iterators): Likewise. Update initialization of iterator_use.
7834 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7835 and record_potential_iterator_use.
7836 (rtx_reader::read_rtx_operand): Likewise.
7838 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7839 Alan Hayward <alan.hayward@arm.com>
7840 David Sherwood <david.sherwood@arm.com>
7842 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7845 2017-08-21 Richard Biener <rguenther@suse.de>
7848 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7849 at struct end conservatively when comparing common bases.
7851 2017-08-21 Richard Biener <rguenther@suse.de>
7853 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7854 (mem_ref_in_stmt): Remove.
7855 (determine_max_movement): Use ref index to get at the reference.
7856 (invariantness_dom_walker::before_dom_children): Deal with
7857 lim data already initialized.
7858 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7860 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7862 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7863 (TARGET_ISA_ROUND): Ditto.
7864 (TARGET_ROUND): Ditto.
7865 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7866 * config/i386/i386.md: Ditto.
7867 * config/i386/sse.md: Ditto.
7868 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7869 with OPTION_MASK_ISA_SSE4_1.
7871 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7874 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7875 __builtin_ia32_lzcnt_u16.
7877 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7880 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7881 (rs6000_set_current_function): Rewrite function to use it.
7883 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7886 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7887 and TYPE_WARN_IF_NOT_ALIGN.
7888 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7889 (handle_warn_if_not_align): New.
7890 (place_union_field): Call handle_warn_if_not_align.
7891 (place_field): Call handle_warn_if_not_align.
7892 Copy TYPE_WARN_IF_NOT_ALIGN.
7893 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7894 (layout_type): Likewise.
7895 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7897 (tree_decl_common): Add warn_if_not_align.
7898 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7899 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7900 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7901 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7902 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7903 * doc/extend.texi: Document warn_if_not_aligned attribute.
7904 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7906 2017-08-17 Martin Liska <mliska@suse.cz>
7909 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7910 (get_data_dependence): Use it as pointer type.
7911 (distribute_loop): Likewise.
7913 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7915 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7916 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7917 (p8_vmrgow_<mode>_direct): New define_insn.
7918 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7919 handle endianness for vmrgew and vmrgow permute patterns.
7921 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7923 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7924 * config/rs6000/rs6000-cpus.def: Remove comment.
7925 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7926 (POWERPC_MASKS): Likewise.
7927 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7928 use of TARGET_VSX_TIMODE.
7929 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7930 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7931 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7932 (rs6000_option_override_internal): Remove dead code.
7933 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7934 (rs6000_legitimize_reload_address): Likewise.
7935 (rs6000_legitimate_address_p): Likewise.
7936 (rs6000_opt_masks): Delete "vsx-timode".
7937 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7938 from function comment.
7939 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7940 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7941 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7943 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7944 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7945 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7948 (define_peephole2): Likewise.
7950 2017-08-17 Martin Sebor <msebor@redhat.com>
7953 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7954 past the end of an array.
7955 (test_pp_format): Add test cases.
7957 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7959 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7960 missing ECF_NOTHROW flags.
7962 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7965 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7966 operands residing in integer registers.
7967 (*vsx_le_perm_load_<mode>): Likewise.
7968 (*vsx_le_perm_store_<mode>): Likewise.
7969 (define_peephole2): Add peepholes to optimize the above.
7971 2017-08-17 Marek Polacek <polacek@redhat.com>
7974 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7975 to mimic what shorten_compare did. Change the return type to bool.
7976 (fold_cond_expr_with_comparison): Update call to
7977 operand_equal_for_comparison_p.
7978 (fold_ternary_loc): Likewise.
7980 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7982 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7984 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7986 2017-08-17 Richard Biener <rguenther@suse.de>
7988 * tree-ssa-structalias.c (solve_graph): When propagating
7989 to successors update the graphs succ edges and avoid duplicate work.
7991 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7994 * config/i386/i386.c (ix86_option_override_internal): Save target
7995 specific options after ix86_stack_protector_guard_reg was changed.
7997 2017-08-17 Richard Biener <rguenther@suse.de>
7999 PR tree-optimization/81827
8000 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8001 (new_var_info): Initialize it conservatively.
8002 (get_call_vi): Mark register vars.
8003 (new_scalar_tmp_constraint_exp): Likewise.
8004 (handle_rhs_call): Likewise.
8005 (handle_const_call): Likewise.
8006 (create_function_info_for): Likewise.
8007 (solve_constraints): Sort varinfos to separate register from
8008 non-register vars to pack points-to solution bitmaps during
8011 2017-08-17 Marek Polacek <polacek@redhat.com>
8013 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8015 2017-08-17 Richard Biener <rguenther@suse.de>
8017 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8018 to true when overflow is undefined and we saturated the result.
8020 2017-08-17 Alan Modra <amodra@gmail.com>
8023 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8024 Don't use store multiple if only one reg needs saving.
8025 (interesting_frame_related_regno): New function.
8026 (rs6000_frame_related): Don't emit frame info for regs that
8028 (rs6000_emit_epilogue): Likewise.
8030 2017-08-16 Nathan Sidwell <nathan@acm.org>
8032 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8033 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8034 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8035 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8036 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8037 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8038 (verify_type): Adjust for TYPE_BINFO move.
8039 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8040 process TYPE_BINFO directly.
8041 (hash_tree): Likewise.
8042 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8044 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8047 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8049 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8051 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8054 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8055 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8056 Add ix86_binary_operator_ok to insn constraint.
8057 (*iordi_1_bts): Ditto.
8058 (*xordi_1_btc): Ditto.
8059 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8060 Update corresponding peephole2 pattern.
8064 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8066 PR tree-optimization/81832
8067 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8068 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8070 2017-08-16 Marek Polacek <polacek@redhat.com>
8073 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8074 regarding size_zero_node.
8076 2017-08-16 Martin Liska <mliska@suse.cz>
8079 * config.gcc: Respect previously set extra_objs in case
8082 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8084 PR tree-optimization/81835
8085 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8086 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8087 not depend on the phi.
8089 2017-08-16 Alan Modra <amodra@gmail.com>
8091 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8094 2017-08-16 Alan Modra <amodra@gmail.com>
8096 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8097 (save_reg_p): ..into this. Update all callers.
8098 (first_reg_to_save): Simplify.
8100 2017-08-16 Alan Modra <amodra@gmail.com>
8102 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8105 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8109 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8110 constant count if that count is less than 32.
8112 2017-08-15 Nathan Sidwell <nathan@acm.org>
8114 * gcc.c (execute): Emit friendlier message if inferior is killed
8115 by an external cause.
8117 2017-08-15 Richard Biener <rguenther@suse.de>
8119 PR tree-optimization/81790
8120 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8121 CONSTRUCTORs from simplifying and VN.
8123 2017-08-14 Martin Sebor <msebor@redhat.com>
8125 * builtin-attrs.def: Add comments.
8127 2017-08-14 Martin Sebor <msebor@redhat.com>
8130 * doc/extend.texi (attribute nonstring): Document new attribute.
8132 2017-08-14 Martin Sebor <msebor@redhat.com>
8135 * tree-diagnostic.c (default_tree_printer): Handle %G.
8136 * gimple-pretty-print.h (percent_G_format): Declare new function.
8137 * gimple-pretty-print.c (percent_G_format): Define.
8138 * tree-pretty-print.c (percent_K_format): Add argument.
8140 2017-08-14 Martin Sebor <msebor@redhat.com>
8142 PR translation/79998
8143 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8144 Remove a stray space.
8146 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8149 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8150 (*iordi_1_bts): Ditto.
8151 (*xordi_1_btc): Ditto.
8153 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8156 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8157 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8159 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8160 quoted strings, and make more translator-friendly.
8161 (darwin_rs6000_override_options): Likewise.
8162 (rs6000_option_override_internal): Likewise.
8163 (rs6000_return_in_memory): Fix overlong line.
8164 (init_cmulative_args): Use quoted strings, and make more
8165 translator-friendly.
8166 (rs6000_pass_by_reference): Fix overlong line.
8167 (def_builtin): Use quoted strings.
8168 (altivec_expand_predicate_builtin): Use quoted strings, and make
8169 more translator-friendly.
8170 (htm_expand_builtin): Use quoted strings.
8171 (cpu_expand_builtin): Use quoted strings, and make more
8172 translator-friendly.
8173 (altivec_expand_builtin): Likewise.
8174 (paired_expand_predicate_builtin): Likewise.
8175 (rs6000_invalid_builtin): Likewise.
8176 (builtin_function_type): Use quoted strings.
8177 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8178 more translator-friendly.
8179 (rs6000_trampoline_init): Likewise.
8180 (rs6000_handle_altivec_attribute): Likewise.
8181 (rs6000_inner_target_options): Use quoted strings.
8182 (rs6000_disable_incompatible_switches): Likewise.
8183 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8184 strings, and make more translator-friendly.
8185 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8187 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8189 PR tree-optimization/81799
8190 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8191 cond_expr to simple gimple operand.
8193 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8196 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8198 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8202 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8203 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8204 TARGET_GASISR_PROLOGUES.
8205 * config/avr/avr.c (avr_option_override): Same.
8206 (avr_pass_pre_proep::execute): Same.
8208 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8211 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8212 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8214 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8216 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8217 define_insn_and_split. Split to a memory load from 0 in
8218 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8219 using PTR mode iterator.
8220 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8221 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8222 (*add_tp_<mode>): Redefine as define_insn_and_split.
8223 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8224 address space. Merge with *add_tp_x32 using PTR mode iterator.
8225 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8226 Split to an add with a memory load from 0 in
8227 DEFAULT_TLS_SEG_REG address space.
8229 2017-08-12 Andrew Pinski <apinski@cavium.com>
8231 * config/aarch64/aarch64-option-extensions.def (rdma):
8232 Fix feature string to what Linux prints out in /proc/cpuinfo.
8234 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8237 * dwarf2out.c (modified_type_die): For C typedef types that have
8238 an ultimate origin, process the ultimate origin instead of the
8240 (gen_typedef_die): Assert that input DECLs have no ultimate
8242 (gen_type_die_with_usage): For typedef variants that have an
8243 ultimate origin, just call gen_decl_die on the original DECL.
8244 (process_scope_var): Avoid creating DIEs for local typedefs and
8245 concrete static variables.
8247 2017-08-12 Alan Modra <amodra@gmail.com>
8251 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8252 match gnu-user.h startfile.
8253 (ENDFILE_LINUX_SPEC): Similarly.
8255 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8258 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8260 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8262 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8263 * config/aarch64/aarch64.md (mov<mode>): Change.
8264 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8265 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8266 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8268 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8270 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8271 for storage order barriers.
8273 2017-08-11 Martin Liska <mliska@suse.cz>
8276 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8277 variables of void type.
8279 2017-08-11 Martin Liska <mliska@suse.cz>
8281 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8282 TARGET_SUPPORTS_ALIASES.
8283 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8284 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8285 (optimize_weakref): Likewise.
8286 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8287 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8288 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8290 2017-08-11 Martin Liska <mliska@suse.cz>
8293 * config/i386/i386.c (make_resolver_func): Do complete
8294 refactoring of the function.
8296 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8299 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8300 * config/i386/i386.c (ix86_stack_protect_guard): Use
8301 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8302 * doc/invoke.texi (x86 Options): Document
8303 -mstack-protector-guard-symbol= option.
8305 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8307 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8308 * config/i386/i386.c (ix86_split_stack_guard): New function.
8309 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8310 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8311 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8312 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8313 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8314 (split_stack_space_check): Call ix86_split_stack_guard.
8316 2017-08-10 Martin Sebor <msebor@redhat.com>
8318 * print-tree.c (print_node): Print location using the established
8320 Replace spaces with colons.
8321 (debug_raw, debug): Ditto.
8323 2017-08-10 Martin Sebor <msebor@redhat.com>
8326 * pretty-print.c (pp_format): Correct the handling of %s precision.
8328 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8331 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8333 (ix86_finalize_stack_frame_flags): This. Also clear
8334 frame_pointer_needed if -fno-omit-frame-pointer is used without
8336 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8337 with ix86_finalize_stack_frame_flags.
8338 (ix86_expand_epilogue): Likewise.
8339 (ix86_expand_split_stack_prologue): Likewise.
8340 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8342 2017-08-10 Martin Liska <mliska@suse.cz>
8345 * c-attribs.c (handle_target_attribute):
8346 Report warning for an empty string argument of target attribute.
8348 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8351 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8353 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8354 or DECL_NONLOCAL labels.
8355 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8356 or DECL_NONLOCAL labels here.
8358 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8360 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8361 to indicate when early gimple folding has been disabled.
8362 (rs6000_gimple_fold_builtin): Add debug content.
8363 (rs6000_invalid_builtin): Fix whitespace.
8364 (rs6000_expand_builtin): Fix whitespace.
8365 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8367 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8370 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8371 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8373 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8375 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8376 (qdf24xx): Likewise.
8377 * config/aarch64/aarch64.md: Include falkor.md.
8378 * config/aarch64/falkor.md: New.
8380 2017-08-09 Marek Polacek <polacek@redhat.com>
8383 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8384 * diagnostic.c (emit_diagnostic): Add a comment.
8385 (emit_diagnostic_valist): New function.
8387 2017-08-09 Marek Polacek <polacek@redhat.com>
8390 * input.c (make_location): New overload.
8391 * input.h (make_location): Declare.
8393 2017-08-08 Alan Modra <amodra@gmail.com>
8394 H.J. Lu <hongjiu.lu@intel.com>
8397 * gcc.c (NO_PIE_SPEC): Delete.
8398 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8400 (LINK_PIE_SPEC): ..to here.
8401 (LINK_COMMAND_SPEC): Support -no-pie.
8402 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8403 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8405 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8406 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8407 (ENDFILE_CRTEND_SPEC): Similarly.
8409 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8412 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8413 (mstack-protector-guard-offset=): Ditto.
8414 * config/i386/i386.c (ix86_option_override): Handle
8415 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8417 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8418 ix86_stack_protect_guard_offset variables.
8419 (TARGET_STACK_PROTECT_GUARD): Always define.
8420 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8421 and -mstack-protector-guard-offset= options.
8423 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8425 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8426 boundary case for the last candidate.
8428 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8430 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8431 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8433 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8436 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8437 * config/aarch64/aarch64-builtins.c
8438 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8439 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8440 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8442 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8443 Andrew Pinski <pinskia@gmail.com>
8446 * internal-fn.def (XORSIGN): New.
8447 * optabs.def (xorsign_optab): New.
8448 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8449 (convert_expand_mult_copysign): New.
8450 (pass_optimize_widening_mul::execute): Call
8451 convert_expand_mult_copysign.
8453 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8455 PR tree-optimization/81354
8456 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8457 Insert on edges rather than explicitly creating landing pads.
8458 (analyze_candidates_and_replace): Commit edge inserts.
8460 2017-08-08 Richard Biener <rguenther@suse.de>
8463 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8464 (expand_simple_operations): Also look through ADDR_EXPRs with
8465 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8467 2017-08-08 Richard Biener <rguenther@suse.de>
8469 PR tree-optimization/81723
8470 * tree-vect-slp.c (struct bst_traits): New hash traits.
8471 (bst_fail): New global.
8472 (vect_build_slp_tree_2): New worker, split out from ...
8473 (vect_build_slp_tree): ... this now wrapping it with using
8474 bst_fail set to cache SLP tree build fails. Properly handle
8476 (vect_analyze_slp_instance): Allocate and free bst_fail.
8478 2017-08-08 Martin Liska <mliska@suse.cz>
8481 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8482 LABEL_DECLs that can be from a different function.
8484 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8486 PR tree-optimization/81744
8487 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8488 loop's number of iterations.
8490 2017-08-08 Martin Liska <mliska@suse.cz>
8492 * asan.c: Include header files.
8493 * attribs.c (build_decl_attribute_variant): New function moved
8495 (build_type_attribute_qual_variant): Likewise.
8496 (cmp_attrib_identifiers): Likewise.
8497 (simple_cst_list_equal): Likewise.
8498 (omp_declare_simd_clauses_equal): Likewise.
8499 (attribute_value_equal): Likewise.
8500 (comp_type_attributes): Likewise.
8501 (build_type_attribute_variant): Likewise.
8502 (lookup_ident_attribute): Likewise.
8503 (remove_attribute): Likewise.
8504 (merge_attributes): Likewise.
8505 (merge_type_attributes): Likewise.
8506 (merge_decl_attributes): Likewise.
8507 (merge_dllimport_decl_attributes): Likewise.
8508 (handle_dll_attribute): Likewise.
8509 (attribute_list_equal): Likewise.
8510 (attribute_list_contained): Likewise.
8511 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8512 (lookup_attribute_by_prefix): Likewise.
8513 * bb-reorder.c: Include header files.
8514 * builtins.c: Likewise.
8515 * calls.c: Likewise.
8516 * cfgexpand.c: Likewise.
8517 * cgraph.c: Likewise.
8518 * cgraphunit.c: Likewise.
8519 * convert.c: Likewise.
8520 * dwarf2out.c: Likewise.
8521 * final.c: Likewise.
8522 * fold-const.c: Likewise.
8523 * function.c: Likewise.
8524 * gimple-expr.c: Likewise.
8525 * gimple-fold.c: Likewise.
8526 * gimple-pretty-print.c: Likewise.
8527 * gimple.c: Likewise.
8528 * gimplify.c: Likewise.
8529 * hsa-common.c: Likewise.
8530 * hsa-gen.c: Likewise.
8531 * internal-fn.c: Likewise.
8532 * ipa-chkp.c: Likewise.
8533 * ipa-cp.c: Likewise.
8534 * ipa-devirt.c: Likewise.
8535 * ipa-fnsummary.c: Likewise.
8536 * ipa-inline.c: Likewise.
8537 * ipa-visibility.c: Likewise.
8539 * lto-cgraph.c: Likewise.
8540 * omp-expand.c: Likewise.
8541 * omp-general.c: Likewise.
8542 * omp-low.c: Likewise.
8543 * omp-offload.c: Likewise.
8544 * omp-simd-clone.c: Likewise.
8545 * opts-global.c: Likewise.
8546 * passes.c: Likewise.
8547 * predict.c: Likewise.
8548 * sancov.c: Likewise.
8549 * sanopt.c: Likewise.
8550 * symtab.c: Likewise.
8551 * toplev.c: Likewise.
8552 * trans-mem.c: Likewise.
8553 * tree-chkp.c: Likewise.
8554 * tree-eh.c: Likewise.
8555 * tree-into-ssa.c: Likewise.
8556 * tree-object-size.c: Likewise.
8557 * tree-parloops.c: Likewise.
8558 * tree-profile.c: Likewise.
8559 * tree-ssa-ccp.c: Likewise.
8560 * tree-ssa-live.c: Likewise.
8561 * tree-ssa-loop.c: Likewise.
8562 * tree-ssa-sccvn.c: Likewise.
8563 * tree-ssa-structalias.c: Likewise.
8564 * tree-ssa.c: Likewise.
8565 * tree-streamer-in.c: Likewise.
8566 * tree-vectorizer.c: Likewise.
8567 * tree-vrp.c: Likewise.
8569 * ubsan.c: Likewise.
8570 * varasm.c: Likewise.
8571 * varpool.c: Likewise.
8572 * tree.c: Remove functions moved to attribs.[ch].
8574 * config/aarch64/aarch64.c: Add attrs.h header file.
8575 * config/alpha/alpha.c: Likewise.
8576 * config/arc/arc.c: Likewise.
8577 * config/arm/arm.c: Likewise.
8578 * config/avr/avr.c: Likewise.
8579 * config/bfin/bfin.c: Likewise.
8580 * config/c6x/c6x.c: Likewise.
8581 * config/cr16/cr16.c: Likewise.
8582 * config/cris/cris.c: Likewise.
8583 * config/darwin.c: Likewise.
8584 * config/epiphany/epiphany.c: Likewise.
8585 * config/fr30/fr30.c: Likewise.
8586 * config/frv/frv.c: Likewise.
8587 * config/ft32/ft32.c: Likewise.
8588 * config/h8300/h8300.c: Likewise.
8589 * config/i386/winnt.c: Likewise.
8590 * config/ia64/ia64.c: Likewise.
8591 * config/iq2000/iq2000.c: Likewise.
8592 * config/lm32/lm32.c: Likewise.
8593 * config/m32c/m32c.c: Likewise.
8594 * config/m32r/m32r.c: Likewise.
8595 * config/m68k/m68k.c: Likewise.
8596 * config/mcore/mcore.c: Likewise.
8597 * config/microblaze/microblaze.c: Likewise.
8598 * config/mips/mips.c: Likewise.
8599 * config/mmix/mmix.c: Likewise.
8600 * config/mn10300/mn10300.c: Likewise.
8601 * config/moxie/moxie.c: Likewise.
8602 * config/msp430/msp430.c: Likewise.
8603 * config/nds32/nds32-isr.c: Likewise.
8604 * config/nds32/nds32.c: Likewise.
8605 * config/nios2/nios2.c: Likewise.
8606 * config/nvptx/nvptx.c: Likewise.
8607 * config/pa/pa.c: Likewise.
8608 * config/pdp11/pdp11.c: Likewise.
8609 * config/powerpcspe/powerpcspe.c: Likewise.
8610 * config/riscv/riscv.c: Likewise.
8611 * config/rl78/rl78.c: Likewise.
8612 * config/rx/rx.c: Likewise.
8613 * config/s390/s390.c: Likewise.
8614 * config/sh/sh.c: Likewise.
8615 * config/sol2.c: Likewise.
8616 * config/sparc/sparc.c: Likewise.
8617 * config/spu/spu.c: Likewise.
8618 * config/stormy16/stormy16.c: Likewise.
8619 * config/tilegx/tilegx.c: Likewise.
8620 * config/tilepro/tilepro.c: Likewise.
8621 * config/v850/v850.c: Likewise.
8622 * config/vax/vax.c: Likewise.
8623 * config/visium/visium.c: Likewise.
8624 * config/xtensa/xtensa.c: Likewise.
8626 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8629 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8630 constraints since the -mupper-regs-* switches have been
8632 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8633 into a vector from a double word element that was extracted from
8634 another vector, and eliminate extra XXPERMDI instructions.
8635 (vsx_concat_<mode>_2): Likewise.
8636 (vsx_concat_<mode>_3): Likewise.
8637 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8638 concat to allow optimizing inserts from previous extracts.
8640 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8642 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8643 memory reference to a SSP offset in TLS address space.
8644 (ix86_print_operand) <case '@'>: Remove.
8645 (ix86_print_operand_punct_valid_p): Remove '@' code.
8646 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8648 (stack_tls_protect_set_<mode>): Remove.
8649 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8650 (stack_tls_protect_test_<mode>): Remove.
8651 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8653 2017-08-07 Olivier Hainque <hainque@adacore.com>
8656 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8658 2017-08-07 Douglas Rupp <rupp@adacore.com>
8660 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8661 variable was referenced as multidir in command.
8663 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8666 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8669 2017-08-07 Martin Liska <mliska@suse.cz>
8671 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8672 * config/rl78/rl78.c: Add include of attribs.h.
8673 * config/sh/sh.c: Likewise.
8674 * config/v850/v850.c: Likewise.
8676 2017-08-07 Tom de Vries <tom@codesourcery.com>
8679 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8681 2017-08-07 Martin Liska <mliska@suse.cz>
8683 * config/mips/mips.c: Include attribs.h.
8685 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8688 * doc/invoke.texi: Document change in behvaior for -Ofast for
8691 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8693 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8695 (aarch64_pop_regs): Likewise.
8696 (aarch64_gen_load_pair): Likewise.
8697 (aarch64_save_callee_saves): Likewise.
8698 (aarch64_restore_callee_saves): Likewise.
8700 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8702 * config/i386/i386.c: Revert the last change.
8704 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8707 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8709 (ix86_finalize_stack_frame_flags): This. Also clear
8710 frame_pointer_needed if -fno-omit-frame-pointer is used without
8712 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8713 with ix86_finalize_stack_frame_flags.
8714 (ix86_expand_epilogue): Likewise.
8715 (ix86_expand_split_stack_prologue): Likewise.
8717 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8720 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8721 to P_AES for Westmere.
8723 2017-08-07 Jonathan Yong <10walls@gmail.com>
8725 * config/i386/mingw.opt (fset-stack-executable): Removed.
8726 * config/i386/cygming.opt (fset-stack-executable): Moved
8728 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8730 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8732 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8734 2017-08-07 Marek Polacek <polacek@redhat.com>
8737 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8739 2017-08-07 Martin Liska <mliska@suse.cz>
8741 * attribs.h (canonicalize_attr_name): New function.
8742 (cmp_attribs): Move from c-format.c and adjusted.
8743 (is_attribute_p): Moved from tree.h.
8744 * tree-inline.c: Add new includes.
8745 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8746 (private_is_attribute_p): Remove.
8747 (private_lookup_attribute): Likewise.
8748 (private_lookup_attribute_by_prefix): Simplify.
8749 (remove_attribute): Use is_attribute_p.
8750 * tree.h: Remove removed declarations.
8752 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8755 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8756 instead of computing it in the function. Formatting fix.
8757 (expand_case): Don't rely on default_edge being the first edge,
8758 clear it if removing it, pass default_edge to
8759 emit_case_dispatch_table.
8760 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8763 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8765 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8766 insn in the function, emit NOP after the insn.
8768 2017-08-06 Tom de Vries <tom@codesourcery.com>
8770 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8773 2017-08-06 Tom de Vries <tom@codesourcery.com>
8775 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8778 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8780 PR tree-optimization/57371
8781 * match.pd: New pattern.
8783 2017-08-04 Marek Polacek <polacek@redhat.com>
8786 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8787 perform the computation in offset_int.
8789 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8791 PR tree-optimization/81136
8792 * tree-vectorizer.h: Include tree-hash-traits.h.
8793 (vec_base_alignments): New typedef.
8794 (vec_info): Add a base_alignments field.
8795 (vect_record_base_alignments): Declare.
8796 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8798 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8799 (create_data_ref): Add an is_conditional_in_stmt argument.
8800 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8801 the is_conditional_in_stmt field.
8802 (data_ref_loc): Add an is_conditional_in_stmt field.
8803 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8804 (find_data_references_in_stmt): Update call to create_data_ref.
8805 (graphite_find_data_references_in_stmt): Likewise.
8806 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8807 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8808 (vect_record_base_alignment): New function.
8809 (vect_record_base_alignments): Likewise.
8810 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8811 for nested statements even if we fail to compute a misalignment.
8812 Use pooled base alignments for unconditional references.
8813 (vect_find_same_alignment_drs): Compare base addresses instead
8815 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8816 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8818 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8820 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8821 Add an explicit name for the enum. Use auto_vec for slp_instances
8823 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8825 (_bb_vec_info): Add a constructor and destructor.
8826 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8827 (destroy_loop_vec_info): Delete.
8828 (vect_destroy_datarefs): Likewise.
8829 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8830 (vec_info::vec_info): New function.
8831 (vec_info::~vec_info): Likewise.
8832 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8833 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8834 destroy_loop_vec_info.
8835 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8836 (_loop_vec_info::_loop_vec_info): ...this.
8837 (destroy_loop_vec_info): Replace with...
8838 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8839 the stmt_vec_infos. Leave handling of vec_info information to its
8840 destructor. Remove explicit vector releases.
8841 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8842 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8843 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8844 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8845 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8846 (destroy_bb_vec_info): Replace with...
8847 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8848 information to its destructor.
8849 (vect_slp_analyze_bb_1): Use new and delete instead of
8850 new_bb_vec_info and destroy_bb_vec_info.
8851 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8854 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8856 * tree-data-ref.h (subscript): Add access_fn field.
8857 (data_dependence_relation): Add could_be_independent_p.
8858 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8859 (same_access_functions): Move to tree-data-ref.c.
8860 * tree-data-ref.c (ref_contains_union_access_p): New function.
8861 (access_fn_component_p): Likewise.
8862 (access_fn_components_comparable_p): Likewise.
8863 (dr_analyze_indices): Add a reference to access_fn_component_p.
8864 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8866 (constant_access_functions): Likewise.
8867 (add_other_self_distances): Likewise.
8868 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8869 (initialize_data_dependence_relation): Use XCNEW and remove
8870 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8871 of access functions that have the same type. Allow the
8872 subsequence to end with different bases in some circumstances.
8873 Record the chosen access functions in SUB_ACCESS_FN.
8874 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8875 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8876 (subscript_dependence_tester_1): Likewise dra and drb.
8877 (build_classic_dist_vector): Update calls accordingly.
8878 (subscript_dependence_tester): Likewise.
8879 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8880 DDR_COULD_BE_INDEPENDENT_P.
8881 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8882 comp_alias_ddrs instead of may_alias_ddrs.
8883 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8885 (vect_analyze_data_ref_dependence): Use it if
8886 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8887 distance vectors if that fails.
8888 (dependence_distance_ge_vf): New function.
8889 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8890 LOOP_VINFO_MAY_ALIAS_DDRS.
8892 2017-08-04 Richard Biener <rguenther@suse.de>
8895 * fold-const.c (fold_binary_loc): Properly restrict
8896 minus_var0 && minus_var1 case when associating undefined overflow
8899 2017-08-04 Tom de Vries <tom@codesourcery.com>
8901 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8903 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8905 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8906 Don't start diagnostic messages with a capital letter.
8907 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8909 (rs6000_invalid_builtin): Likewise.
8910 (rs6000_trampoline_init): Likewise.
8912 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8915 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8916 after setting changeable df flags.
8918 2017-08-03 Richard Biener <rguenther@suse.de>
8920 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8921 up if the use is in USE - X.
8923 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8925 * toplev.c (dumpfile.h): New include.
8926 (internal_error_reentered): New static function. Use it...
8927 (internal_error_function): ...here to handle reentered internal_error.
8929 2017-08-03 Richard Biener <rguenther@suse.de>
8932 * fold-const.c (split_tree): Add minus_var and minus_con
8933 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8934 here but always use minus_*.
8935 (associate_trees): Assert we never associate with MINUS_EXPR
8936 and NULL first operand. Do not recurse for PLUS_EXPR operands
8937 when associating as MINUS_EXPR either.
8938 (fold_binary_loc): Track minus_var and minus_con.
8940 2017-08-03 Tom de Vries <tom@codesourcery.com>
8943 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8944 ACCEL_COMPILER, apply finish_options on
8945 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8947 2017-08-03 Tom de Vries <tom@codesourcery.com>
8950 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8951 function_entry_patch_area_size > 0.
8953 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8956 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8957 instead of 10??LU, perform unit multiplication in wide_int,
8958 don't change alloc_object_size_limit if the limit is larger
8961 PR tree-optimization/81655
8962 PR tree-optimization/81588
8963 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8964 the case when ranges[i].low and high are 1 for unsigned type with
8968 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8969 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8971 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8973 * tree-vrp.h: Add include guard.
8975 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8978 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8979 (ud2): New insn pattern.
8980 * config/i386/i386.c (ix86_expand_epilogue):
8981 For naked functions, generate ud2 instead of trap insn.
8983 2017-08-02 Marek Polacek <polacek@redhat.com>
8986 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8988 2017-08-02 Tom de Vries <tom@codesourcery.com>
8989 Cesar Philippidis <cesar@codesourcery.com>
8991 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8992 Add missing edge probabilities.
8994 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8996 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8999 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9002 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9003 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9004 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9005 prologue_seq sequences - if any.
9007 2017-08-02 Richard Biener <rguenther@suse.de>
9009 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9010 via vectors if supported, integer extracts via punning if supported
9011 or otherwise vector extracts.
9013 2017-08-02 Richard Biener <rguenther@suse.de>
9015 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9017 (bitmap_insert_into_set): ... this.
9019 2017-08-02 Richard Biener <rguenther@suse.de>
9021 PR tree-optimization/81633
9023 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9025 PR tree-optimization/71752
9026 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9028 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9030 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9031 (machine_function::call_ms2sysv_pad_out): Remove field.
9032 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9033 (ix86_compute_frame_layout): Likewise.
9035 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9038 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9039 attribute with interrupt attribute.
9041 2017-08-01 Andrew Pinski <apinski@cavium.com>
9043 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9044 BIT_INSERT_EXPR's operand 1
9045 to see if the types precision matches.
9047 2017-08-01 Martin Liska <mliska@suse.cz>
9050 * builtins.c (expand_builtin_memcpy_args): Remove.
9051 (expand_builtin_memcpy): Call newly added function
9052 expand_builtin_memory_copy_args.
9053 (expand_builtin_memcpy_with_bounds): Likewise.
9054 (expand_builtin_mempcpy): Remove last argument.
9055 (expand_builtin_mempcpy_with_bounds): Likewise.
9056 (expand_builtin_memory_copy_args): New function created from
9057 expand_builtin_mempcpy_args with small modifications.
9058 (expand_builtin_mempcpy_args): Remove.
9059 (expand_builtin_stpcpy): Remove unused argument.
9060 (expand_builtin): Likewise.
9061 (expand_builtin_with_bounds): Likewise.
9063 2017-08-01 Martin Liska <mliska@suse.cz>
9066 Make mempcpy more optimal (PR middle-end/70140).
9068 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9071 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9072 __builtin_vec_cmpne verify both arguments are compatible vectors
9073 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9074 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9075 move computation of aligned to after checking the argument types.
9079 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9080 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9083 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9084 Doug Rupp <rupp@adacore.com>
9085 Olivier Hainque <hainque@adacore.com>
9087 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9088 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9090 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9091 for TARGET_OS_CPP_BUILTIN.
9092 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9093 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9095 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9096 passing required abi options to the assembler for EABI configurations.
9097 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9098 of .text.hot and .text.unlikely sections for kernel modules when
9099 using ARM style exceptions.
9100 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9101 options. Add EXTRA_CC1_SPEC.
9102 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9104 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9106 (ARM_TARGET2_DWARF_FORMAT): Define.
9107 * config/arm/t-vxworks: Adjust multilib control to removal of the
9108 Diab command line options.
9110 2017-08-01 Martin Liska <mliska@suse.cz>
9112 PR gcov-profile/81561
9113 * gcov.c (unblock): Make unblocking safe as we need to preserve
9114 index correspondence of blocks and block_lists.
9116 2017-08-01 Richard Biener <rguenther@suse.de>
9118 PR tree-optimization/81181
9119 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9120 (compute_antic): ... end of iteration here.
9122 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9124 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9125 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9126 (ftree-slp-vectorize): Likewise.
9127 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9128 can no longer be set independent of flag_tree_loop_vectorize.
9129 * omp-general.c (emp_max_vf): Likewise.
9130 * opts.c (enable_fdo_optimizations): Remove references to
9131 flag_tree_vectorize, these are now implicit.
9132 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9133 and leave it for the options machinery.
9135 2017-08-01 Martin Liska <mliska@suse.cz>
9138 * builtins.c (expand_builtin_memcpy_args): Remove.
9139 (expand_builtin_memcpy): Call newly added function
9140 expand_builtin_memory_copy_args.
9141 (expand_builtin_memcpy_with_bounds): Likewise.
9142 (expand_builtin_mempcpy): Remove last argument.
9143 (expand_builtin_mempcpy_with_bounds): Likewise.
9144 (expand_builtin_memory_copy_args): New function created from
9145 expand_builtin_mempcpy_args with small modifications.
9146 (expand_builtin_mempcpy_args): Remove.
9147 (expand_builtin_stpcpy): Remove unused argument.
9148 (expand_builtin): Likewise.
9149 (expand_builtin_with_bounds): Likewise.
9151 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9154 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9155 print "ds:" only for immediates in generic address space.
9157 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9160 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9161 (ix86_function_ok_for_sibcall): Return false for naked functions.
9163 2017-08-01 Richard Biener <rguenther@suse.de>
9165 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9166 (compute_antic): Seed worklist with exit block predecessors.
9167 * cfganal.c (dfs_find_deadend): For a cycle return the source
9168 of the edge closing it.
9170 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9172 * config/aarch64/aarch64.c
9173 (aarch64_can_const_movi_rtx_p): Move 0 check.
9175 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9177 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9178 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9180 * match.pd: Ditto in address comparison pattern.
9182 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9184 PR tree-optimization/81627
9185 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9186 closed ssa form for store-store chain.
9188 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9190 PR tree-optimization/81620
9191 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9192 for store-store chain.
9194 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9196 PR tree-optimization/81588
9197 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9198 ranges[i].in_p, invert comparison code ccode. For >/>=,
9199 swap rhs1 and rhs2 and comparison code unconditionally,
9200 for </<= don't do that. Don't swap rhs1/rhs2 again if
9201 ranges[i].in_p, instead invert comparison code ccode if
9202 opcode or oe->rank is BIT_IOR_EXPR.
9205 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9206 a direct optab to conversion optab.
9207 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9208 with GET_MODE_INNER as last argument instead of optab_handler.
9209 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9210 vector extraction if possible and optab is available.
9211 * expr.c (store_constructor): Use convert_optab_handler instead
9212 of optab_handler. Use vector initialization from smaller
9213 vectors if possible and optab is available.
9214 * tree-vect-stmts.c (vectorizable_load): Likewise.
9215 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9217 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9218 of vec_init from half-sized vectors with the same element mode.
9219 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9220 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9221 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9222 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9223 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9224 after mode in gen_vec_extract* calls.
9225 (vec_extract<mode>): Renamed to ...
9226 (vec_extract<mode><ssescalarmodelower>): ... this.
9227 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9228 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9229 element mode after mode in gen_vec_init* calls.
9230 (VEC_INIT_HALF_MODE): New mode iterator.
9231 (vec_init<mode>): Renamed to ...
9232 (vec_init<mode><ssescalarmodelower>): ... this.
9233 (vec_init<mode><ssehalfvecmodelower>): New expander.
9234 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9235 (vec_extractv2sfsf): ... this.
9236 (vec_initv2sf): Renamed to ...
9237 (vec_initv2sfsf): ... this.
9238 (vec_extractv2si): Renamed to ...
9239 (vec_extractv2sisi): ... this.
9240 (vec_initv2si): Renamed to ...
9241 (vec_initv2sisi): ... this.
9242 (vec_extractv4hi): Renamed to ...
9243 (vec_extractv4hihi): ... this.
9244 (vec_initv4hi): Renamed to ...
9245 (vec_initv4hihi): ... this.
9246 (vec_extractv8qi): Renamed to ...
9247 (vec_extractv8qiqi): ... this.
9248 (vec_initv8qi): Renamed to ...
9249 (vec_initv8qiqi): ... this.
9250 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9251 (vec_init<mode>): Renamed to ...
9252 (vec_init<mode><VEC_base_l>): ... this.
9253 (vec_extract<mode>): Renamed to ...
9254 (vec_extract<mode><VEC_base_l>): ... this.
9255 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9256 (vec_initv2sfsf): ... this.
9257 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9258 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9259 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9260 element mode after mode in gen_vec_init* calls.
9261 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9262 (vec_init<mode><Vel>): ... this.
9263 (vec_extract<mode>): Renamed to ...
9264 (vec_extract<mode><Vel>): ... this.
9265 * config/aarch64/iterators.md (Vel): New mode attribute.
9266 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9267 Add element mode after mode in gen_vec_extract* calls.
9268 * config/s390/vector.md (non_vec_l): New mode attribute.
9269 (vec_extract<mode>): Renamed to ...
9270 (vec_extract<mode><non_vec_l>): ... this.
9271 (vec_init<mode>): Renamed to ...
9272 (vec_init<mode><non_vec_l>): ... this.
9273 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9274 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9276 * config/arm/iterators.md (V_elem_l): New mode attribute.
9277 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9278 (vec_extract<mode><V_elem_l>): ... this.
9279 (vec_extractv2di): Renamed to ...
9280 (vec_extractv2didi): ... this.
9281 (vec_init<mode>): Renamed to ...
9282 (vec_init<mode><V_elem_l>): ... this.
9283 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9284 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9285 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9286 Add element mode after gen_vec_extract* calls.
9287 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9288 (vec_init<mode><unitmode>): ... this.
9289 (vec_extract<mode>): Renamed to ...
9290 (vec_extract<mode><unitmode>): ... this.
9291 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9292 (vec_init<mode><unitmode>): ... this.
9293 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9294 (vec_initv2sfsf): ... this.
9295 (vec_extractv2sf): Renamed to ...
9296 (vec_extractv2sfsf): ... this.
9297 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9298 Add element mode after gen_vec_extract* calls.
9299 * config/mips/mips.md (unitmode): New mode iterator.
9300 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9301 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9302 * config/spu/spu.md (inner_l): New mode attribute.
9303 (vec_init<mode>): Renamed to ...
9304 (vec_init<mode><inner_l>): ... this.
9305 (vec_extract<mode>): Renamed to ...
9306 (vec_extract<mode><inner_l>): ... this.
9307 * config/sparc/sparc.md (veltmode): New mode iterator.
9308 (vec_init<VMALL:mode>): Renamed to ...
9309 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9310 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9311 (vec_initv2sisi): ... this.
9312 (vec_initv2sf): Renamed to ...
9313 (vec_initv2sfsf): ... this.
9314 (vec_extractv2sf): Renamed to ...
9315 (vec_extractv2sfsf): ... this.
9316 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9317 (vec_init<mode>): Renamed to ...
9318 (vec_init<mode><VEC_base_l>): ... this.
9319 (vec_extract<mode>): Renamed to ...
9320 (vec_extract<mode><VEC_base_l>): ... this.
9321 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9322 (vec_initv2sfsf): ... this.
9323 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9324 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9325 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9326 gen_vec_init* calls.
9328 2017-08-01 Richard Biener <rguenther@suse.de>
9330 PR tree-optimization/81297
9331 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9332 TREE_OVERFLOW from INTEGER_CSTs.
9334 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9336 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9338 2017-07-31 Carl Love <cel@us.ibm.com>
9340 * config/rs6000/rs6000-c: Add support for built-in functions
9341 vector signed char vec_xl_be (signed long long, signed char *);
9342 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9343 vector signed int vec_xl_be (signed long long, signed int *);
9344 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9345 vector signed long long vec_xl_be (signed long long, signed long long *);
9346 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9347 vector signed short vec_xl_be (signed long long, signed short *);
9348 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9349 vector double vec_xl_be (signed long long, double *);
9350 vector float vec_xl_be (signed long long, float *);
9351 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9352 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9353 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9355 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9356 (altivec_expand_builtin): Add switch statement to call
9357 altivec_expand_xl_be for each builtin.
9358 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9359 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9360 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9361 __builtin_vsx_le_be_v16qi.
9362 * doc/extend.texi: Update the built-in documentation file for the
9363 new built-in functions.
9365 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9368 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9370 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9372 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9374 * config.gcc: Add z14.
9375 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9376 CPU model numbers for z13s and z14.
9377 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9379 * config/s390/s390-opts.h (enum processor_type): Rename
9380 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9381 * config/s390/s390.c (processor_table): Add field for CPU name to
9382 be passed to Binutils.
9383 (s390_asm_output_machine_for_arch): Use the new field in
9384 processor_table for Binutils.
9385 (s390_expand_builtin): Replace arch12 with z14.
9386 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9387 (s390_get_sched_attrmask): Likewise.
9388 (s390_get_unit_mask): Likewise.
9389 * config/s390/s390.opt: Add z14 to processor_type enum.
9391 2017-07-31 Martin Jambor <mjambor@suse.cz>
9394 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9395 regardless of optimization level.
9397 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9398 Martin Liska <mliska@suse.cz>
9400 * predict.def: Remove old comment and adjust probability.
9401 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9404 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9407 * config/i386/i386.c (ix86_function_naked): New function.
9408 (ix86_can_use_return_insn_p): Return false for naked functions.
9409 (ix86_expand_prologue): Skip prologue for naked functions.
9410 (ix86_expand_epilogue): Skip epilogue for naked functions
9411 and emit trap instruction.
9412 (ix86_warn_func_return): New function.
9413 (ix86_attribute_table): Add "naked" attribute specification.
9414 (TARGET_WARN_FUNC_RETURN): Define.
9415 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9417 2017-07-31 Martin Liska <mliska@suse.cz>
9419 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9420 (dump_gimple_bb_header): Always dump BB info.
9421 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9423 2017-07-31 Martin Liska <mliska@suse.cz>
9426 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9427 also with current_function_decl non-null equality.
9429 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9432 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9433 change type to the element type, instead add eltype variable and
9434 use it where we are interested in the element type.
9436 PR tree-optimization/81603
9437 * ipa-polymorphic-call.c
9438 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9439 offset arithmetic in offset_int, bail out if the resulting bit offset
9440 doesn't fit into shwi.
9442 2017-07-31 Martin Liska <mliska@suse.cz>
9444 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9445 (gimplify_save_expr): Fix comment.
9447 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9450 * config/i386/i386.c (ix86_function_arg): Update arguments for
9452 (ix86_compute_frame_layout): Set the initial stack offset to
9453 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9454 INCOMING_FRAME_SP_OFFSET.
9455 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9456 stack before exception handler returns.
9457 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9458 the 'ERROR_CODE' for exception handler.
9460 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9462 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9463 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9464 (ASM_OUTPUT_REG_POP): Ditto.
9465 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9466 instead of asm_fprintf to output pure string.
9468 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9470 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9471 to imported_module_or_decl hook.
9472 (debug_nothing_tree_tree_tree_bool): Remove.
9473 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9474 * debug.c (do_nothing_debug_hooks): Use
9475 debug_nothing_tree_tree_tree_bool_bool instead of
9476 debug_nothing_tree_tree_tree_bool.
9477 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9478 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9479 * sdbout.c (sdb_debug_hooks): Likewise.
9480 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9481 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9483 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9484 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9485 attribute, don't add anything.
9487 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9489 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9490 (fold_build2_stat_loc): Likewise.
9491 (fold_build3_stat_loc): Likewise.
9492 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9493 (fold_build1_loc): Remove macro.
9494 (fold_build2_loc): Likewise.
9495 (fold_build3_loc): Likewise.
9497 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9499 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9500 (gimple_build_debug_bind_source_stat): Likewise.
9501 * gimple.h (gimple_build_debug_bind): Remove macro.
9502 (gimple_build_debug_bind_source): Likewise.
9504 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9506 * bitmap.c (bitmap_alloc): Adjust.
9507 (bitmap_gc_alloc): Likewise.
9508 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9510 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9512 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9513 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9514 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9515 (bitmap_gc_alloc_stat): Likewise.
9516 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9520 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9521 * rtl.h (shallow_copy_rtx): Remove macro.
9523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9525 * emit-rtl.c (gen_raw_REG): Adjust.
9526 * gengenrtl.c (gendef): Likewise.
9527 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9528 * rtl.h (rtx_alloc): Remove macro.
9530 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9532 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9533 (build_tree_list_stat): Likewise.
9534 * tree.h (build_tree_list): Remove macro.
9535 (build_tree_list_vec): Likewise.
9537 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9539 * tree.c (make_vector_stat): Remove _stat from name.
9540 (build_vector_stat): Likewise.
9541 * tree.h (make_vector_stat): Remove macro.
9542 (build_vector_stat): Likewise.
9544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9546 * tree.h (build_var_debug_value): Remove prototype.
9548 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9550 * tree.c (tree_cons_stat): Remove _stat from name.
9551 * tree.h (tree_cons): Remove macro.
9553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9555 * tree.c (build_vl_exp_stat): Remove _stat from name.
9556 * tree.h (build_vl_exp): Remove macro.
9558 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9560 * tree.c (build_decl_stat): Remove _stat from name.
9561 * tree.h (build_decl): Remove macro.
9563 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9565 * gimple.c (gimple_build_with_ops_stat): Adjust.
9566 (gimple_alloc_stat): Remove _stat from name.
9567 * gimple.h (gimple_alloc): Remove macro.
9569 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9571 * tree.c (make_tree_vec_stat): Remove _stat from name.
9572 (grow_tree_vec_stat): Likewise.
9573 * tree.h (make_tree_vec_stat): Adjust prototype.
9574 (grow_tree_vec_stat): Likewise.
9575 (make_tree_vec): Remove macro.
9576 (grow_tree_vec): Likewise.
9578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9580 * fold-const.c (fold_build1_stat_loc): Adjust.
9581 (fold_build2_stat_loc): Likewise.
9582 (fold_build3_stat_loc): Likewise.
9583 * tree.c (build0_stat): Remove _stat from name.
9584 (build1_stat): Likewise.
9585 (build2_stat): Likewise.
9586 (build3_stat): Likewise.
9587 (build4_stat): Likewise.
9588 (build5_stat): Likewise.
9589 * tree.h (build1_loc): Remove macro, and rename _stat function
9591 (build2_loc): Likewise.
9592 (build3_loc): Likewise.
9593 (build4_loc): Likewise.
9594 (build5_loc): Likewise.
9596 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9598 * tree.c (make_int_cst_stat): Remove _stat from name.
9599 * tree.h (make_int_cst_stat): Adjust prototype.
9600 (make_int_cst): Remove macro.
9602 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9604 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9605 * tree.h (make_tree_binfo_stat): Adjust prototype.
9606 (make_tree_binfo): Remove.
9608 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9610 * tree.c (copy_node_stat): Rename to copy_node.
9611 (build_distinct_type_copy): Adjust.
9612 * tree.h (copy_node_stat): Adjust prototype.
9613 (copy_node): Remove macro.
9615 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9617 * tree.c (make_node_stat): rename to make_node.
9618 (build_tree_list_stat): Adjust.
9619 (build0_stat): Likewise.
9620 (build2_stat): Likewise.
9621 (build3_stat): Likewise.
9622 (build4_stat): Likewise.
9623 (build5_stat): Likewise.
9624 (build_decl_stat): Likewise.
9625 * tree.h (make_node_stat): Adjust prototype.
9626 (make_node): remove macro.
9628 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9630 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9631 (PPC_FEATURE2_SCV): Likewise.
9632 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9634 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9636 * config/aarch64/aarch64.c
9637 (aarch64_internal_mov_immediate): Add new special pattern.
9638 * config/aarch64/aarch64.md (*movdi_aarch64):
9639 Add reg/32bit const mov case.
9641 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9642 Richard Sandiford <richard.sandiford@linaro.org>
9644 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9645 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9646 Add integer and movi cases.
9647 (movi-split-hf-df-sf split, fp16): New.
9648 (enabled): Added TARGET_FP_F16INST.
9649 * config/aarch64/iterators.md (GPF_HF): New.
9650 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9652 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9654 * config/aarch64/aarch64.c
9655 (aarch64_simd_container_mode): Add prototype.
9656 (aarch64_expand_mov_immediate): Add HI support.
9657 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9658 (aarch64_can_const_movi_rtx_p): New.
9659 (aarch64_preferred_reload_class):
9660 Remove restrictions of using FP registers for certain SIMD operations.
9661 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9662 (aarch64_valid_floating_const): Add integer move validation.
9663 (aarch64_simd_imm_scalar_p): Remove.
9664 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9665 (aarch64_legitimate_constant_p): Expand list of supported cases.
9666 * config/aarch64/aarch64-protos.h
9667 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9668 (aarch64_reinterpret_float_as_int): New.
9669 (aarch64_simd_imm_scalar_p): Remove.
9670 * config/aarch64/constraints.md (Uvi): New.
9671 (Dd): Split into Ds and new Dd.
9672 * config/aarch64/aarch64.md (*movsi_aarch64):
9674 (*movdi_aarch64): Add SIMD mov case.
9676 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9678 * tree-predcom.c: (struct chain): Handle store-store chain in which
9679 stores for elimination only store loop invariant values.
9680 (execute_pred_commoning_chain): Ditto.
9681 (prepare_initializers_chain_store_elim): Ditto.
9682 (prepare_finalizers): Ditto.
9683 (is_inv_store_elimination_chain): New function.
9684 (initialize_root_vars_store_elim_1): New function.
9686 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9688 * tree-predcom.c: Revise general description of the pass.
9689 (enum chain_type): New enum type for store elimination.
9690 (struct chain): New field supporting store elimination.
9691 (struct component): Ditto.
9692 (dump_chain): Dump store-stores chain.
9693 (release_chain): Release resources.
9694 (split_data_refs_to_components): Compute and create component
9695 contains only stores for elimination.
9696 (get_chain_last_ref_at): New function.
9697 (make_invariant_chain): Initialization.
9698 (make_rooted_chain): Specify chain type in parameter and record it.
9699 (add_looparound_copies): Skip for store-stores chain.
9700 (determine_roots_comp): Compute type of chain and pass it to
9702 (initialize_root_vars_store_elim_2): New function.
9703 (finalize_eliminated_stores): New function.
9704 (remove_stmt): Handle store for elimination.
9705 (execute_pred_commoning_chain): Execute predictive commoning on
9707 (determine_unroll_factor): Skip unroll for store-stores chain.
9708 (prepare_initializers_chain_store_elim): New function.
9709 (prepare_initializers_chain): Hanlde store-store chain.
9710 (prepare_finalizers_chain, prepare_finalizers): New function.
9711 (tree_predictive_commoning_loop): Return integer value indicating
9712 if loop is unrolled or lcssa form is corrupted.
9713 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9715 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9717 * tree-predcom.c (initialize_root): Delete.
9718 (execute_pred_commoning_chain): Initialize root vars and replace
9719 reference of non-combined chain directly, rather than call above
9722 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9724 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9725 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9727 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9729 * tree-predcom.c (struct chain): New field init_seq.
9730 (release_chain): Release init_seq.
9731 (prepare_initializers_chain): Record intialization stmts in above
9733 (insert_init_seqs): New function.
9734 (tree_predictive_commoning_loop): Call insert_init_seqs.
9736 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9738 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9740 2017-07-28 Richard Biener <rguenther@suse.de>
9742 * match.pd: Remove superfluous :c.
9743 * genmatch.c (simplify::id): Add member.
9744 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9746 (current_id): New global.
9747 (dt_node::parent): Move from ...
9748 (dt_operand::parent): ... here. Add for_id member.
9749 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9750 (decision_tree::find_node): Relax order requirement when
9751 merging DT_TRUE nodes to ones inbetween the current simplify
9752 and the one we try to merge with. Add diagnostic whenever
9753 we need to enforce pattern order by not merging.
9754 (decision_tree::insert): Set current_id.
9755 (decision_tree::print_node): Dump parent node and for_id.
9756 (parser::last_id): Add member.
9757 (parser::push_simplify): Assign unique id.
9758 (parser::parser): Initialize last_id.
9760 2017-07-28 Martin Liska <mliska@suse.cz>
9763 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9764 gimple_build_debug_bind.
9766 2017-07-28 Richard Biener <rguenther@suse.de>
9768 PR tree-optimization/81502
9769 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9771 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9773 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9774 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9775 for BIT_FIELD_REF args.
9776 * fold-const.c (make_bit_field_ref): Likewise.
9777 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9779 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9782 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9783 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9784 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9785 Or it into SANITIZER_UNDEFINED.
9786 * ubsan.c: Include gimple-fold.h and varasm.h.
9787 (ubsan_expand_ptr_ifn): New function.
9788 (instrument_pointer_overflow): New function.
9789 (maybe_instrument_pointer_overflow): New function.
9790 (instrument_object_size): Formatting fix.
9791 (pass_ubsan::execute): Call instrument_pointer_overflow
9792 and maybe_instrument_pointer_overflow.
9793 * internal-fn.c (expand_UBSAN_PTR): New function.
9794 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9795 * sanitizer.def (__ubsan_handle_pointer_overflow,
9796 __ubsan_handle_pointer_overflow_abort): New builtins.
9797 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9798 * internal-fn.def (UBSAN_PTR): New internal function.
9799 * opts.c (sanitizer_opts): Add pointer-overflow.
9800 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9801 * fold-const.c (build_range_check): Compute pointer range check in
9802 integral type if pointer arithmetics would be needed. Formatting
9805 2017-07-28 Martin Liska <mliska@suse.cz>
9808 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9809 parameters that are of a variable-length.
9811 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9813 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9815 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9816 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9817 (CRT_CALL_STATIC_FUNCTION): Likewise.
9818 (ASM_DEFAULT_SPEC): New define.
9819 (ASM_SPEC32): Likewise.
9820 (ASM_SPEC64): Likewise.
9821 (ASM_SPEC_COMMON): Likewise.
9822 (ASM_SPEC): Likewise.
9823 (INVALID_64BIT): Likewise.
9824 (LINK_OS_DEFAULT_SPEC): Likewise.
9825 (LINK_OS_SPEC32): Likewise.
9826 (LINK_OS_SPEC64): Likewise.
9827 (POWERPC_LINUX): Likewise.
9828 (PTRDIFF_TYPE): Likewise.
9829 (RESTORE_FP_PREFIX): Likewise.
9830 (RESTORE_FP_SUFFIX): Likewise.
9831 (SAVE_FP_PREFIX): Likewise.
9832 (SAVE_FP_SUFFIX): Likewise.
9833 (SIZE_TYPE): Likewise.
9834 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9835 (TARGET_64BIT): Likewise.
9836 (TARGET_64BIT): Likewise.
9837 (TARGET_AIX): Likewise.
9838 (WCHAR_TYPE_SIZE): Likewise.
9839 (WCHAR_TYPE): Undefine.
9840 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9841 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9842 (CPP_OS_RTEMS_SPEC): Delete.
9843 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9844 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9846 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9848 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9850 PR tree-optimization/81578
9851 * tree-parloops.c (build_new_reduction): Bail out if
9852 reduction_code isn't one of the standard OpenMP reductions.
9853 Move the details printing after that decision.
9855 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9857 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9858 related to reload_in_progress.
9859 (splat_input_operand): Likewise.
9860 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9862 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9864 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9865 (TARGET_INSTANTIATE_DECLS): Likewise.
9866 (legitimate_indexed_address_p): Delete reload_in_progress code.
9867 (rs6000_debug_legitimate_address_p): Likewise.
9868 (rs6000_eliminate_indexed_memrefs): Likewise.
9869 (rs6000_emit_le_vsx_store): Likewise.
9870 (rs6000_emit_move_si_sf_subreg): Likewise.
9871 (rs6000_emit_move): Likewise.
9872 (register_to_reg_type): Likewise.
9873 (rs6000_pre_atomic_barrier): Likewise.
9874 (rs6000_machopic_legitimize_pic_address): Likewise.
9875 (rs6000_allocate_stack_temp): Likewise.
9876 (rs6000_address_for_fpconvert): Likewise.
9877 (rs6000_address_for_altivec): Likewise.
9878 (rs6000_secondary_memory_needed_rtx): Delete function.
9879 (rs6000_check_sdmode): Likewise.
9880 (rs6000_alloc_sdmode_stack_slot): Likewise.
9881 (rs6000_instantiate_decls): Likewise.
9882 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9883 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9884 Delete reload_in_progress.
9885 (*vec_reload_and_plus_<mptrsize>): Likewise.
9886 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9887 (vsx_div_v2di): Likewise.
9888 (vsx_udiv_v2di): Likewise.
9890 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9892 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9893 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9894 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9895 (rs6000_debug_reg_global): Delete print of LRA status.
9896 (rs6000_option_override_internal): Delete dead LRA related code.
9897 (rs6000_lra_p): Delete function.
9898 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9900 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9902 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9903 * config/riscv/rtems.h: New file.
9905 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9906 Sudakshina Das <sudi.das@arm.com>
9908 * config/aarch64/aarch64.md
9909 (define_split for and<mode>3nr_compare): Move
9910 non aarch64_logical_operand to a register.
9911 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9912 register immediate operand to a register.
9913 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9915 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9918 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9920 2017-07-27 Richard Biener <rguenther@suse.de>
9922 PR tree-optimization/81573
9923 PR tree-optimization/81494
9924 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9925 multi defuse cycle case.
9927 2017-07-27 Richard Biener <rguenther@suse.de>
9929 PR tree-optimization/81571
9930 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9933 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9935 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9936 earlier and only if MASK_FPU is set. Adjust formatting.
9938 2017-07-27 Martin Liska <mliska@suse.cz>
9940 * opt-functions.awk: Add validation of value of Init.
9941 * optc-gen.awk: Pass new argument.
9943 2017-07-27 Martin Liska <mliska@suse.cz>
9945 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9946 Fix wrong condition.
9948 2017-07-27 Martin Liska <mliska@suse.cz>
9950 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9951 BBs and edges seen by autoFDO.
9953 2017-07-27 Richard Biener <rguenther@suse.de>
9955 PR tree-optimization/81502
9956 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9957 with incompatible but same sized type.
9958 (execute_update_addresses_taken): Likewise.
9960 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9962 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9963 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9965 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9968 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9969 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9970 Change s_operand to memory_operand.
9972 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9974 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9975 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9976 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9977 Emit instructions rather than returning an expression. Handle TFmode
9978 and KFmode by casting to TImode.
9979 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9980 (rs6000_emit_le_vsx_store): Likewise.
9981 * config/rs6000/vsx.md (VSX_TI): New iterator.
9982 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9983 (*vsx_le_undo_permute_<mode>): Likewise.
9984 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9985 emit the split sequence.
9986 (*vsx_le_perm_store_<mode>): Likewise.
9988 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9990 PR tree-optimization/81555
9991 PR tree-optimization/81556
9992 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9993 if true, force CHANGED for the recursive invocation.
9994 (reassociate_bb): Remember original length of ops array, pass
9995 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9997 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9998 attributes for noipa attribute. For naked attribute use
9999 lookup_attribute first before lookup_attribute_spec.
10000 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10002 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10003 for functions with noipa attribute.
10004 (cgraph_externally_visible_p): Return true for functions with noipa
10006 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10007 for functions with noipa attribute.
10008 * doc/extend.texi: Document noipa function attribute.
10009 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10010 also for functions with noipa attribute.
10011 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10013 2017-07-26 Andrew Pinski <apinski@cavium.com>
10015 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10016 vec_unalign_load_cost and vec_unalign_store_cost.
10018 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10020 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10021 -mvsx-small-integer option.
10022 (ISA_3_0_MASKS_IEEE): Likewise.
10023 (OTHER_VSX_VECTOR_MASKS): Likewise.
10024 (POWERPC_MASKS): Likewise.
10025 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10026 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10027 code, only testing for DImode being allowed in non-VSX floating
10029 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10030 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10032 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10033 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10034 TARGET_P8_VECTOR test.
10035 (rs6000_secondary_reload_simple_move): Likewise.
10036 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10037 since TARGET_P9_VECTOR was already tested.
10038 (rs6000_opt_masks): Remove -mvsx-small-integer.
10039 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10040 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10042 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10043 test for TARGET_VEXTRACTUB was used, and that uses
10045 (p9 extract splitter): Likewise.
10046 (vsx_extract_<mode>_di_p9): Likewise.
10047 (vsx_extract_<mode>_store_p9): Likewise.
10048 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10049 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10050 the elimination of TARGET_VSX_SMALL_INTEGER.
10051 (vsx_extract_<mode>_p8): Likewise.
10052 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10053 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10054 (vsx_set_<mode>_p9): Likewise.
10055 (vsx_set_v4sf_p9): Likewise.
10056 (vsx_set_v4sf_p9_zero): Likewise.
10057 (vsx_insert_extract_v4sf_p9): Likewise.
10058 (vsx_insert_extract_v4sf_p9_2): Likewise.
10059 * config/rs6000/rs6000.md (sign extend splitter): Change
10060 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10061 (floatsi<mode>2_lfiwax_mem): Likewise.
10062 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10063 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10064 since a test for TARGET_P9_VECTOR was used.
10065 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10066 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10067 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10068 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10069 TARGET_P8_VECTOR test.
10070 (fix_trunc<mode>si2_stfiwx): Likewise.
10071 (fix_trunc<mode>si2_internal): Likewise.
10072 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10073 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10075 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10076 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10077 TARGET_P8_VECTOR test.
10078 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10079 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10080 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10082 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10083 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10084 since a test for TARGET_P9_VECTOR was used.
10085 (splitter for loading small constants): Likewise.
10087 2017-07-26 Andrew Pinski <apinski@cavium.com>
10089 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10092 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10095 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10096 (fp_valid_at): Likewise.
10098 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10100 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10101 (qdf24xx_addrcost_table): Likewise.
10102 (cortexa57_tunings): Update to use generic_branch_cost.
10103 (cortexa72_tunings): Likewise.
10104 (cortexa73_tunings): Likewise.
10105 (qdf24xx_tunings): Likewise.
10107 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10109 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10110 (thunderx2t99_branch_cost): Likewise.
10111 (cortexa35_tunings): Update to use generic_branch_cost.
10112 (cortexa53_tunings): Likewise.
10113 (cortexa57_tunings): Likewise.
10114 (cortexa72_tunings): Likewise.
10115 (cortexa73_tunings): Likewise.
10116 (thunderx2t99_tunings): Likewise.
10118 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10120 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10121 (sparc_option_override): Honour MASK_FSMULD.
10122 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10123 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10124 * config/sparc/sparc.opt (mfsmuld): New option.
10125 * doc/invoke.texi (mfsmuld): Document option.
10127 2017-07-26 Marek Polacek <polacek@redhat.com>
10129 PR middle-end/70992
10130 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10132 2017-07-26 Richard Biener <rguenther@suse.de>
10134 * gimple-match-head.c (do_valueize): Return OP if valueize
10136 (get_def): New helper to get at the def stmt of a SSA name
10137 if valueize allows.
10138 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10139 do_valueize to get at the def stmt.
10140 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10142 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10144 PR middle-end/46932
10145 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10147 2017-07-26 Martin Liska <mliska@suse.cz>
10150 * function.c (expand_function_start): Make expansion of
10151 nonlocal_goto_save_area after parm_birth_insn.
10153 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10155 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10156 from all CPU target flags enable members.
10158 2017-07-26 Richard Biener <rguenther@suse.de>
10160 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10161 (decision_tree::gen): Make 'type' const.
10162 (write_predicate): Likewise.
10164 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10166 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10167 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10168 (rs6000_option_override_internal): Likewise.
10169 (rs6000_expand_vector_set): Likewise.
10170 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10171 (TARGET_UPPER_REGS_SF): Likewise.
10172 (TARGET_UPPER_REGS_DI): Likewise.
10173 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10174 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10175 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10176 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10177 (Splitters for DI constants in Altivec registers): Likewise.
10178 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10179 (vsx_set_v4sf_p9): Likewise.
10180 (vsx_set_v4sf_p9_zero): Likewise.
10181 (vsx_insert_extract_v4sf_p9): Likewise.
10182 (vsx_insert_extract_v4sf_p9_2): Likewise.
10184 2017-07-25 Carl Love <cel@us.ibm.com>
10186 * doc/extend.texi: Update the built-in documentation file for the
10187 existing built-in functions
10188 vector signed char vec_cnttz (vector signed char);
10189 vector unsigned char vec_cnttz (vector unsigned char);
10190 vector signed short vec_cnttz (vector signed short);
10191 vector unsigned short vec_cnttz (vector unsigned short);
10192 vector signed int vec_cnttz (vector signed int);
10193 vector unsigned int vec_cnttz (vector unsigned int);
10194 vector signed long long vec_cnttz (vector signed long long);
10195 vector unsigned long long vec_cnttz (vector unsigned long long);
10197 2017-07-25 Andrew Pinski <apinski@cavium.com>
10199 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10200 accesses where the use is for the first operand of a BIT_INSERT.
10202 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10205 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10206 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10208 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10210 * config/i386/gstabs.h: Delete.
10211 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10213 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10215 * config/i386/i386.c (ix86_decompose_address): Do not check for
10216 register RTX when looking at index_reg or base_reg.
10217 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10219 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10221 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10222 to update EH info here.
10224 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10226 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10228 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10230 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10232 2017-07-25 Torsten Duwe <duwe@suse.de>
10234 * common.opt: Introduce -fpatchable-function-entry
10235 command line option, and its variables function_entry_patch_area_size
10236 and function_entry_patch_area_start.
10237 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10238 including a two-value parser.
10239 * target.def (print_patchable_function_entry): New target hook.
10240 * targhooks.h (default_print_patchable_function_entry): New function.
10241 * targhooks.c (default_print_patchable_function_entry): Likewise.
10242 * toplev.c (process_options): Switch off IPA-RA if
10243 patchable function entries are being generated.
10244 * varasm.c (assemble_start_function): Look at the
10245 patchable-function-entry command line switch and current
10246 function attributes and maybe generate NOP instructions by
10247 calling the print_patchable_function_entry hook.
10248 * doc/extend.texi: Document patchable_function_entry attribute.
10249 * doc/invoke.texi: Document -fpatchable_function_entry
10250 command line option.
10251 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10253 * doc/tm.texi: Re-generate.
10255 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10258 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10259 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10261 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10263 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10265 2017-07-25 Richard Biener <rguenther@suse.de>
10267 PR tree-optimization/81455
10268 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10269 not walk in cycles when looking for guards.
10271 2017-07-25 Richard Biener <rguenther@suse.de>
10273 PR tree-optimization/81529
10274 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10275 when optimizing backedge uses.
10277 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10279 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10281 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10282 to dl_section_ref. On AIX, append an expression to subtract
10283 the size of the section length to dl_section_ref.
10285 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10287 * configure.ac: If any of the config.* scripts fail, exit 1.
10288 * configure: Regenerate.
10290 2017-07-25 Richard Biener <rguenther@suse.de>
10292 PR middle-end/81546
10293 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10294 of immediate uses, be more verbose on errors.
10296 2017-07-25 Richard Biener <rguenther@suse.de>
10298 PR tree-optimization/81510
10299 * tree-vect-loop.c (vect_is_simple_reduction): When the
10300 reduction stmt is not inside the loop bail out.
10302 2017-07-25 Richard Biener <rguenther@suse.de>
10304 PR tree-optimization/81303
10305 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10306 profitability check against LOOP_VINFO_NITERSM1.
10308 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10310 * domwalk.c (cmp_bb_postorder): Simplify.
10311 (sort_bbs_postorder): New function. Use it...
10312 (dom_walker::walk): ...here to optimize common cases.
10314 2017-07-25 Martin Liska <mliska@suse.cz>
10317 * ipa-visibility.c (function_and_variable_visibility): Make the
10318 redirection just on target that supports aliasing.
10319 Fix GNU coding style.
10321 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10324 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10326 * config/arm/rtems.h: Likewise.
10327 * config/bfin/rtems.h: Likewise.
10328 * config/i386/rtemself.h: Likewise.
10329 * config/lm32/rtems.h: Likewise.
10330 * config/m32c/rtems.h: Likewise.
10331 * config/m68k/rtemself.h: Likewise.
10332 * config/microblaze/rtems.h: Likewise.
10333 * config/mips/rtems.h: Likewise.
10334 * config/moxie/rtems.h: Likewise.
10335 * config/nios2/rtems.h: Likewise.
10336 * config/powerpcspe/rtems.h: Likewise.
10337 * config/rs6000/rtems.h: Likewise.
10338 * config/rtems.h: Likewise.
10339 * config/sh/rtems.h: Likewise.
10340 * config/sh/rtemself.h: Likewise.
10341 * config/sparc/rtemself.h: Likewise.
10343 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10346 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10347 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10348 * tree-ssa-structalias.c (alias_get_name): Same.
10350 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10353 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10354 instructions if no du chain is found.
10356 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10358 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10360 2017-07-25 Richard Biener <rguenther@suse.de>
10362 PR middle-end/81505
10363 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10366 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10368 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10369 upper-regs options.
10370 (ISA_2_7_MASKS_SERVER): Likewise.
10371 (ISA_3_0_MASKS_IEEE): Likewise.
10372 (OTHER_P8_VECTOR_MASKS): Likewise.
10373 (OTHER_VSX_VECTOR_MASKS): Likewise.
10374 (POWERPC_MASKS): Likewise.
10375 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10376 duplicate list of options.
10377 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10378 explicit -mupper-regs options.
10379 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10380 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10381 alias for -mupper-regs-df.
10382 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10383 (rs6000_init_hard_regno_mode_ok): Likewise.
10384 (rs6000_option_override_internal): Likewise.
10385 (rs6000_opt_masks): Likewise.
10386 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10387 options in terms of whether -mvsx or -mpower8-vector was used.
10388 (TARGET_UPPER_REGS_DI): Likewise.
10389 (TARGET_UPPER_REGS_SF): Likewise.
10390 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10391 -mupper-regs-* options.
10393 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10395 * passes.c (emergency_dump_function): Print some empty lines and a
10396 header before the RTL dump.
10398 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10400 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10402 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10405 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10406 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10408 2017-07-24 Carl Love <cel@us.ibm.com>
10410 * config/rs6000/rs6000-c.c: Add support for built-in functions
10411 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10412 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10413 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10414 vec_extract_fp_from_shortl): Add defines for the two builtins.
10415 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10416 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10418 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10419 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10420 * doc/extend.texi: Update the built-in documentation file for the
10421 new built-in function.
10423 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10426 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10428 * doc/generic.texi: Likewise.
10429 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10430 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10432 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10434 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10435 (aarch64_mls_elt_merge<mode>): Likewise.
10437 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10439 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10440 having __cxa_atexit.
10442 2017-07-23 Michael Collison <michael.collison@arm.com>
10444 * config/arm/arm.c (arm_option_override): Deprecate
10445 use of -mstructure-size-boundary.
10446 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10447 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10449 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10452 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10453 Reduce cost estimate for direct moves.
10455 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10458 * config/i386/i386.c (ix86_option_override_internal): Disable
10459 BMI, BMI2 and TBM instructions for -m16.
10461 2017-07-21 Carl Love <cel@us.ibm.com>
10463 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10464 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10465 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10466 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10467 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10468 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10469 VMULOSW): New enum "unspec" values.
10470 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10471 altivec_vmulosw): New patterns.
10472 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10473 VMULOSW): Add definitions.
10475 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10477 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10478 (qdf24xx): Likewise.
10479 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10480 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10481 (AARCH64_FL_V8_1): Renumber.
10482 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10483 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10484 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10485 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10486 rdma to feature modifiers list.
10488 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10490 PR middle-end/56727
10491 * ipa-visibility (function_and_variable_visibility): Convert
10492 recursive PLT call to direct call if appropriate.
10494 2017-07-21 Andrew Pinski <apinski@cavium.com>
10496 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10497 operand 1 to see if the types precision matches.
10498 * fold-const.c (operand_equal_p): Likewise.
10500 2017-07-21 Richard Biener <rguenther@suse.de>
10502 PR tree-optimization/81303
10503 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10504 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10505 (vect_peeling_hash_get_lowest_cost): Adjust.
10506 (vect_enhance_data_refs_alignment): Likewise. Use
10507 vect_get_peeling_costs_all_drs to compute the penalty for no
10508 peeling to match up costs.
10510 2017-07-21 Richard Biener <rguenther@suse.de>
10512 PR tree-optimization/81500
10513 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10514 we didn't identify a reduction path.
10516 2017-07-21 Tom de Vries <tom@codesourcery.com>
10517 Cesar Philippidis <cesar@codesourcery.com>
10519 PR gcov-profile/81442
10520 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10523 2017-07-21 Tom de Vries <tom@codesourcery.com>
10526 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10528 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10529 nvptx_override_options_after_change.
10531 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10533 * dwarf2out.c (output_file_names): Avoid double testing for
10534 dwarf_version >= 5.
10536 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10538 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10540 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10542 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10544 (try_crossjump_to_edge): Do not punt on partitioned functions.
10546 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10548 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10549 Put all BBs reachable only via paths crossing cold region to cold
10551 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10553 2016-07-21 Richard Biener <rguenther@suse.de>
10555 PR tree-optimization/81303
10556 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10557 into account prologue and epilogue iterations when raising
10558 min_profitable_iters to sth at least covering one vector iteration.
10560 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10562 * config/arm/arm.c (arm_test_cpu_arch_dat):
10565 2017-07-20 Nathan Sidwell <nathan@acm.org>
10567 Remove TYPE_METHODS.
10568 * tree.h (TYPE_METHODS): Delete.
10569 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10570 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10571 (dbxout_type_methods): Scan TYPE_FIELDS.
10572 (dbxout_type): Don't check TYPE_METHODS here.
10573 * function.c (use_register_for_decl): Always ignore register for
10574 class types when not optimizing.
10575 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10576 * tree.c (free_lang_data_in_type): Stitch out member functions and
10577 templates from TYPE_FIELDS.
10578 (build_distinct_type_copy, verify_type_variant,
10579 verify_type): Member fns are on TYPE_FIELDS.
10580 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10581 * tree-pretty-print.c (dump_generic_node): Likewise.
10583 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10586 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10587 V2TImode and V4TImode.
10588 (ix86_expand_vector_extract): Likewise.
10589 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10590 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10591 (ssescalarmode): Handle V4TImode and V2TImode.
10592 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10593 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10594 (VEXTRACTI128_MODE): New mode iterator.
10595 (splitter for *vec_extractv?ti first element): New.
10596 (VEC_INIT_MODE): New mode iterator.
10597 (vec_init<mode>): Consolidate 3 expanders into one using
10598 VEC_INIT_MODE mode iterator.
10600 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10602 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10603 non_spilled_static_chain_regno_p.
10605 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10607 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10609 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10611 * bb-reorder.c (connect_traces): Allow copying of blocks within
10614 2017-07-20 Richard Biener <rguenther@suse.de>
10616 * gimple.h (gimple_phi_result): Add gphi * overload.
10617 (gimple_phi_result_ptr): Likewise.
10618 (gimple_phi_arg): Likewise. Adjust index assert to only
10619 allow actual argument accesses rather than all slots available
10621 (gimple_phi_arg_def): Add gphi * overload.
10622 * tree-phinodes.c (make_phi_node): Initialize only actual
10624 (resize_phi_node): Clear memory not covered by old node,
10625 do not initialize excess argument slots.
10626 (reserve_phi_args_for_new_edge): Initialize new argument slot
10629 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10631 PR tree-optimization/81388
10633 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10635 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10636 by removing computation of may_be_zero.
10638 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10639 Tom de Vries <tom@codesourcery.com>
10641 PR middle-end/81030
10642 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10643 when gimple level profile disagrees with what RTL expander did.
10645 2017-07-20 Richard Biener <rguenther@suse.de>
10647 PR tree-optimization/61171
10648 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10649 (vect_analyze_stmt): Add slp instance parameter.
10650 (vectorizable_reduction): Likewise.
10651 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10652 (vect_is_simple_reduction): Deal with chains not detected
10653 as SLP reduction chain, specifically not properly associated
10654 chains containing a mix of plus/minus.
10655 (get_reduction_op): Remove.
10656 (get_initial_defs_for_reduction): Simplify, pass in whether
10657 this is a reduction chain, pass in the SLP node for the PHIs.
10658 (vect_create_epilog_for_reduction): Get the SLP instance as
10660 (vectorizable_reduction): Get the SLP instance as arg.
10661 During analysis remember the SLP node with the PHIs in the
10662 instance. Simplify getting at the vectorized reduction PHIs.
10663 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10664 through SLP instance.
10665 (vect_slp_analyze_operations): Likewise.
10666 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10667 (vect_transform_stmt): Likewise.
10669 2017-07-20 Tom de Vries <tom@codesourcery.com>
10671 PR tree-optimization/81489
10672 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10673 read of phi arg location to before loop that modifies phi.
10675 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10677 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10680 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10682 PR middle-end/81331
10683 * except.c (execute): Fix ordering issue.
10685 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10687 PR rtl-optimization/81423
10688 * combine.c (make_compound_operation_int): Don't try to optimize
10689 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10691 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10693 PR rtl-optimization/81423
10694 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10695 with a constant that is -1 in the truncated to mode.
10697 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10699 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10700 (determine_unlikely_bbs): ... here.
10701 * predict.h (propagate_unlikely_bbs_forward): Declare.
10702 * cfgexpand.c (pass_expand::execute): Use it.
10703 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10705 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10706 propagate_unlikely_bbs_forward.
10708 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10710 PR middle-end/81331
10711 * except.c (maybe_add_nop_after_section_switch): New function.
10714 2017-07-19 Tom de Vries <tom@codesourcery.com>
10716 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10718 2017-07-19 Tom de Vries <tom@codesourcery.com>
10720 * gimple.h (gimple_phi_arg): Make assert more strict.
10722 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10724 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10725 * config/rs6000/mmintrin.h: New file.
10726 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10728 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10730 PR tree-optimization/81346
10731 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10733 2017-07-19 Tom de Vries <tom@codesourcery.com>
10735 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10737 2017-07-19 Tom de Vries <tom@codesourcery.com>
10739 * config/nvptx/nvptx-modes.def: Add V2DImode.
10740 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10741 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10742 (nvptx_output_mov_insn): Handle lack of mov.b128.
10743 (nvptx_print_operand): Handle 'H' and 'L' codes.
10744 (nvptx_vector_mode_supported): Allow V2DImode.
10745 (nvptx_preferred_simd_mode): New function.
10746 (nvptx_data_alignment): New function.
10747 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10748 nvptx_preferred_simd_mode.
10749 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10751 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10753 2017-07-19 Tom de Vries <tom@codesourcery.com>
10755 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10756 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10757 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10758 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10759 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10760 (mov<VECIM>_insn): New define_insn.
10761 (define_expand "mov<VECIM>): New define_expand.
10763 2017-07-19 Tom de Vries <tom@codesourcery.com>
10765 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10767 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10769 PR tree-optimization/81346
10770 * fold-const.h (fold_div_compare, range_check_type): Declare.
10771 * fold-const.c (range_check_type): New function.
10772 (build_range_check): Use range_check_type.
10773 (fold_div_compare): No longer static, rewritten into
10774 a match.pd helper function.
10775 (fold_comparison): Don't call fold_div_compare here.
10776 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10777 as helper function.
10779 2017-07-19 Nathan Sidwell <nathan@acm.org>
10781 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10782 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10783 * tree.c (find_decls_types_r, verify_type): Use
10784 TYPE_{MIN,MAX}_VALUE_RAW.
10785 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10786 (hash_tree): Likewise.
10787 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10789 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10792 2017-07-18 Tom de Vries <tom@codesourcery.com>
10794 PR middle-end/81464
10795 * omp-expand.c (expand_omp_for_static_chunk): Handle
10796 equal-argument loop exit phi.
10798 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10801 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10802 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10803 operand 2 predicate.
10804 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10805 operand 2 predicate.
10806 (ror,rol -> rorx splitters): Use const_int_operand as
10807 operand 2 predicate.
10809 2017-06-18 Richard Biener <rguenther@suse.de>
10811 PR tree-optimization/81410
10812 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10813 the gap in the ! slp_perm SLP case after each group.
10815 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10817 PR middle-end/81463
10818 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10821 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10823 PR middle-end/81462
10824 * predict.c (set_even_probabilities): Cleanup; do not affect
10825 probabilities that are already known.
10826 (combine_predictions_for_bb): Call even when count is set.
10828 2017-07-18 Nathan Sidwell <nathan@acm.org>
10830 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10833 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10836 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10837 optimization for loop niter analysis.
10839 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10842 * config/avr/avr.c (avr_optimize_casesi): Don't use
10843 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10845 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10847 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10848 body_cost_vec from _vect_peel_extended_info.
10849 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10850 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10853 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10855 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10857 2017-07-18 Richard Biener <rguenther@suse.de>
10859 PR tree-optimization/80620
10860 PR tree-optimization/81403
10861 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10862 info when re-using a VN table entry.
10864 2017-07-18 Richard Biener <rguenther@suse.de>
10866 PR tree-optimization/81418
10867 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10868 vectype_in. Verify that with lane-reducing reduction operations
10869 we have a single def-use cycle.
10871 2017-07-17 Carl Love <cel@us.ibm.com>
10873 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10875 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10876 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10877 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10878 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10879 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10880 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10881 VMULOSW): New enum "unspec" values.
10882 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10883 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10884 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10885 altivec_vmulosw): New patterns.
10886 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10887 VMULOSW): Add definitions.
10888 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10890 * config/alpha/alpha.c: Include predict.h.
10892 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10894 * tree-vrp.c (compare_assert_loc): Fix comparison function
10895 to return predictable results.
10897 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10899 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10901 (subdi3): Likewise.
10902 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10903 * doc/invoke.texi (mexpand-adddi): Update text.
10905 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10907 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10908 that also clobbers the CC register. The old expand code is moved
10910 (*arc_clzsi2): ... here.
10911 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10912 the CC register. The old expand code is moved to ...
10913 (arc_ctzsi2): ... here.
10915 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10917 * config/arc/arc.opt (mindexed-loads): Use initial value
10918 TARGET_INDEXED_LOADS_DEFAULT.
10919 (mauto-modify-reg): Use initial value
10920 TARGET_AUTO_MODIFY_REG_DEFAULT.
10921 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10922 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10923 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10924 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10926 2017-07-17 Martin Liska <mliska@suse.cz>
10929 * opts.c (finish_options): Do not allow -fgnu-tm
10930 w/ -fsanitize={kernel-,}address. Say sorry.
10932 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10935 * tree-loop-distribution.c (classify_partition): Only assert on
10936 numer of iterations.
10937 (merge_dep_scc_partitions): Delete prameter. Update function call.
10938 (distribute_loop): Remove code handling loop with unknown niters.
10939 (pass_loop_distribution::execute): Skip loop with unknown niters.
10941 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10944 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10945 function sort_partitions_by_post_order.
10947 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10949 PR tree-optimization/81374
10950 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10951 the max index of basic blocks, rather than number of basic blocks.
10953 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10955 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10957 (arc_legitimate_pic_operand_p): Likewise.
10958 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10960 (arc_needs_pcl_p): Likewise.
10961 (arc_legitimate_pc_offset_p): Likewise.
10962 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10963 function is also used in constrains.md.
10964 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10965 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10966 PLUS. Only return true/false in known cases, otherwise assert.
10967 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10968 is already called in arc_legitimate_constant_p.
10969 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10971 (LEGITIMATE_PIC_OPERAND_P): Use
10972 arc_raw_symbolic_reference_mentioned_p function.
10973 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10978 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10979 Andrew Burgess <andrew.burgess@embecosm.com>
10981 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10982 (arc_return_address_register): New function.
10983 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10984 (arc_handle_fndecl_attribute): Add naked attribute.
10985 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10986 (TARGET_WARN_FUNC_RETURN): Likewise.
10987 (arc_allocate_stack_slots_for_args): New function.
10988 (arc_warn_func_return): Likewise.
10989 (machine_function): Change type fn_type.
10990 (arc_compute_function_type): Consider new naked function type,
10991 change function return type.
10992 (arc_must_save_register): Adapt to handle new
10993 arc_compute_function_type's return type.
10994 (arc_expand_prologue): Likewise.
10995 (arc_expand_epilogue): Likewise.
10996 (arc_return_address_regs): Delete.
10997 (arc_return_address_register): New function.
10998 (arc_epilogue_uses): Use above function.
10999 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11000 (arc_function_type): Change encoding, add naked type.
11001 (ARC_INTERRUPT_P): Change to handle the new encoding.
11002 (ARC_FAST_INTERRUPT_P): Likewise.
11003 (ARC_NORMAL_P): Define.
11004 (ARC_NAKED_P): Likewise.
11005 (arc_compute_function_type): Delete prototype.
11006 * config/arc/arc.md (in_ret_delay_slot): Use
11007 arc_return_address_register function.
11008 (simple_return): Likewise.
11009 (p_return_i): Likewise.
11011 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11013 PR tree-optimization/81428
11014 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11015 can't be built for those types.
11017 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11019 Remove stuff dead since r239246.
11021 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11022 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11023 (avr_inform_devices): Remove dead stuff.
11025 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11027 * config/arm/arm_neon.h: Fix softp typo.
11029 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11031 PR tree-optimization/81365
11032 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11033 aggregate moves onto bb predecessor edges, make sure there are no
11034 loads that could alias the lhs in between the start of bb and the
11037 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11040 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11041 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11042 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11044 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11046 PR tree-optimization/81396
11047 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11048 (init_symbolic_number): Initialize it to 1.
11049 (perform_symbolic_merge): Add n_ops from both operands into the new
11051 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11052 without base_addr as useless if they need more than one operation.
11053 (bswap_replace): Handle !bswap case for NULL base_addr.
11055 2017-07-17 Tom de Vries <tom@codesourcery.com>
11058 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11061 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11063 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11064 conditional builtin define __FIX_LEON3FT_B2BST.
11066 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11068 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11069 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11072 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11074 PR rtl-optimization/81424
11075 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11076 to remove potential trapping from operands if -fnon-call-exceptions.
11078 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11080 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11081 profile_proability for scalling.
11082 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11084 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11086 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11088 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11090 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11091 fixpoint arithmetics.
11093 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11095 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11096 fixpoint arithmetics.
11098 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11100 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11101 fixpoint arithmetics.
11103 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11105 * profile-count.h (profile_probability::from_reg_br_prob_note,
11106 profile_probability::to_reg_br_prob_note): New functions.
11107 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11108 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11109 * predict.c (probability_reliable_p): Update.
11110 (edge_probability_reliable_p): Update.
11111 (br_prob_note_reliable_p): Update.
11112 (invert_br_probabilities): Update.
11113 (add_reg_br_prob_note): New function.
11114 (combine_predictions_for_insn): Update.
11115 * asan.c (asan_clear_shadow): Update.
11116 * cfgbuild.c (compute_outgoing_frequencies): Update.
11117 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11118 (update_br_prob_note): Update.
11119 (rtl_verify_edges): Update.
11120 (purge_dead_edges): Update.
11121 (fixup_reorder_chain): Update.
11122 * emit-rtl.c (try_split): Update.
11123 * ifcvt.c (cond_exec_process_insns): Update.
11124 (cond_exec_process_if_block): Update.
11125 (dead_or_predicable): Update.
11126 * internal-fn.c (expand_addsub_overflow): Update.
11127 (expand_neg_overflow): Update.
11128 (expand_mul_overflow): Update.
11129 * loop-doloop.c (doloop_modify): Update.
11130 * loop-unroll.c (compare_and_jump_seq): Update.
11131 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11132 * predict.h: Update.
11133 * reorg.c (mostly_true_jump): Update.
11135 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11136 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11137 * config/arc/arc.c: (emit_unlikely_jump): Update.
11138 * config/arm/arm.c: (emit_unlikely_jump): Update.
11139 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11140 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11141 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11142 (ix86_print_operand): Update.
11143 (ix86_split_fp_branch): Update.
11144 (predict_jump): Update.
11145 * config/ia64/ia64.c (ia64_print_operand): Update.
11146 * config/mmix/mmix.c (mmix_print_operand): Update.
11147 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11148 (rs6000_expand_split_stack_prologue): Update.
11149 * config/rs6000/rs6000.c: Update.
11150 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11151 (s390_expand_vec_movstr): Update.
11152 (s390_expand_cs_tdsi): Update.
11153 (s390_expand_split_stack_prologue): Update.
11154 * config/sh/sh.c (sh_print_operand): Update.
11155 (expand_cbranchsi4): Update.
11156 (expand_cbranchdi4): Update.
11157 * config/sparc/sparc.c (output_v9branch): Update.
11158 * config/spu/spu.c (get_branch_target): Update.
11159 (ea_load_store_inline): Update.
11160 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11161 * config/tilepro/tilepro.c: Update.
11163 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11165 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11166 (gimplify_save_expr): Likewise.
11168 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11170 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11172 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11174 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11176 * timevar.def (TV_IPA_FNSUMMARY): Define.
11178 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11180 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11181 to back store errata sensitive sequence from being generated.
11182 (sqrtdf2_fix): Likewise.
11184 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11186 * tree-ssa-threadupdate.c (compute_path_counts,
11187 update_joiner_offpath_counts): Use profile_probability.
11189 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11192 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11194 * config/arm/arm-c.c (arm_cpu_builtins): Define
11195 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11197 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11199 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11200 array entries to represent __ieee128 versions of the
11201 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11202 scalar_extract_sig, and scalar_insert_exp built-in functions.
11203 (altivec_resolve_overloaded_builtin): Add special case handling
11204 for the __builtin_scalar_insert_exp function, as represented by
11205 the P9V_BUILTIN_VEC_VSIEDP constant.
11206 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11207 exponent support for __ieee128 argument.
11208 (VSESQP): Add scalar extract signature support for __ieee128
11210 (VSTDCNQP): Add scalar test negative support for __ieee128
11212 (VSIEQP): Add scalar insert exponent support for __int128 argument
11213 with __ieee128 result.
11214 (VSIEQPF): Add scalar insert exponent support for __ieee128
11215 argument with __ieee128 result.
11216 (VSTDCQP): Add scalar test data class support for __ieee128
11218 (VSTDCNQP): Add overload support for scalar test negative with
11219 __ieee128 argument.
11220 (VSTDCQP): Add overload support for scalar test data class
11221 __ieee128 argument.
11222 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11223 UNSPEC_VSX_SXSIGDP.
11224 (UNSPEC_VSX_SIEXPQP): New constant.
11225 (xsxexpqp): New insn for VSX scalar extract exponent quad
11227 (xsxsigqp): New insn for VSX scalar extract significand quad
11229 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11230 precision with floating point argument.
11231 (xststdcqp): New expand for VSX scalar test data class quad
11233 (xststdcnegqp): New expand for VSX scalar test negative quad
11235 (xststdcqp): New insn to match expansions for VSX scalar test data
11236 class quad precision and VSX scalar test negative quad precision.
11237 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11238 special case operand checking to enforce that second operand of
11239 VSX scalar test data class with quad precision argument is a 7-bit
11241 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11242 prototypes and descriptions of __ieee128 versions of
11243 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11244 scalar_test_data_class, and scalar_test_neg built-in functions.
11246 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11248 PR tree-optimization/81162
11249 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11250 replace a negate with an add.
11252 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11254 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11256 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11258 * config/arm/arm-c.c (arm_cpu_builtins): Define
11259 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11261 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11263 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11264 (armv8-r): Set ARM Cortex-R52 as default CPU.
11265 * config/arm/arm-tables.opt: Regenerate.
11266 * config/arm/arm-tune.md: Regenerate.
11267 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11269 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11270 extension for -mcpu=cortex-r52.
11272 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11274 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11275 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11276 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11277 (fp-armv8): Define it as FP_ARMv8 only.
11278 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11279 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11281 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11282 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11283 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11284 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11285 than TARGET_FPU_ARMV8.
11286 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11287 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11288 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11290 * config/arm/neon.md (neon_vrint): Likewise.
11291 (neon_vcvt): Likewise.
11292 (neon_<fmaxmin_op><mode>): Likewise.
11293 (<fmaxmin><mode>3): Likewise.
11294 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11295 * config/arm/predicates.md (arm_cond_move_operator): Check against
11296 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11298 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11300 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11301 to top of function.
11303 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11305 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11306 loop in comment with memset.
11308 2017-07-14 Martin Liska <mliska@suse.cz>
11310 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11311 * dwarf2out.c (is_java): Remove the function.
11312 (output_pubname): Remove usage of the function.
11313 (lower_bound_default): Remove usage of DW_LANG_Java.
11314 (gen_compile_unit_die): Likewise.
11315 * gcc.c: Remove compiler defaults for .java and .zip files.
11316 * gimple-expr.c (remove_suffix): Change as there's no longer
11317 extension than 4-letter one.
11318 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11319 (gimplify_save_expr): Likewise.
11320 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11321 as it's possible even for other languages than Java.
11322 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11323 * lto-opts.c (lto_write_options): Remove reference to Java.
11324 * opts.c (strip_off_ending): Update file extension handling.
11325 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11326 * tree-eh.c (lower_resx): Likewise.
11327 * tree.c (free_lang_data_in_type): Remove dead code.
11328 (find_decls_types_r): Likewise.
11329 (build_common_builtin_nodes): Remove Java from a comment.
11330 (verify_type): Remove dead code.
11331 * varasm.c (assemble_external): Remove Java from a comment.
11333 2017-07-14 Martin Liska <mliska@suse.cz>
11335 * opts.c (finish_options): Add quotes.
11336 (common_handle_option): Likewise.
11338 2017-07-14 Martin Liska <mliska@suse.cz>
11340 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11341 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11342 Remove N_SO_PASCAL.
11343 * dwarf2out.c (lower_bound_default): Do not handle
11345 (gen_compile_unit_die): Likewise.
11346 * gcc.c: Remove default extension binding for GNU Pascal.
11347 * stmt.c: Remove Pascal language from a comment.
11348 * xcoffout.c: Likewise.
11350 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11353 * diagnostic-show-locus.c (fixit_cmp): New function.
11354 (layout::layout): Sort m_fixit_hints.
11355 (column_range::column_range): Assert that the values are valid.
11356 (struct char_span): New struct.
11357 (correction::overwrite): New method.
11358 (struct source_line): New struct.
11359 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11360 calls in terms of classes source_line and char_span, and
11361 correction::overwrite.
11362 (selftest::test_overlapped_fixit_printing_2): New function.
11363 (selftest::diagnostic_show_locus_c_tests): Call it.
11365 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11367 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11368 early if there is no lhs.
11370 2017-07-13 Martin Liska <mliska@suse.cz>
11372 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11373 (gen_reference_type_die): Likewise.
11374 * stor-layout.c: Remove Pascal-related comment.
11376 2017-07-13 Martin Liska <mliska@suse.cz>
11378 * opts.c (finish_options): Add quotes to error messages.
11379 (parse_sanitizer_options): Likewise.
11381 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11383 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11385 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11387 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11389 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11391 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11393 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11395 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11398 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11399 provides the hardware capability bits, define the macro
11400 __BUILTIN_CPU_SUPPORTS__.
11401 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11402 if GLIBC does not provide the hardware capability bits. Add a
11403 gcc_unreachable call if the built-in cpu function is neither
11404 __builtin_cpu_is nor __builtin_cpu_supports.
11405 (rs6000_get_function_versions_dispatcher): Change the warning
11406 that an old GLIBC is used which does not export the capability
11407 bits to be an error.
11408 * doc/extend.texi (target_clones attribute): Document the
11409 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11410 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11411 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11412 the macros defined by GCC if the newer GLIBC is available.
11414 2017-07-12 Jeff Law <law@redhat.com>
11416 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11417 remaining includes slightly.
11418 * config/riscv/riscv-builtins.c: Include profile-count.h.
11420 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11423 * config/avr/avr.c (avr_set_current_function): In diagnostic
11424 messages: Quote keywords and (parts of) identifiers.
11425 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11428 2017-07-12 Carl Love <cel@us.ibm.com>
11430 * config/rs6000/rs6000-c.c: Add support for built-in functions
11431 vector bool char vec_revb (vector bool char);
11432 vector bool short vec_revb (vector short char);
11433 vector bool int vec_revb (vector bool int);
11434 vector bool long long vec_revb (vector bool long long);
11435 * doc/extend.texi: Update the built-in documentation file for the
11436 new built-in functions.
11438 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11440 * config/s390/s390.md: Remove movcc splitter.
11442 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11444 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11445 load/store on condition.
11447 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11450 * config/avr/avr.c (avr_encode_section_info)
11451 [progmem && !TREE_READONLY]: Error if progmem object needs
11454 2017-07-11 Michael Collison <michael.collison@arm.com>
11456 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11459 2017-07-11 Carl Love <cel@us.ibm.com>
11461 * config/rs6000/rs6000-c.c: Add support for builtins
11462 vector unsigned int vec_parity_lsbb (vector signed int);
11463 vector unsigned int vec_parity_lsbb (vector unsigned int);
11464 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11465 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11466 vector unsigned long long vec_parity_lsbb (vector signed long long);
11467 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11468 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11469 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11470 * doc/extend.texi: Update the built-in documentation file for the
11471 new built-in functions.
11473 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11475 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11476 (layout::m_primary_loc): New field.
11477 (layout::layout): Initialize new field. Move location filtering
11478 logic from here to...
11479 (layout::maybe_add_location_range): ...this new method. Add
11480 support for filtering to just the lines already specified by other
11482 (layout::will_show_line_p): New method.
11483 (gcc_rich_location::add_location_if_nearby): New method.
11484 (selftest::test_add_location_if_nearby): New test function.
11485 (selftest::diagnostic_show_locus_c_tests): Call it.
11486 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11489 2017-07-11 Tom de Vries <tom@codesourcery.com>
11491 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11492 (bb_first_real_insn): New function.
11493 (nvptx_single): Add extra initialization of broadcasted condition
11496 2017-07-11 Nathan Sidwell <nathan@acm.org>
11498 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11500 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11502 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11503 Binutils doc as TEXI will mess them up.
11504 * doc/invoke.texi (AVR Options): Same here.
11506 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11508 * config/sparc/sparc.opt (mfix-ut700): New option.
11509 (mfix-gr712rc): Likewise.
11510 (sparc_fix_b2bst): New variable.
11511 * doc/invoke.texi (SPARC options): Document them.
11512 (ARM options): Fix warnings.
11513 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11514 instructions to prevent sequences that can trigger the store-store
11515 errata for certain LEON3FT processors.
11516 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11517 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11518 * config/sparc/sparc.md (fix_b2bst): New attribute.
11519 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11521 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11524 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11526 (*rsqrtsf2_sse): Ditto.
11528 (div<mode>3): Macroize insn from divdf3 and divsf3
11529 using MODEF mode iterator.
11531 2017-07-10 Martin Sebor <msebor@redhat.com>
11533 PR tree-optimization/80397
11534 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11535 instead of testing for equality to INTEGER_TYPE.
11537 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11539 * config.gcc: Remove uclibc from arc target spec.
11541 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11543 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11545 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11548 * lto-wrapper.c (remove_option): New function.
11549 (merge_and_complain): Merge PIC/PIE options more realistically.
11551 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11553 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11557 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11558 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11559 * config.in: Regenerate.
11560 * configure: Regenerate.
11561 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11562 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11563 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11564 (TARGET_GASISR_PROLOGUES): ...target mask.
11565 * common/config/avr/avr-common.c
11566 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11567 Set -mgas-isr-prologues.
11568 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11569 INSERT_PASS_BEFORE for it.
11570 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11571 * config/avr/avr.c (avr_option_override)
11572 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11573 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11574 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11575 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11576 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11578 (make_avr_pass_pre_proep): New function.
11579 (emit_push_sfr) <treg>: Add argument to function and use it
11580 instead of TMP_REG.
11581 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11582 and set machine->gasisr.yes.
11583 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11584 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11585 __gcc_isr.n_pushed to .L__stack_usage.
11586 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11587 (avr_asm_final_postscan_insn): ...this new static function.
11588 * config/avr/avr.h (machine_function)
11589 <is_no_gccisr, use_L__stack_usage>: New fields.
11590 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11591 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11592 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11593 (gasisr, *gasisr): New expander and insn.
11594 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11595 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11596 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11598 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11600 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11603 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11605 Move jump-tables out of .text again.
11608 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11609 (ASM_OUTPUT_ADDR_VEC): New function.
11610 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11611 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11612 INSN_ADDRESSes as asm comment.
11613 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11614 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11615 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11616 * config/avr/avr.md (*tablejump): Adjust comment.
11617 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11618 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11620 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11621 (avr_output_addr_vec): New proto.
11622 (avr_log_t) <insn_addresses>: New field.
11624 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11627 * config/i386/i386.c (ix86_function_arg_advance): Set
11628 outgoing_args_on_stack to true if there are outgoing arguments
11630 (ix86_function_arg): Likewise.
11631 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11632 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11633 * config/i386/i386.h (machine_function): Add
11634 outgoing_args_on_stack.
11636 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11638 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11639 supporting pthreds.
11640 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11642 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11644 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11645 (REAL_H): Remove $(MACHMODE_H).
11646 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11648 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11649 $(MACHMODE_H) and double-int.h.
11650 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11652 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11655 2017-07-07 Andrew Pinski <apinski@cavium.com>
11657 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11658 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11660 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11662 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11663 Add warning if GCC was not configured to link against a GLIBC that
11664 exports the hardware capability bits.
11665 (make_resolver_func): Make resolver function private and not a
11666 COMDAT function. Create the name with clone_function_name instead
11667 of make_unique_name.
11670 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11671 correct operand in doing the split.
11673 2017-07-07 Carl Love <cel@us.ibm.com>
11675 * config/rs6000/rs6000-c: Add support for built-in function
11676 vector unsigned short vec_pack_to_short_fp32 (vector float,
11678 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11679 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11680 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11681 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11682 (convert_4f32_8i16): Add define_expand.
11683 * doc/extend.texi: Update the built-in documentation file for the
11684 new built-in function.
11686 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11688 * config/sparc/m8.md: New file.
11689 * config/sparc/sparc.md: Include m8.md.
11691 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11693 * config/sparc/sparc.opt: New option -mvis4b.
11694 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11695 (sparc_option_override): Handle VIS4B.
11696 (enum sparc_builtins): Define
11697 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11698 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11699 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11700 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11701 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11702 (check_constant_argument): New function.
11703 (sparc_vis_init_builtins): Define builtins
11704 __builtin_vis_dictunpack{8,16,32},
11705 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11706 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11707 __builtin_vis_fpcmpde{8,16,32}shl and
11708 __builtin_vis_fpcmpur{8,16,32}shl.
11709 (sparc_expand_builtin): Check that the constant operands to
11710 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11711 constant and in range.
11712 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11714 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11715 (SPARC_IMM5_P): Likewise.
11716 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11717 (enabled): Handle vis4b.
11718 (UNSPEC_DICTUNPACK): New unspec.
11719 (UNSPEC_FPCMPSHL): Likewise.
11720 (UNSPEC_FPUCMPSHL): Likewise.
11721 (UNSPEC_FPCMPDESHL): Likewise.
11722 (UNSPEC_FPCMPURSHL): Likewise.
11723 (cpu_feature): New CPU feature `vis4b'.
11724 (dictunpack{8,16,32}): New insns.
11725 (FPCSMODE): New mode iterator.
11726 (fpcscond): New code iterator.
11727 (fpcsucond): Likewise.
11728 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11729 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11730 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11731 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11732 * config/sparc/constraints.md: Define constraints `q' for unsigned
11733 2-bit integer constants and `t' for unsigned 5-bit integer
11735 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11737 (imm5_operand_dictunpack16): Likewise.
11738 (imm5_operand_dictunpack32): Likewise.
11739 (imm2_operand): Likewise.
11740 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11741 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11742 ditunpack* and fpcmp*shl builtins.
11744 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11746 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11747 * config.in: Add HAVE_AS_SPARC6 define.
11748 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11750 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11752 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11753 (CPP_CPU_SPEC): Handle m8.
11754 (ASM_CPU_SPEC): Likewise.
11755 * config/sparc/sparc-opts.h (enum processor_type): Add
11757 * config/sparc/sparc.c (m8_costs): New struct.
11758 (sparc_option_override): Handle TARGET_CPU_m8.
11759 (sparc32_initialize_trampoline): Likewise.
11760 (sparc64_initialize_trampoline): Likewise.
11761 (sparc_issue_rate): Likewise.
11762 (sparc_register_move_cost): Likewise.
11763 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11764 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11765 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11766 (CPP_CPU_SPEC): Handle M8.
11767 (ASM_CPU_SPEC): Likewise.
11768 (AS_M8_FLAG): Define.
11769 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11770 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11771 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11773 * configure: Regenerate.
11774 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11777 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11779 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11781 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11782 ("*movdi_insn_sp32"): Do not set v3pipe.
11783 ("*movsi_insn"): Likewise.
11784 ("*movdi_insn_sp64"): Likewise.
11785 ("*movsf_insn"): Likewise.
11786 ("*movdf_insn_sp32"): Likewise.
11787 ("*movdf_insn_sp64"): Likewise.
11788 ("*zero_extendsidi2_insn_sp64"): Likewise.
11789 ("*sign_extendsidi2_insn"): Likewise.
11790 ("*mov<VM32:mode>_insn"): Likewise.
11791 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11792 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11793 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11794 ("<vlop:code><VL:mode>3"): Likewise.
11795 ("*not_<vlop:code><VL:mode>3"): Likewise.
11796 ("*nand<VL:mode>_vis"): Likewise.
11797 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11798 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11799 ("one_cmpl<VL:mode>2"): Likewise.
11800 ("faligndata<VM64:mode>_vis"): Likewise.
11801 ("alignaddrsi_vis"): Likewise.
11802 ("alignaddrdi_vis"): Likweise.
11803 ("alignaddrlsi_vis"): Likewise.
11804 ("alignaddrldi_vis"): Likewise.
11805 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11806 ("bmaskdi_vis"): Likewise.
11807 ("bmasksi_vis"): Likewise.
11808 ("bshuffle<VM64:mode>_vis"): Likewise.
11809 ("cmask8<P:mode>_vis"): Likewise.
11810 ("cmask16<P:mode>_vis"): Likewise.
11811 ("cmask32<P:mode>_vis"): Likewise.
11812 ("pdistn<P:mode>_vis"): Likewise.
11813 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11815 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11817 * config/sparc/sparc.md ("subtype"): New insn attribute.
11818 ("*wrgsr_sp64"): Set insn subtype.
11819 ("*rdgsr_sp64"): Likewise.
11820 ("alignaddrsi_vis"): Likewise.
11821 ("alignaddrdi_vis"): Likewise.
11822 ("alignaddrlsi_vis"): Likewise.
11823 ("alignaddrldi_vis"): Likewise.
11824 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11825 ("fexpand_vis"): Likewise.
11826 ("fpmerge_vis"): Likewise.
11827 ("faligndata<VM64:mode>_vis"): Likewise.
11828 ("bshuffle<VM64:mode>_vis"): Likewise.
11829 ("cmask8<P:mode>_vis"): Likewise.
11830 ("cmask16<P:mode>_vis"): Likewise.
11831 ("cmask32<P:mode>_vis"): Likewise.
11832 ("fchksm16_vis"): Likewise.
11833 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11834 ("fmean16_vis"): Likewise.
11835 ("fp<plusminus_insn>64_vis"): Likewise.
11836 ("<plusminus_insn>v8qi3"): Likewise.
11837 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11838 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11839 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11840 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11841 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11842 ("*movqi_insn"): Likewise.
11843 ("*movhi_insn"): Likewise.
11844 ("*movsi_insn"): Likewise.
11845 ("movsi_pic_gotdata_op"): Likewise.
11846 ("*movdi_insn_sp32"): Likewise.
11847 ("*movdi_insn_sp64"): Likewise.
11848 ("movdi_pic_gotdata_op"): Likewise.
11849 ("*movsf_insn"): Likewise.
11850 ("*movdf_insn_sp32"): Likewise.
11851 ("*movdf_insn_sp64"): Likewise.
11852 ("*zero_extendhisi2_insn"): Likewise.
11853 ("*zero_extendqihi2_insn"): Likewise.
11854 ("*zero_extendqisi2_insn"): Likewise.
11855 ("*zero_extendqidi2_insn"): Likewise.
11856 ("*zero_extendhidi2_insn"): Likewise.
11857 ("*zero_extendsidi2_insn_sp64"): Likewise.
11858 ("ldfsr"): Likewise.
11859 ("prefetch_64"): Likewise.
11860 ("prefetch_32"): Likewise.
11861 ("tie_ld32"): Likewise.
11862 ("tie_ld64"): Likewise.
11863 ("*tldo_ldub_sp32"): Likewise.
11864 ("*tldo_ldub1_sp32"): Likewise.
11865 ("*tldo_ldub2_sp32"): Likewise.
11866 ("*tldo_ldub_sp64"): Likewise.
11867 ("*tldo_ldub1_sp64"): Likewise.
11868 ("*tldo_ldub2_sp64"): Likewise.
11869 ("*tldo_ldub3_sp64"): Likewise.
11870 ("*tldo_lduh_sp32"): Likewise.
11871 ("*tldo_lduh1_sp32"): Likewise.
11872 ("*tldo_lduh_sp64"): Likewise.
11873 ("*tldo_lduh1_sp64"): Likewise.
11874 ("*tldo_lduh2_sp64"): Likewise.
11875 ("*tldo_lduw_sp32"): Likewise.
11876 ("*tldo_lduw_sp64"): Likewise.
11877 ("*tldo_lduw1_sp64"): Likewise.
11878 ("*tldo_ldx_sp64"): Likewise.
11879 ("*mov<VM32:mode>_insn"): Likewise.
11880 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11881 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11883 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11885 * config/sparc/sparc.md ("type"): New insn type viscmp.
11886 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11888 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11889 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11890 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11891 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11893 ("n7_vis_logical_11cycle"): Likewise.
11894 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11895 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11896 * config/sparc/niagara.md ("niag_vis"): Likewise.
11897 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11898 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11900 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11902 * config/sparc/sparc.md: New instruction type `bmask'.
11903 (bmaskdi_vis): Use the `bmask' type.
11904 (bmasksi_vis): Likewise.
11905 * config/sparc/ultra3.md (us3_array): Likewise.
11906 * config/sparc/niagara7.md (n7_array): Likewise.
11907 * config/sparc/niagara4.md (n4_array): Likewise.
11908 * config/sparc/niagara2.md (niag2_vis): Likewise.
11909 (niag3_vis): Likewise.
11910 * config/sparc/niagara.md (niag_vis): Likewise.
11912 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11914 * ipa-comdats.c: Remove optimize check from gate.
11915 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11916 for functions not optimized.
11917 (ipa_fn_summary_read): Skip optimize check.
11918 (ipa_fn_summary_write): Likewise.
11919 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11921 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11923 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11924 (check_callers): Give up on caller which is not optimized.
11925 (inline_small_functions): Likewise.
11926 (ipa_inline): Do not give up when not optimizing.
11927 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11928 away unoptimizes cdtors.
11929 (whole_program_function_and_variable_visibility): Do
11930 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11931 * ipa.c (process_references): Do not check optimize.
11932 (symbol_table::remove_unreachable_nodes): Update optimize check.
11933 (set_writeonly_bit): Update optimize check.
11934 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11935 (pass_ipa_single_use::gate): Remove.
11937 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11939 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11940 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11941 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11942 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11943 permute_load, permute_store, adjust_extract, adjust_splat,
11944 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11945 replace_swap_with_copy, dump_swap_insn_table,
11946 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11947 recombine_lvx_pattern, recombine_stvx_pattern,
11948 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11949 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11950 to file rs6000-p8swap.c.
11951 * config/rs6000/rs6000-p8swap.c: New file.
11952 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11953 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11954 and rs6000*-*-* targets.
11956 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11958 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11960 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11962 * lto-wrapper.c (merge_and_complain): Do not merge
11963 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11964 fsigned_zeros, ftrapping_math, fwrapv.
11965 (append_compiler_options): Do not track these options.
11966 (append_linker_options): Likewie
11968 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11970 * cgraphunit.c (cgraph_node::finalize_function): When
11971 !flag_toplevel_reorde set no_reorder flag.
11972 (varpool_node::finalize_decl): Likewise.
11973 (symbol_table::compile): Drop no toplevel reorder path.
11975 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11977 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11978 edges; zero probability is not better than uninitialized.
11980 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11982 * asan.h (asan_sanitize_allocas_p): Declare.
11983 * asan.c (asan_sanitize_allocas_p): New function.
11984 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11985 (handle_builtin_alloca): Likewise.
11986 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11987 if !asan_sanitize_allocas_p.
11988 * params.def (asan-instrument-allocas): Add new option.
11989 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11990 * opts.c (common_handle_option): Disable allocas sanitization for
11993 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11995 * asan.c: Include gimple-fold.h.
11996 (get_last_alloca_addr): New function.
11997 (handle_builtin_stackrestore): Likewise.
11998 (handle_builtin_alloca): Likewise.
11999 (asan_emit_allocas_unpoison): Likewise.
12000 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12001 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12002 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12003 (instrument_builtin_call): Pass gimple iterator to
12004 get_mem_refs_of_builtin_call.
12005 (last_alloca_addr): New global.
12006 * asan.h (asan_emit_allocas_unpoison): Declare.
12007 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12008 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12009 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12010 if function calls alloca.
12011 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12012 * gimple-fold.h (replace_call_with_value): Declare.
12013 * internal-fn.c: Include asan.h.
12014 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12015 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12017 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12019 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12020 (C_SELFTEST_FLAGS): New.
12021 (CPP_SELFTEST_FLAGS): New.
12022 (SELFTEST_DEPS): New, from deps of s-selftest.
12023 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12024 (CPP_SELFTEST_DEPS): New.
12025 (selftest): Add dependency on s-selftest-c++.
12026 (s-selftest): Rename to...
12027 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12028 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12029 than SELFTEST_FLAGS.
12030 (selftest-gdb): Rename to...
12031 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12033 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12034 (selftest-valgrind): Rename to...
12035 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12037 (selftest-valgrind): Reintroduce as an alias for
12038 selftest-c-valgrind.
12039 (s-selftest-c++): New.
12040 (selftest-c++-gdb): New.
12041 (selftest-c++-valgrind): New.
12043 2017-07-06 Olivier Hainque <hainque@adacore.com>
12045 * gcc.c (process_command): When deciding if undefined variables
12046 should be ignored when processing specs, accept "gcc -v" as well.
12048 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12050 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12051 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12053 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12055 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12056 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12057 * config/arm/arm-tables.opt: Regenerate.
12058 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12060 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12062 2017-07-06 Carl Love <cel@us.ibm.com>
12064 * ChangeLog: Clean up from mid air collision
12066 2017-07-06 Carl Love <cel@us.ibm.com>
12068 * config/rs6000/rs6000-c.c: Add support for built-in functions
12069 vector signed int vec_subc (vector signed int, vector signed int);
12070 vector signed __int128 vec_subc (vector signed __int128,
12071 vector signed __int128);
12072 vector unsigned __int128 vec_subc (vector unsigned __int128,
12073 vector unsigned __int128);
12074 vector signed int vec_sube (vector signed int, vector signed int,
12075 vector signed int);
12076 vector unsigned int vec_sube (vector unsigned int,
12077 vector unsigned int,
12078 vector unsigned int);
12079 vector signed __int128 vec_sube (vector signed __int128,
12080 vector signed __int128,
12081 vector signed__int128);
12082 vector unsigned __int128 vec_sube (vector unsigned __int128,
12083 vector unsigned __int128,
12084 vector unsigned __int128);
12085 vector signed int vec_subec (vector signed int, vector signed int,
12086 vector signed int);
12087 vector unsigned int vec_subec (vector unsigned int,
12088 vector unsigned int,
12089 vector unsigned int);
12090 vector signed __int128 vec_subec (vector signed __int128,
12091 vector signed __int128,
12092 vector signed__int128);
12093 vector unsigned __int128 vec_subec (vector unsigned __int128,
12094 vector unsigned __int128,
12095 vector unsigned __int128);
12096 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12097 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12098 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12099 BU_ALTIVEC_OVERLOAD_X definitions.
12100 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12101 * doc/extend.texi: Update the built-in documentation file for the new
12102 built-in functions.
12104 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12107 * diagnostic-show-locus.c (layout::layout): Use start and finish
12108 spelling location for the start and finish of each range.
12109 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12110 Add unused aspect param.
12111 * input.c (expand_location_1): Add "aspect" param, and use it
12112 to access the correct part of the location.
12113 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12115 (expand_location_to_spelling_point): Likewise.
12116 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12117 param, and pass it to expand_location_1.
12119 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12121 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12122 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12123 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12124 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12125 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12126 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12127 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12128 _mm_maskz_getmant_ss): New intrinsics.
12129 (__builtin_ia32_getexpss128_mask): Changed to ...
12130 __builtin_ia32_getexpss128_round ... this.
12131 (__builtin_ia32_getexpsd128_mask): Changed to ...
12132 __builtin_ia32_getexpsd128_round ... this.
12133 * config/i386/i386-builtin-types.def
12134 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12135 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12136 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12137 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12138 __builtin_ia32_getmantss_mask_round): New builtins.
12139 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12140 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12141 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12142 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12143 * config/i386/sse.md
12144 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12145 avx512f_sgetexp<mode><mask_scalar_name>
12146 <round_saeonly_scalar_name> ... this.
12147 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12148 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12149 vgetexp<ssescalarmodesuffix>
12150 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12151 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12152 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12153 avx512f_vgetmant<mode><mask_scalar_name>
12154 <round_saeonly_scalar_name> ... this.
12155 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12156 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12157 vgetmant<ssescalarmodesuffix>
12158 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12159 %0<mask_scalar_operand4>, %1, %2
12160 <round_saeonly_scalar_mask_op4>, %3} ... this.
12161 * config/i386/subst.md (mask_scalar_operand4,
12162 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12163 round_saeonly_scalar_nimm_predicate): New subst attributes.
12165 2017-07-06 Julia Koval <julia.koval@intel.com>
12167 * config/i386/i386.c (ix86_erase_embedded_rounding):
12168 Remove code for old rounding pattern.
12170 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12172 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12174 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12176 * doc/sourcebuild.texi (Test Directives, Variants of
12177 dg-require-support): Add documentation for dg-require-stack-check.
12179 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12181 * config/i386/subst.md (mask_scalar, round_scalar,
12182 round_saeonly_scalar): New meta-templates.
12183 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12184 round_scalar_mask_operand3, round_scalar_mask_op3,
12185 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12186 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12187 round_saeonly_scalar_constraint,
12188 round_saeonly_scalar_prefix): New subst attribute.
12189 * config/i386/sse.md
12190 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12191 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12192 <round_scalar_name> ... this.
12193 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12194 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12195 <round_scalar_name> ... this.
12196 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12197 <sse>_vm<code><mode>3<mask_scalar_name>
12198 <round_saeonly_scalar_name> ... this.
12199 (v<plusminus_mnemonic><ssescalarmodesuffix>
12200 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12201 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12202 v<plusminus_mnemonic><ssescalarmodesuffix>
12203 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12204 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12205 (v<multdiv_mnemonic><ssescalarmodesuffix>
12206 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12207 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12208 v<multdiv_mnemonic><ssescalarmodesuffix>
12209 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12210 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12211 (v<maxmin_float><ssescalarmodesuffix>
12212 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12213 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12214 v<maxmin_float><ssescalarmodesuffix>
12215 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12216 %0<mask_scalar_operand3>, %1, %<iptr>2
12217 <round_saeonly_scalar_mask_op3>} ... this.
12219 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12221 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12222 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12224 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12225 Alan Hayward <alan.hayward@arm.com>
12226 David Sherwood <david.sherwood@arm.com>
12228 * combine.c (simplify_if_then_else): Remove "enum" before
12230 * compare-elim.c (can_eliminate_compare): Likewise.
12231 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12233 (aarch64_lookup_simd_builtin_type): Likewise.
12234 (aarch64_simd_builtin_type): Likewise.
12235 (aarch64_init_simd_builtin_types): Likewise.
12236 (aarch64_simd_expand_args): Likewise.
12237 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12239 (aarch64_reverse_mask): Likewise.
12240 (aarch64_simd_emit_reg_reg_move): Likewise.
12241 (aarch64_gen_adjusted_ldpstp): Likewise.
12242 (aarch64_ccmp_mode_to_code): Likewise.
12243 (aarch64_operands_ok_for_ldpstp): Likewise.
12244 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12245 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12247 (aarch64_min_divisions_for_recip_mul): Likewise.
12248 (aarch64_reassociation_width): Likewise.
12249 (aarch64_get_condition_code_1): Likewise.
12250 (aarch64_simd_emit_reg_reg_move): Likewise.
12251 (aarch64_simd_attr_length_rglist): Likewise.
12252 (aarch64_reverse_mask): Likewise.
12253 (aarch64_operands_ok_for_ldpstp): Likewise.
12254 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12255 (aarch64_gen_adjusted_ldpstp): Likewise.
12256 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12258 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12259 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12260 (arm_lookup_simd_builtin_type): Likewise.
12261 (arm_simd_builtin_type): Likewise.
12262 (arm_init_simd_builtin_types): Likewise.
12263 (arm_expand_builtin_args): Likewise.
12264 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12265 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12266 (ft32_setup_incoming_varargs): Likewise.
12267 (ft32_function_arg): Likewise.
12268 (ft32_function_arg_advance): Likewise.
12269 (ft32_pass_by_reference): Likewise.
12270 (ft32_arg_partial_bytes): Likewise.
12271 (ft32_valid_pointer_mode): Likewise.
12272 (ft32_addr_space_pointer_mode): Likewise.
12273 (ft32_addr_space_legitimate_address_p): Likewise.
12274 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12276 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12277 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12278 (iamcu_alignment): Likewise.
12279 (canonicalize_vector_int_perm): Likewise.
12280 (ix86_noce_conversion_profitable_p): Likewise.
12281 (ix86_mpx_bound_mode): Likewise.
12282 (ix86_operands_ok_for_move_multiple): Likewise.
12283 * config/microblaze/microblaze-protos.h
12284 (microblaze_expand_conditional_branch_reg): Likewise.
12285 * config/microblaze/microblaze.c
12286 (microblaze_expand_conditional_branch_reg): Likewise.
12287 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12289 (rs6000_reassociation_width): Likewise.
12290 (rs6000_invalid_binary_op): Likewise.
12291 (fusion_p9_p): Likewise.
12292 (emit_fusion_p9_load): Likewise.
12293 (emit_fusion_p9_store): Likewise.
12294 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12296 (riscv_hard_regno_mode_ok_p): Likewise.
12297 (riscv_address_insns): Likewise.
12298 (riscv_split_symbol): Likewise.
12299 (riscv_legitimize_move): Likewise.
12300 (riscv_function_value): Likewise.
12301 (riscv_hard_regno_nregs): Likewise.
12302 (riscv_expand_builtin): Likewise.
12303 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12304 (riscv_build_integer): Likewise.
12305 (riscv_split_integer): Likewise.
12306 (riscv_legitimate_constant_p): Likewise.
12307 (riscv_cannot_force_const_mem): Likewise.
12308 (riscv_regno_mode_ok_for_base_p): Likewise.
12309 (riscv_valid_base_register_p): Likewise.
12310 (riscv_valid_offset_p): Likewise.
12311 (riscv_valid_lo_sum_p): Likewise.
12312 (riscv_classify_address): Likewise.
12313 (riscv_legitimate_address_p): Likewise.
12314 (riscv_address_insns): Likewise.
12315 (riscv_load_store_insns): Likewise.
12316 (riscv_force_binary): Likewise.
12317 (riscv_split_symbol): Likewise.
12318 (riscv_force_address): Likewise.
12319 (riscv_legitimize_address): Likewise.
12320 (riscv_move_integer): Likewise.
12321 (riscv_legitimize_const_move): Likewise.
12322 (riscv_legitimize_move): Likewise.
12323 (riscv_address_cost): Likewise.
12324 (riscv_subword): Likewise.
12325 (riscv_output_move): Likewise.
12326 (riscv_canonicalize_int_order_test): Likewise.
12327 (riscv_emit_int_order_test): Likewise.
12328 (riscv_function_arg_boundary): Likewise.
12329 (riscv_pass_mode_in_fpr_p): Likewise.
12330 (riscv_pass_fpr_single): Likewise.
12331 (riscv_pass_fpr_pair): Likewise.
12332 (riscv_get_arg_info): Likewise.
12333 (riscv_function_arg): Likewise.
12334 (riscv_function_arg_advance): Likewise.
12335 (riscv_arg_partial_bytes): Likewise.
12336 (riscv_function_value): Likewise.
12337 (riscv_pass_by_reference): Likewise.
12338 (riscv_setup_incoming_varargs): Likewise.
12339 (riscv_print_operand): Likewise.
12340 (riscv_elf_select_rtx_section): Likewise.
12341 (riscv_save_restore_reg): Likewise.
12342 (riscv_for_each_saved_reg): Likewise.
12343 (riscv_register_move_cost): Likewise.
12344 (riscv_hard_regno_mode_ok_p): Likewise.
12345 (riscv_hard_regno_nregs): Likewise.
12346 (riscv_class_max_nregs): Likewise.
12347 (riscv_memory_move_cost): Likewise.
12348 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12349 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12350 (rl78_addr_space_address_mode): Likewise.
12351 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12353 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12354 (rs6000_reassociation_width): Likewise.
12355 (rs6000_invalid_binary_op): Likewise.
12356 (fusion_p9_p): Likewise.
12357 (emit_fusion_p9_load): Likewise.
12358 (emit_fusion_p9_store): Likewise.
12359 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12360 (ok_for_simple_move_operands): Likewise.
12361 (ok_for_simple_move_strict_operands): Likewise.
12362 (ok_for_simple_arith_logic_operands): Likewise.
12363 (visium_legitimize_reload_address): Likewise.
12364 (visium_select_cc_mode): Likewise.
12365 (output_cbranch): Likewise.
12366 (visium_split_double_move): Likewise.
12367 (visium_expand_copysign): Likewise.
12368 (visium_expand_int_cstore): Likewise.
12369 (visium_expand_fp_cstore): Likewise.
12370 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12371 (visium_function_arg): Likewise.
12372 (visium_function_arg_advance): Likewise.
12373 (visium_libcall_value): Likewise.
12374 (visium_setup_incoming_varargs): Likewise.
12375 (visium_legitimate_constant_p): Likewise.
12376 (visium_legitimate_address_p): Likewise.
12377 (visium_legitimize_address): Likewise.
12378 (visium_secondary_reload): Likewise.
12379 (visium_register_move_cost): Likewise.
12380 (visium_memory_move_cost): Likewise.
12381 (prepare_move_operands): Likewise.
12382 (ok_for_simple_move_operands): Likewise.
12383 (ok_for_simple_move_strict_operands): Likewise.
12384 (ok_for_simple_arith_logic_operands): Likewise.
12385 (visium_function_value_1): Likewise.
12386 (rtx_ok_for_offset_p): Likewise.
12387 (visium_legitimize_reload_address): Likewise.
12388 (visium_split_double_move): Likewise.
12389 (visium_expand_copysign): Likewise.
12390 (visium_expand_int_cstore): Likewise.
12391 (visium_expand_fp_cstore): Likewise.
12392 (visium_split_cstore): Likewise.
12393 (visium_select_cc_mode): Likewise.
12394 (visium_split_cbranch): Likewise.
12395 (output_cbranch): Likewise.
12396 (visium_print_operand_address): Likewise.
12397 * expmed.c (flip_storage_order): Likewise.
12398 * expmed.h (emit_cstore): Likewise.
12399 (flip_storage_order): Likewise.
12400 * genrecog.c (validate_pattern): Likewise.
12401 * hsa-gen.c (gen_hsa_addr): Likewise.
12402 * internal-fn.c (expand_arith_overflow): Likewise.
12403 * ira-color.c (allocno_copy_cost_saving): Likewise.
12404 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12405 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12406 (process_invariant_for_inheritance): Likewise.
12407 * lra-eliminations.c (move_plus_up): Likewise.
12408 * omp-low.c (lower_oacc_reductions): Likewise.
12409 * simplify-rtx.c (simplify_subreg): Likewise.
12410 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12411 (TARGET_CHKP_BOUND_MODE): Likewise..
12412 * targhooks.c (default_chkp_bound_mode): Likewise.
12413 (default_setup_incoming_vararg_bounds): Likewise.
12414 * targhooks.h (default_chkp_bound_mode): Likewise.
12415 (default_setup_incoming_vararg_bounds): Likewise.
12416 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12417 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12418 (have_whole_vector_shift): Likewise.
12419 * tree-vect-stmts.c (vectorizable_load): Likewise.
12420 * doc/tm.texi: Regenerate.
12422 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12424 Graceful degrade if Binutils PR21472 is not available.
12427 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12428 .rodata in flash test fails.
12429 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12430 * confgure: Regenerate.
12431 * config.in: Regenerate.
12432 * config/avr/avr.c (avr_asm_named_section)
12433 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12434 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12435 (avr_asm_init_sections): Same.
12437 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12439 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12440 (fma<VH:mode>4_intrinsic): Likewise.
12441 (*fmsub<VCVTF:mode>4): Likewise.
12442 (*fmsub<VH:mode>4_intrinsic): Likewise.
12444 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12447 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12448 Don't depend on "optimize > 0".
12449 (out_movhi_r_mr, out_movqi_mr_r): Same.
12450 (out_movhi_mr_r, out_movqi_r_mr): Same.
12451 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12452 io_address_operand on "optimize > 0".
12454 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12456 * tree-loop-distribution.c: Add general explanantion on the pass.
12457 (generate_loops_for_partition): Mark distributed loop.
12458 (pg_add_dependence_edges): New parameter. Handle alias data
12459 dependence specially and record it in the parameter if asked.
12460 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12461 (init_partition_graph_vertices, add_partition_graph_edge): New.
12462 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12463 (free_partition_graph_vdata, build_partition_graph): New.
12464 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12465 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12466 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12467 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12468 (version_for_distribution_p, finalize_partitions): New.
12469 (distribute_loop): Handle alias data dependence specially. Factor
12470 out loop fusion code as functions and call these functions.
12472 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12474 * tree-loop-distribution.c (classify_partition): New parameter and
12475 better handle reduction statement.
12476 (rdg_build_partitions): Revise comment.
12477 (distribute_loop): Compute statements in all partitions and pass it
12478 to classify_partition.
12480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12482 * tree-loop-distribution.c (enum partition_type): New.
12483 (struct partition): New field type.
12484 (partition_merge_into): Add parameter. Update partition type.
12485 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12486 (build_rdg_partition_for_vertex): Compute partition type.
12487 (rdg_build_partitions): Dump partition type.
12488 (distribute_loop): Update calls to partition_merge_into.
12490 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12492 * tree-loop-distribution.c (struct ddr_hasher): New.
12493 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12495 (classify_partition): Call get_data_dependence.
12496 (pg_add_dependence_edges): Ditto.
12497 (distribute_loop): Release data dependence hash table.
12499 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12501 * tree-loop-distribution.c (ref_base_address): Delete.
12502 (similar_memory_accesses): Rename ...
12503 (share_memory_accesses): ... to this. Check if partitions access
12504 the same memory reference.
12505 (distribute_loop): Call share_memory_accesses.
12507 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12509 * tree-loop-distribution.c (struct partition): New field recording
12510 its data reference.
12511 (partition_alloc, partition_free): Init and release data refs.
12512 (partition_merge_into): Merge data refs.
12513 (build_rdg_partition_for_vertex): Collect data refs for partition.
12514 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12516 (distribute_loop): Remve data refs from vertice data of partition
12519 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12521 * tree-loop-distribution.c (params.h): Include header file.
12522 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12523 (datarefs_vec): New global var.
12524 (create_rdg_vertices): Use datarefs_vec directly.
12525 (free_rdg): Don't free data references.
12526 (build_rdg): Update use. Don't free data references.
12527 (distribute_loop): Compute global variable for data references.
12528 Bail out if there are too many data references.
12530 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12532 * tree-loop-distribution.c (loop_nest): New global var.
12533 (build_rdg): Use loop directly, rather than loop nest.
12534 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12536 (distribute_loop): Compute global variable loop nest. Update use.
12538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12540 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12541 (partition_merge_into): New parameter. Dump reason for fusion.
12542 (distribute_loop): Update use of partition_merge_into.
12544 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12546 * tree-loop-distribution.c (bb_top_order_index): New.
12547 (bb_top_order_index_size, bb_top_order_cmp): New.
12548 (stmts_from_loop): Use topological order.
12549 (pass_loop_distribution::execute): Compute and release topological
12550 order for basic blocks.
12552 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12554 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12557 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12559 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12560 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12561 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12562 * internal-fn.def (LOOP_DIST_ALIAS): New.
12563 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12564 (fold_loop_internal_call): ... this.
12565 (vect_loop_dist_alias_call): New function.
12566 (set_uid_loop_bbs): Call fold_loop_internal_call.
12567 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12570 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12573 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12574 Require dead FLAGS_REG at the beginning of a peephole.
12576 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12579 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12580 arguments in the call to __builtin_ia32_sbb_u32.
12581 (_subborrow_u64): Swap _X and _Y arguments in the call to
12582 __builtin_ia32_sbb_u64.
12584 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12587 * tree-vrp.c (compare_assert_loc): Turn into a function template
12588 with stable template parameter. Only test if a->e is NULL,
12589 !a->e == !b->e has been verified already. Use e == NULL or
12590 e != NULL instead of e or ! e tests. If stable is true, don't use
12591 iterative_hash_expr, on the other side allow a or b or both NULL
12592 and sort the NULLs last.
12593 (process_assert_insertions): Sort using compare_assert_loc<false>
12594 instead of compare_assert_loc, later sort using
12595 compare_assert_loc<true> before calling process_assert_insertions_for
12596 in a loop. Use break instead of continue once seen NULL pointer.
12598 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12600 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12601 Cortex-R7 and Cortex-R8 processors.
12603 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12605 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12606 uninitialized while src is not.
12608 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12610 * common/config/arm/arm-common.c: Adjust include path for
12612 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12613 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12614 (arm-cpu-data.h): Likewise.
12615 (arm-cpu-cdata.h): Likewise.
12616 * config/arm/arm-cpu.h: Delete.
12617 * config/arm/arm-cpu-cdata.h: Delete.
12618 * config/arm/arm-cpu-data.h: Delete.
12620 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12622 * config/arm/arm-cpus.in (cortex-a55): New.
12623 (cortex-a75): Likewise.
12624 (cortex-a75.cortex-a55): Likewise.
12625 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12627 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12628 * config/arm/arm-cpu-cdata.h: Regenerate.
12629 * config/arm/arm-cpu-data.h: Regenerate.
12630 * config/arm/arm-cpu.h: Regenerate.
12631 * config/arm/arm-tables.opt: Regenerate.
12632 * config/arm/arm-tune.md: Regenerate.
12634 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12636 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12638 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12640 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12643 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12645 PR tree-optimization/81292
12646 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12647 full_string_p, also call adjust_related_strinfos if the adjustment
12648 is simple, otherwise invalidate related strinfos.
12650 2017-07-04 Martin Liska <mliska@suse.cz>
12653 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12654 newly created variable as DECL_IGNORED_P.
12656 2017-07-04 Martin Liska <mliska@suse.cz>
12659 * ipa-inline.c (inline_small_functions):
12660 Use xstrdup_for_dump.
12662 2017-07-04 Tom de Vries <tom@codesourcery.com>
12664 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12666 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12669 * config/darwin.c (darwin_function_switched_text_sections):
12672 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12674 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12676 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12678 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12680 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12682 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12683 min_profitable_iters, and th as inclusive lower bounds.
12684 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12685 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12686 for min_profitable_iters and min_profitable_estimate.
12687 (vect_transform_loop): Treat th as an inclusive lower bound.
12688 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12690 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12693 * config/darwin.c (darwin_function_switched_text_sections):
12694 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12695 in two pieces, and suppress the use of buf.
12697 2017-07-03 Nathan Sidwell <nathan@acm.org>
12699 * hash-table.h (hash_table_mod1): Fix indentation.
12701 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12703 PR middle-end/81290
12704 * predict.c (force_edge_cold): Be more careful about propagation
12706 * profile-count.h (profile_probability::guessed,
12707 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12709 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12711 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12713 * doc/invoke.texi (rcpc architecture extension): Document it.
12715 2017-07-03 Richard Biener <rguenther@suse.de>
12717 PR tree-optimization/60510
12718 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12719 the scalar reduction PHI and use it.
12720 (vectorizable_reduction): Properly guard the single_defuse_cycle
12721 path for non-SLP reduction chains where we cannot use it.
12722 Rework reduc_def/index and vector type deduction. Rework
12723 vector operand gathering during reduction op code-gen.
12724 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12725 chains dissolve the chain and leave it to non-SLP reduction
12728 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12730 * tree-data-ref.h (dr_alignment): Declare.
12731 * tree-data-ref.c (dr_alignment): New function.
12732 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12733 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12735 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12737 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12739 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12740 and base_misalignment fields.
12741 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12742 * tree-data-ref.c: Include builtins.h.
12743 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12744 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12745 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12746 * tree-vect-data-refs.c: Include tree-cfg.h.
12747 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12748 fields instead of calculating an alignment here.
12749 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12750 innermost_loop_behavior fields.
12752 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12754 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12756 (DR_STEP_ALIGNMENT): New macro.
12757 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12758 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12759 (create_data_ref): Print it.
12760 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12761 to tell whether the step preserves vector (mis)alignment.
12762 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12763 Move the check for an integer step and generalise to all INTEGER_CST.
12764 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12765 Print the outer step alignment.
12767 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12769 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12770 with offset_alignment.
12771 (DR_ALIGNED_TO): Delete.
12772 (DR_OFFSET_ALIGNMENT): New macro.
12773 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12774 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12775 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12776 (create_data_ref): Likewise.
12777 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12778 (vect_analyze_data_refs): Likewise.
12779 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12780 creating dummy innermost behavior.
12782 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12784 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12785 with a "innermost_loop_behavior *" and refeence tree.
12786 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12787 (create_data_ref): Update call accordingly.
12788 * tree-predcom.c (find_looparound_phi): Likewise.
12790 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12792 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12793 fields with dr_wrt_vec_loop.
12794 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12795 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12796 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12797 (vect_dr_behavior): New function.
12798 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12799 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12800 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12801 track whether the step preserves the misalignment.
12802 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12803 Use vect_dr_behavior.
12804 (vect_setup_realignment): Update call accordingly.
12805 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12806 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12807 call to vect_create_addr_base_for_vector_ref.
12808 (vect_create_cond_for_align_checks): Likewise.
12809 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12810 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12811 (vect_recog_mask_conversion_pattern): Likewise.
12812 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12813 (new_stmt_vec_info): Remove redundant zeroing.
12815 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12817 * common/config/arm/arm-common.c (arm_be8_option): New function.
12818 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12819 (ISA_ARMv6): Add isa_bit_be8.
12820 * config/arm/arm.h (arm_be8_option): Add prototype.
12821 (BE8_SPEC_FUNCTION): New define.
12822 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12823 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12824 (mlittle-endian): Similarly.
12825 (mbe8, mbe32): New options.
12826 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12827 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12829 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12831 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12833 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12835 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12836 (cleanup_tree_cfg_bb): Use it.
12837 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12839 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12841 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12844 * loop-doloop.c (add_test): Update profile.
12846 2017-07-03 Martin Liska <mliska@suse.cz>
12849 * sanopt.c (rewrite_usage_of_param): New function.
12850 (sanitize_rewrite_addressable_params): Likewise.
12851 (pass_sanopt::execute): Call rewrite_usage_of_param.
12853 2017-07-03 Richard Biener <rguenther@suse.de>
12855 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12856 back to using VIEW_CONVERT_EXPR.
12858 2017-07-03 Martin Liska <mliska@suse.cz>
12861 * doc/extend.texi: Document when a resolver function is
12862 generated for target_clones.
12864 2017-07-03 Martin Liska <mliska@suse.cz>
12866 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12867 and shadow memory of auto variables which are subject of
12868 use-after-scope sanitization.
12869 (asan_expand_mark_ifn): Add do set only when is_poison.
12871 2016-07-03 Richard Biener <rguenther@suse.de>
12873 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12875 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12876 (vectorizable_reduction): Perform reduction PHI creation when
12877 visiting a reduction PHI and adjust and simplify code generation
12878 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12879 (vect_transform_loop): Visit reduction PHIs.
12880 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12881 defs into the SLP tree.
12882 (vect_build_slp_tree): Reduction defs terminate the recursion.
12883 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12885 (vect_get_vec_defs_for_stmt_copy): Export.
12886 (vect_get_vec_defs): Likewise.
12887 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12889 (vect_get_vec_defs_for_stmt_copy): Declare.
12890 (vect_get_vec_defs): Likewise.
12892 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12894 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12895 parameter with a "loop" parameter and use it instead of the
12896 loop containing DR_STMT. Don't check simple_iv when doing
12897 BB analysis. Describe the two analysis modes in the comment.
12899 2017-07-03 Tom de Vries <tom@codesourcery.com>
12901 PR tree-optimization/69468
12902 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12903 (find_same_succ_bb): Handle ignore_edge_flags.
12905 2017-07-03 Tom de Vries <tom@codesourcery.com>
12907 PR tree-optimization/81192
12908 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12910 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12912 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12914 2017-07-03 Tom de Vries <tom@codesourcery.com>
12916 PR tree-optimization/81192
12917 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12918 BB_SAME_SUCC (bb) == NULL.
12920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12922 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12925 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12927 * dumpfile.c: Include profile-count.h
12928 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12930 (insert_cond_bb): Update profile.
12931 * tree-cfg.h (insert_cond_bb): Update prototype.
12932 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12933 * tree-dump.c: Do not include tree-cfg.
12935 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12937 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12939 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12941 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12944 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12946 * tree-complex.c (expand_complex_div_wide): update profile.
12948 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12949 Alan Hayward <alan.hayward@arm.com>
12950 David Sherwood <david.sherwood@arm.com>
12952 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12953 (CORETYPES_H): New define.
12954 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12955 (insn-modes-inline.h, s-modes-inline-h): New rules.
12956 (generated_files): Add insn-modes-inline.h.
12957 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12958 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12959 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12960 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12961 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12962 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12963 (build/gencodes.o, build/genconditions.o): Likewise.
12964 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12965 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12966 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12967 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12968 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12969 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12970 * coretypes.h: Include everything up to real.h for generators.
12971 Include insn-modes.h first. Include wide-int-print.h after
12972 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12973 * machmode.h: Don't include insn-modes.h here.
12974 * function-tests.c: Remove includes of signop.h, machmode.h,
12975 double-int.h and wide-int.h.
12977 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12979 * optc-save-gen.awk: Likewise.
12980 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12981 * godump.c: Remove include of wide-int-print.h.
12982 * pretty-print.h: Likewise.
12983 * wide-int-print.cc: Likewise.
12984 * wide-int.cc: Likewise.
12985 * hash-map-tests.c: Remove include of signop.h.
12986 * hash-set-tests.c: Likewise.
12987 * rtl-tests.c: Likewise.
12988 * mkconfig.sh: Remove include of machmode.h.
12989 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12991 (emit_insn_modes_inline_h): ...this new function. Emit the code
12992 into an insn-modes-inline.h header file, adding appropriate
12993 include guards and end comments.
12994 (emit_insn_modes_c_header): Remove include of machmode.h.
12995 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12997 (main): Handle -i flag and call emit_insn_modes_inline_h when
13000 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13002 * tree-ssa-strlen.c (strinfo): Rename the length field to
13003 nonzero_chars. Add a full_string_p field.
13004 (compare_nonzero_chars, zero_length_string_p): New functions.
13005 (get_addr_stridx): Add an offset_out parameter.
13006 Use compare_nonzero_chars.
13007 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13008 (new_strinfo): Update after above changes to strinfo.
13009 (set_endptr_and_length): Set full_string_p.
13010 (get_string_length): Update after above changes to strinfo.
13011 (unshare_strinfo): Update call to new_strinfo.
13012 (maybe_invalidate): Likewise.
13013 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13014 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13015 as a uhwi instead of an shwi. Update after above changes to
13016 strinfo and new_strinfo.
13017 (zero_length_string): Assert that chainsi contains full strings.
13018 Use zero_length_string_p. Update call to new_strinfo.
13019 (adjust_related_strinfos): Update after above changes to strinfo.
13020 Copy full_string_p from origsi.
13021 (adjust_last_stmt): Use zero_length_string_p.
13022 (handle_builtin_strlen): Update after above changes to strinfo and
13023 new_strinfo. Install the lhs as the string length if the previous
13024 entry didn't describe a full string.
13025 (handle_builtin_strchr): Update after above changes to strinfo
13027 (handle_builtin_strcpy): Likewise.
13028 (handle_builtin_strcat): Likewise.
13029 (handle_builtin_malloc): Likewise.
13030 (handle_pointer_plus): Likewise.
13031 (handle_builtin_memcpy): Likewise. Track nonzero characters
13032 that aren't necessarily followed by a nul terminator.
13033 (handle_char_store): Likewise.
13035 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13037 PR tree-optimization/80769
13038 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13039 for malloc and calloc. Document the new invariant that all related
13040 strinfos have delayed lengths or none do.
13041 (verify_related_strinfos): Move earlier in file.
13042 (set_endptr_and_length): New function, split out from...
13043 (get_string_length): ...here. Also set the lengths of related
13045 (zero_length_string): Assert that chainsi has known (rather than
13047 (adjust_related_strinfos): Likewise.
13049 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13051 PR tree-optimization/81136
13052 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13053 assert that two references with the same misalignment have the same
13054 compile-time misalignment if those compile-time misalignments
13057 2017-07-01 Andi Kleen <ak@linux.intel.com>
13059 * print-tree.c (print_node): Print all attributes.
13061 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13063 * cfg.c (scale_bbs_frequencies): New function.
13064 * cfg.h (scale_bbs_frequencies): Declare it.
13065 * cfgloopanal.c (single_likely_exit): Cleanup.
13066 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13068 (scale_loop_profile): Likewise.
13069 (loop_version): Likewise.
13070 (create_empty_loop_on_edge): Update.
13071 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13072 scale_loop_frequencies, scale_loop_profile, loopify,
13073 loop_version): Update prototypes.
13074 * modulo-sched.c (sms_schedule): Update.
13075 * predict.c (unlikely_executed_edge_p): Also check probability.
13076 (probably_never_executed_edge_p): Fix typo.
13077 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13078 * tree-parloops.c (gen_parallel_loop): Update.
13079 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13080 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13081 * tree-ssa-loop-split.c (split_loop): Update.
13082 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13083 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13084 (vect_loop_versioning): Update.
13085 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13087 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13089 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13091 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13093 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13094 to keep profile consistent.
13096 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13098 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13099 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13100 * profile-count.h (max_safe_multiplier): Make unsigned.
13101 (profile_count::guessed_zero): New.
13103 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13105 * bb-reorder.c (fix_up_crossing_landing_pad,
13106 fix_crossing_conditional_branches): Use make_single_succ_edge
13107 to keep profile consistent.
13109 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13111 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13114 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13117 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13118 the right scopes, make sure cond_jump isn't preserved between multiple
13119 iterations. Search for fallthru edge whenever there are 3+ edges and
13120 use find_fallthru_edge for it.
13122 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13124 Patch by Alexander Monakov <amonakov@ispras.ru>
13125 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13126 probabilities consistently.
13128 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13130 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13131 emit_cmp_and_jump_insns.
13133 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13136 * tree-inline.c (expand_call_inline): Combine profile statuses.
13138 2017-06-30 Andrew Pinski <apinski@cavium.com>
13140 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13141 fold_stmt returned true.
13143 2017-06-30 Nathan Sidwell <nathan@acm.org>
13145 * ggc.h (empty_string): Delete.
13146 * cfgexpand.c (expand_asm_stmt): Use plain "".
13147 * optabs.c (expand_asm_memory_barrier): Likewise.
13148 * stringpool.c (empty_string): Delete.
13149 (digit_vector, digit_string): Delete.
13150 (ggc_alloc_string): Use plain "", don't optimize single digit
13151 strings. Use ggc_alloc_atomic.
13153 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13155 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13156 comparison set and one other set, use the cost of the non-comparison
13159 2017-06-30 Nathan Sidwell <nathan@acm.org>
13161 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13164 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13166 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13167 loops. Remove now unneeded calls to gimple_switch_set_label() that
13168 just set removed labels to NULL_TREE.
13170 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13172 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13173 (set_range_info): ...here. Only call set_range_info_raw if domain
13175 (set_nonzero_bits): Call set_range_info_raw.
13176 * tree-ssanames.h (set_range_info_raw): New.
13178 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13181 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13182 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13183 of nonimmediate_operand and <store_mask_constraint> instead of m
13184 for the input operand. For V8FI iterator, always split if input
13185 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13186 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13187 <store_mask_predicate> instead of register_operand and
13188 <store_mask_constraint> instead of v for the input operand. Make
13189 sure both operands aren't MEMs for if not <mask_applied>.
13191 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13193 * lto-wrapper.c (copy_file) Close both file descriptors before
13196 2017-06-30 Martin Liska <mliska@suse.cz>
13199 * multiple_target.c (create_dispatcher_calls): Make ifunc
13200 also for function that don't have calls or are not referenced.
13202 2017-06-30 Richard Biener <rguenther@suse.de>
13204 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13205 analyze the first scalar stmt. Move vector type computation
13206 for the BB case here from ...
13207 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13208 live operation processing in the SLP case properly.
13210 2017-06-30 Richard Biener <rguenther@suse.de>
13212 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13214 2017-06-30 Martin Liska <mliska@suse.cz>
13217 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13218 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13220 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13222 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13225 2017-06-30 Richard Biener <rguenther@suse.de>
13227 PR tree-optimization/81249
13228 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13229 condition reduction result to original scalar type.
13231 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13233 * profile-count.h (enum profile_quality): Fix typos and whitespace
13236 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13238 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13239 type for branch probabilities.
13241 2017-06-29 Julian Brown <julian@codesourcery.com>
13242 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13244 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13245 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13246 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13247 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13249 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13251 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13252 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13253 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13254 CC usage from generic code to here.
13255 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13256 CC usage into the target macros.
13258 2017-06-29 Maya Rashish <coypu@sdf.org>
13260 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13263 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13265 * arm/arm-builtins.c: Include profile-count.h
13266 * except.c (sjlj_emit_function_enter): Use
13267 profile_probability::unlikely.
13269 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13271 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13272 and tocrel_offset be pointer args rather than implicitly using
13274 (legitimate_constant_pool_address_p, rs6000_emit_move,
13275 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13276 tocrel_offset and use in toc_relative_expr_p call.
13277 (print_operand, print_operand_address): Use static tocrel_base_oac
13278 and tocrel_offset_oac.
13279 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13282 2017-06-29 Maya Rashish <coypu@sdf.org>
13284 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13286 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13288 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13289 objects, take into account only the alignment of 'op0' and 'mode1' if
13292 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13294 * ccmp.c (ccmp_tree_comparison_p): New function.
13295 (ccmp_candidate_p): Update to use above function.
13296 (get_compare_parts): New function.
13297 (expand_ccmp_next): Update to use new functions.
13298 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13300 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13301 take mode as argument.
13302 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13303 * expr.c (expand_expr_real_1): Pass mode as argument.
13305 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13307 * combine.c (combine_instructions): Print insns to dump_file, together
13310 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13312 * asan.c (asan_emit_stack_protection): Update.
13313 (create_cond_insert_point): Update.
13314 * auto-profile.c (afdo_propagate_circuit): Update.
13315 * basic-block.h (struct edge_def): Turn probability to
13316 profile_probability.
13317 (EDGE_FREQUENCY): Update.
13318 * bb-reorder.c (find_traces_1_round): Update.
13319 (better_edge_p): Update.
13320 (sanitize_hot_paths): Update.
13321 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13322 (make_single_succ_edge): Update.
13323 (check_bb_profile): Update.
13324 (dump_edge_info): Update.
13325 (update_bb_profile_for_threading): Update.
13326 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13327 probabilitycount to 0.
13328 * cfgbuild.c (compute_outgoing_frequencies): Update.
13329 * cfgcleanup.c (try_forward_edges): Update.
13330 (outgoing_edges_match): Update.
13331 (try_crossjump_to_edge): Update.
13332 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13333 (expand_gimple_tailcall): Update.
13334 (construct_init_block): Use make_single_succ_edge.
13335 (construct_exit_block): Use make_single_succ_edge.
13336 * cfghooks.c (verify_flow_info): Update.
13337 (redirect_edge_succ_nodup): Update.
13338 (split_edge): Update.
13339 (account_profile_record): Update.
13340 * cfgloopanal.c (single_likely_exit): Update.
13341 * cfgloopmanip.c (scale_loop_profile): Update.
13342 (set_zero_probability): Remove.
13343 (duplicate_loop_to_header_edge): Update.
13344 * cfgloopmanip.h (loop_version): Update prototype.
13345 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13346 (force_nonfallthru_and_redirect): Update.
13347 (update_br_prob_note): Update.
13348 (rtl_verify_edges): Update.
13349 (purge_dead_edges): Update.
13350 (rtl_lv_add_condition_to_bb): Update.
13351 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13352 * cgraphunit.c (init_lowered_empty_function): Update.
13353 (cgraph_node::expand_thunk): Update.
13354 * cilk-common.c: Include profile-count.h
13355 * dojump.c (inv): Remove.
13356 (jumpifnot): Update.
13357 (jumpifnot_1): Update.
13358 (do_jump_1): Update.
13360 (do_jump_by_parts_greater_rtx): Update.
13361 (do_compare_rtx_and_jump): Update.
13362 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13363 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13364 * dwarf2cfi.c: Include profile-count.h
13365 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13366 (sjlj_emit_dispatch_table): Likewise.
13367 * explow.c: Include profile-count.h
13368 * expmed.c (emit_store_flag_force): Update.
13369 (do_cmp_and_jump): Update.
13370 * expr.c (compare_by_pieces_d::generate): Update.
13371 (compare_by_pieces_d::finish_mode): Update.
13372 (emit_block_move_via_loop): Update.
13373 (store_expr_with_bounds): Update.
13374 (store_constructor): Update.
13375 (expand_expr_real_2): Update.
13376 (expand_expr_real_1): Update.
13377 * expr.h (try_casesi, try_tablejump): Update prototypes.
13378 * gimple-pretty-print.c (dump_probability): Update.
13379 (dump_profile): New.
13380 (dump_gimple_label): Update.
13381 (dump_gimple_bb_header): Update.
13382 * graph.c (draw_cfg_node_succ_edges): Update.
13383 * hsa-gen.c (convert_switch_statements): Update.
13384 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13385 (find_if_case_1): Update.
13386 (find_if_case_2): Update.
13387 * internal-fn.c (expand_arith_overflow_result_store): Update.
13388 (expand_addsub_overflow): Update.
13389 (expand_neg_overflow): Update.
13390 (expand_mul_overflow): Update.
13391 (expand_vector_ubsan_overflow): Update.
13392 * ipa-cp.c (good_cloning_opportunity_p): Update.
13393 * ipa-split.c (split_function): Use make_single_succ_edge.
13394 * ipa-utils.c (ipa_merge_profiles): Update.
13395 * loop-doloop.c (add_test): Update.
13396 (doloop_modify): Update.
13397 * loop-unroll.c (compare_and_jump_seq): Update.
13398 (unroll_loop_runtime_iterations): Update.
13399 * lra-constraints.c (lra_inheritance): Update.
13400 * lto-streamer-in.c (input_cfg): Update.
13401 * lto-streamer-out.c (output_cfg): Update.
13402 * mcf.c (adjust_cfg_counts): Update.
13403 * modulo-sched.c (sms_schedule): Update.
13404 * omp-expand.c (expand_omp_for_init_counts): Update.
13405 (extract_omp_for_update_vars): Update.
13406 (expand_omp_ordered_sink): Update.
13407 (expand_omp_for_ordered_loops): Update.
13408 (expand_omp_for_generic): Update.
13409 (expand_omp_for_static_nochunk): Update.
13410 (expand_omp_for_static_chunk): Update.
13411 (expand_cilk_for): Update.
13412 (expand_omp_simd): Update.
13413 (expand_omp_taskloop_for_outer): Update.
13414 (expand_omp_taskloop_for_inner): Update.
13415 * omp-simd-clone.c (simd_clone_adjust): Update.
13416 * optabs.c (expand_doubleword_shift): Update.
13417 (expand_abs): Update.
13418 (emit_cmp_and_jump_insn_1): Update.
13419 (expand_compare_and_swap_loop): Update.
13420 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13421 * predict.c (predictable_edge_p): Update.
13422 (edge_probability_reliable_p): Update.
13423 (set_even_probabilities): Update.
13424 (combine_predictions_for_insn): Update.
13425 (combine_predictions_for_bb): Update.
13426 (propagate_freq): Update.
13427 (estimate_bb_frequencies): Update.
13428 (force_edge_cold): Update.
13429 * profile-count.c (profile_count::dump): Add missing space into dump.
13430 (profile_count::debug): Add newline.
13431 (profile_count::differs_from_p): Explicitly convert to unsigned.
13432 (profile_count::stream_in): Update.
13433 (profile_probability::dump): New member function.
13434 (profile_probability::debug): New member function.
13435 (profile_probability::differs_from_p): New member function.
13436 (profile_probability::differs_lot_from_p): New member function.
13437 (profile_probability::stream_in): New member function.
13438 (profile_probability::stream_out): New member function.
13439 * profile-count.h (profile_count_quality): Rename to ...
13440 (profile_quality): ... this one.
13441 (profile_probability): New.
13442 (profile_count): Update.
13443 * profile.c (compute_branch_probabilities): Update.
13444 * recog.c (peep2_attempt): Update.
13445 * sched-ebb.c (schedule_ebbs): Update.
13446 * sched-rgn.c (find_single_block_region): Update.
13447 (compute_dom_prob_ps): Update.
13448 (schedule_region): Update.
13449 * sel-sched-ir.c (compute_succs_info): Update.
13450 * stmt.c (struct case_node): Update.
13451 (do_jump_if_equal): Update.
13452 (get_outgoing_edge_probs): Update.
13453 (conditional_probability): Update.
13454 (emit_case_dispatch_table): Update.
13455 (expand_case): Update.
13456 (expand_sjlj_dispatch_table): Update.
13457 (emit_case_nodes): Update.
13458 * targhooks.c: Update.
13459 * tracer.c (better_p): Update.
13460 (find_best_successor): Update.
13461 * trans-mem.c (expand_transaction): Update.
13462 * tree-call-cdce.c: Update.
13463 * tree-cfg.c (gimple_split_edge): Upate.
13464 (move_sese_region_to_fn): Upate.
13465 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13466 * tree-eh.c (lower_resx): Upate.
13467 (cleanup_empty_eh_move_lp): Upate.
13468 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13469 * tree-inline.c (copy_edges_for_bb): Update.
13470 (copy_cfg_body): Update.
13471 * tree-parloops.c (gen_parallel_loop): Update.
13472 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13473 (gimple_gen_time_profiler): Update.
13474 * tree-ssa-dce.c (remove_dead_stmt): Update.
13475 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13476 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13477 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13478 (unloop_loops): Update.
13479 (try_peel_loop): Update.
13480 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13481 * tree-ssa-loop-split.c (connect_loops): Update.
13482 (split_loop): Update.
13483 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13484 (hoist_guard): Update.
13485 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13486 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13487 (value_replacement): Update.
13488 * tree-ssa-reassoc.c (branch_fixup): Update.
13489 * tree-ssa-tail-merge.c (replace_block_by): Update.
13490 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13491 (create_edge_and_update_destination_phis): Update.
13492 (compute_path_counts): Update.
13493 (recompute_probabilities): Update.
13494 (update_joiner_offpath_counts): Update.
13495 (freqs_to_counts_path): Update.
13496 (duplicate_thread_path): Update.
13497 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13498 (struct switch_conv_info): Update.
13499 (gen_inbound_check): Update.
13500 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13501 (vect_do_peeling): Update.
13502 (vect_loop_versioning): Update.
13503 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13504 (optimize_mask_stores): Update.
13505 * ubsan.c (ubsan_expand_null_ifn): Update.
13506 * value-prof.c (gimple_divmod_fixed_value): Update.
13507 (gimple_divmod_fixed_value_transform): Update.
13508 (gimple_mod_pow2): Update.
13509 (gimple_mod_pow2_value_transform): Update.
13510 (gimple_mod_subtract): Update.
13511 (gimple_mod_subtract_transform): Update.
13512 (gimple_ic): Update.
13513 (gimple_stringop_fixed_value): Update.
13514 (gimple_stringops_transform): Update.
13515 * value-prof.h: Update.
13517 2017-06-29 Carl Love <cel@us.ibm.com>
13519 * config/rs6000/rs6000-c.c: Add support for built-in functions
13520 vector signed int vec_signed (vector float);
13521 vector signed long long vec_signed (vector double);
13522 vector signed int vec_signed2 (vector double, vector double);
13523 vector signed int vec_signede (vector double);
13524 vector signed int vec_signedo (vector double);
13525 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13526 instruction generator.
13527 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13528 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13529 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13531 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13532 vunsignede_v2df): Add define_expands.
13533 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13534 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13535 VEC_UNSIGNEDO): Add definitions.
13536 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13537 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13538 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13539 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13540 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13541 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13542 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13543 vec_unsignede, vec_unsignedo): Add builtin defines.
13544 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13546 * doc/extend.texi: Update the built-in documentation file for the
13547 new built-in functions.
13549 2017-06-29 Richard Biener <rguenther@suse.de>
13551 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13552 reduction chains to LOOP_VINFO_REDUCTIONS.
13553 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13554 SLP reductions after processing reduction chains.
13556 2017-06-29 Nathan Sidwell <nathan@acm.org>
13558 * builtins.c (fold_builtin_FUNCTION): Use
13559 lang_hooks.decl_printable_name.
13561 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13563 PR middle-end/81194
13564 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13565 with only one label.
13566 * stmt.c (expand_case): Assert NCASES is greater than one.
13568 2017-06-29 Richard Biener <rguenther@suse.de>
13570 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13572 (group_case_labels): Likewise.
13573 (find_taken_edge): Push sanity checking on val to workers...
13574 (find_taken_edge_cond_expr): ... here
13575 (find_taken_edge_switch_expr): ... and here, handle cases
13576 with just a default label.
13577 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13578 (group_case_labels): Likewise.
13579 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13580 group_case_labels does anything cleanup the CFG again.
13582 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13584 PR tree-optimization/81196
13585 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13586 exit condition comparing two IVs.
13588 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13590 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13591 profile to the dummy entry at the end of the list of architectures.
13592 * config/arm/arm-cpu-cdata.h: Regenerated.
13594 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13595 Michael Collison <michael.collison@arm.com>
13598 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13600 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13601 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13602 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13603 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13604 of shift when the shift amount is masked with constant equal to
13605 the size of the mode.
13606 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13609 2017-06-29 Martin Liska <mliska@suse.cz>
13611 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13613 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13616 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13619 (edge_growth_cache_entry::edge_growth_cache_entry): New
13621 (reset_edge_growth_cache): Update to use constructor.
13623 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13625 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13626 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13627 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13629 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13631 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13632 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13634 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13636 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13637 (*-linux-uclibc*): Add t-uclibc tmake_file.
13638 * config/t-musl: New.
13639 * config/t-uclibc: New.
13641 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13643 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13645 (gen_comm_data): Emit architectural setting of arch_prof.
13646 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13648 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13649 (armv8-m.base, armv8-m.main): Likewise.
13650 * arm-protos.h (arm_build_target): Add profile field.
13651 (arch_option): Likewise.
13652 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13654 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13655 arm_active_target.profile.
13657 2017-06-28 Richard Biener <rguenther@suse.de>
13659 PR middle-end/81227
13660 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13661 TYPE_OVERFLOW_WRAPS.
13662 * match.pd (negate_expr_p): Likewise.
13663 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13664 fold_build2, not fold_binary.
13666 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13668 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13669 Convert memory address to Pmode.
13670 (aarch64_print_operand): Assert MEM operands are always Pmode.
13672 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13675 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13676 Remove redundant if.
13677 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13678 * config/arm/aarch-common-protos.h
13679 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13680 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13682 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13685 * multiple_target.c (create_dispatcher_calls): Set the default
13686 clone to be static, not public.
13688 2017-06-28 Richard Biener <rguenther@suse.de>
13690 * tree-vect-loop.c (vectorizable_reduction): Move special
13691 cond reduction IV var creation ...
13692 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13693 parameter. Use STMT_VINFO_VECTYPE.
13694 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13697 2017-06-28 Martin Liska <mliska@suse.cz>
13700 * ipa-visibility.c (non_local_p): Handle visibility.
13702 2017-06-28 Martin Liska <mliska@suse.cz>
13705 * common.opt: Add IntegerRange to various options.
13706 * opt-functions.awk (integer_range_info): New function.
13707 * optc-gen.awk: Add integer_range_info to cl_options struct.
13708 * opts-common.c (decode_cmdline_option): Handle
13709 CL_ERR_INT_RANGE_ARG.
13710 (cmdline_handle_error): Likewise.
13711 * opts.c (print_filtered_help): Show valid interval in
13712 when --help is provided.
13713 * opts.h (struct cl_option): Add range_min and range_max fields.
13714 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13716 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13718 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13719 (x * C EQ/NE y * C): New transformation.
13721 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13723 * genmultilib (combination_space): Accept '+' in option names.
13725 2017-06-28 Martin Liska <mliska@suse.cz>
13728 * asan.c (instrument_derefs): Bail out inner references
13729 that are hard register variables.
13731 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13734 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13735 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13737 2017-06-28 Richard Biener <rguenther@suse.de>
13739 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13740 (vect_get_slp_defs): Adjust.
13741 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13743 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13745 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13746 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13747 (vectorizable_reduction): Adjust.
13748 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13750 (vect_get_slp_defs): Likewise.
13751 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13752 (vectorizable_bswap): Adjust.
13753 (vectorizable_call): Likewise.
13754 (vectorizable_conversion): Likewise.
13755 (vectorizable_assignment): Likewise.
13756 (vectorizable_shift): Likewise.
13757 (vectorizable_operation): Likewise.
13758 (vectorizable_store): Likewise.
13759 (vectorizable_condition): Likewise.
13760 (vectorizable_comparison): Likewise.
13762 2017-06-28 Michael Collison <michael.collison@arm.com>
13765 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13767 (arm_gen_movmemqi): Removed unused variable 'i'.
13768 Convert 'for' loop into 'while' loop.
13769 (arm_expand_prologue): Remove last unnecessary set of insn.
13770 (thumb_pop): Remove unused variable 'pushed_words'.
13771 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13773 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13775 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13776 * config/s390/s390-protos.h: Add prototype of
13777 s390_rel_address_ok_p.
13778 * config/s390/s390.c (s390_got_symbol): New function.
13779 (s390_rel_address_ok_p): New function.
13780 (legitimize_pic_address): Use s390_rel_address_ok_p.
13781 (s390_load_got): Use s390_got_symbol.
13782 (s390_option_override): Issue error if
13783 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13784 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13786 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13788 2017-06-27 Andrew Pinski <apinski@cavium.com>
13790 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13791 (X * copysign (1.0, X)): New pattern.
13792 (X * copysign (1.0, -X)): New pattern.
13793 (copysign (-1.0, CST)): New pattern.
13795 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13797 * genmultilib (combination_space): Remove variable.
13798 Validate reuse rules against regular expression for any sequence
13799 of multilib options in any order.
13801 2017-06-27 Michael Collison <michael.collison@arm.com>
13803 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13804 call aarch64_split_simd_combine.
13805 * (aarch64_combine_internal<mode>): Delete pattern.
13806 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13807 Allow register and subreg operands.
13809 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13811 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13812 specific need, just fallback on defaults.
13813 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13815 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13816 Olivier Hainque <hainque@adacore.com>
13818 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13820 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13821 targets. Pick a default if no particular attempt applied.
13822 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13825 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13827 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13828 (x86_64-wrs-vxworks7): Likewise.
13830 2017-06-27 Marek Polacek <polacek@redhat.com>
13833 * ubsan.c (instrument_null): Check get_base_address's result for null.
13835 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13837 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13839 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13841 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13842 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13843 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13844 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13845 New function types.
13846 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13847 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13848 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13849 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13850 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13851 BUILT_IN_FEUPDATEENV): New builtins.
13852 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13853 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13854 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13855 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13857 (builtin_structptr_types): Adjust size.
13858 * tree.c (builtin_structptr_types): Add four entries.
13860 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13861 Olivier Hainque <hainque@adacore.com>
13863 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13864 (TLS_SYM): New local macro, forcing reference to __tls__ on
13865 link command lines for VxWorks 7 RTPs, triggering initialization
13867 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13868 OS features TLS support, true for RTPs on VxWorks 7.
13869 * config/vxworks.c (vxworks_override_options): Setup emutls
13872 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13874 * predict.c (test_prediction_value_range): Use -1U instead of -1
13875 to avoid narrowing conversion warning.
13876 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13877 to avoid narrowing conversion warning.
13878 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13880 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13882 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13884 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13885 64bit configurations.
13886 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13887 (SIZE_TYPE): Likewise.
13888 * config/vxworks.c (vxworks_emutls_var_fields): Use
13889 long_unsigned_type_node instead of unsigned_type_node as the offset
13890 field type, which is "pointer" mode in emutls.c.
13892 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13895 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13897 PR middle-end/81207
13898 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13899 gimple_vuse copying separately from gimple_vdef copying.
13901 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13903 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13906 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13907 Olivier Hainque <hainque@adacore.com>
13909 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13910 for all vxworks7 targets.
13911 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13912 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13913 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13914 variations for VX6/VX7 and 32/64bits later on in ...
13915 (VXWORKS_LIB_SPEC): Leverage new macros.
13916 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13917 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13919 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13921 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13922 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13924 2017-06-26 Carl Love <cel@us.ibm.com>
13926 * config/rs6000/rs6000-c.c: Add support for built-in functions
13927 vector bool char vec_reve (vector bool char);
13928 vector signed char vec_reve (vector signed char);
13929 vector unsigned char vec_reve (vector unsigned char);
13930 vector bool int vec_reve (vector bool int);
13931 vector signed int vec_reve (vector signed int);
13932 vector unsigned int vec_reve (vector unsigned int);
13933 vector bool long long vec_reve (vector bool long long);
13934 vector signed long long vec_reve (vector signed long long);
13935 vector unsigned long long vec_reve (vector unsigned long long);
13936 vector bool short vec_reve (vector bool short);
13937 vector signed short vec_reve (vector signed short);
13938 vector double vec_reve (vector double);
13939 vector float vec_reve (vector float);
13940 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13941 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13942 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13943 (altivec_vreve): New pattern.
13944 * config/rs6000/altivec.h (vec_reve): New define.
13945 * doc/extend.texi (vec_rev): Update the built-in documentation file
13946 for the new built-in functions.
13948 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13950 PR tree-optimization/71815
13951 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13953 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13955 (slsr_process_phi): Likewise.
13956 (replace_uncond_cands_and_profitable_phis): Don't replace a
13957 multiply candidate with a stride of 1 (copy or cast).
13958 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13960 (lowest_cost_path): Likewise.
13961 (total_savings): Likewise.
13963 2017-06-26 Richard Biener <rguenther@suse.de>
13966 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13967 Use def_builtin_pure for all gather builtins.
13969 2017-06-26 Richard Biener <rguenther@suse.de>
13971 PR tree-optimization/81203
13972 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13973 non-dominating BBs.
13975 2017-06-26 Marek Polacek <polacek@redhat.com>
13978 * doc/invoke.texi: Document -Wmultistatement-macros.
13980 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13982 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13983 arm_neon_ok_no_float_abi effective target.
13985 2017-06-26 Richard Biener <rguenther@suse.de>
13987 PR tree-optimization/80928
13988 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13989 (copy_bbs): Set BB_DUPLICATED flag early.
13990 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13992 (execute_on_shrinking_pred): Likewise.
13993 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13994 BB_DUPLICATED blocks.
13995 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13996 iterate over all PHIs considering removal of *gsi.
13998 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14000 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14003 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14005 * config/rs6000/rs6000-string.c: (expand_block_clear,
14006 do_load_for_compare, select_block_compare_mode,
14007 compute_current_alignment, expand_block_compare,
14008 expand_strncmp_align_check, expand_strn_compare,
14009 expand_block_move, rs6000_output_load_multiple)
14010 Move functions related to string/block move/compare
14011 to a separate file.
14012 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14013 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14014 for this function which is now used in two files.
14015 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14016 * config.gcc: Add rs6000-string.o to extra_objs for
14017 targets powerpc*-*-* and rs6000*-*-*.
14019 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14022 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14023 32-bit, since indexed is not valid for DImode.
14024 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14025 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14026 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14027 (define_peephole2 for Altivec d-form store): Likewise.
14030 * multiple_target.c (create_dispatcher_calls): Only create the
14031 dispatcher call if the function is the default clone of a
14032 versioned function.
14034 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14036 PR middle-end/80902
14037 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14038 a call, force the call to not be a tail call.
14040 2017-06-23 Jeff Law <law@redhat.com>
14042 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14045 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14047 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14048 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14049 handling for early expansion of vector shifts (sl,sr,sra,rl).
14050 (builtin_function_type): Add vector shift right instructions
14051 to the unsigned argument list.
14053 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14055 rtl-optimizatoin/79286
14056 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14057 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14058 trap. PIC register plus a const unspec without offset can never trap.
14060 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14062 * tree.h (builtin_structptr_type): New type.
14063 (builtin_structptr_types): Declare new array.
14064 * tree.c (builtin_structptr_types): New array.
14065 (free_lang_data, build_common_tree_nodes): Use it.
14067 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14070 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14073 2017-06-22 Matt Turner <mattst88@gmail.com>
14075 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14076 Lake models to skylake case. Assume skylake for unknown
14077 models with clflushopt.
14079 2017-06-22 Jeff Law <law@redhat.com>
14081 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14082 frame sizes that do not satisfy aarch64_uimm12_shift.
14084 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14086 * profile-count.h (apply_probability,
14087 apply_scale, probability_in): Fix checks for zero.
14089 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14091 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14092 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14094 2016-06-22 Richard Biener <rguenther@suse.de>
14096 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14097 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14098 REDUC_MAX_EXPR support.
14099 (vectorizable_reduction): Likewise.
14100 (vect_create_epilog_for_reduction): Likewise.
14102 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14104 * match.pd (A / (1 << B) -> A >> B): New.
14105 * generic-match-head.c: Include optabs-tree.h.
14106 * gimple-match-head.c: Likewise.
14107 * optabs-tree.h (target_supports_op_p): New.
14108 * optabs-tree.c (target_supports_op_p): New.
14110 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14112 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14113 $gcc_cv_ld --help output.
14114 (gcc_cv_ld_demangle): Likewise.
14115 (gcc_cv_ld_eh_frame_hdr): Likewise.
14116 (gcc_cv_ld_pie): Likewise.
14117 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14118 (gcc_cv_ld_buildid): Likewise.
14119 (gcc_cv_ld_sysroot): Likewise.
14120 (ld_bndplt_support): Likewise.
14121 (ld_pushpopstate_support): Likewise.
14122 * configure: Regenerate.
14123 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14125 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14128 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14129 operands indexes to avoid gap between operands and match_dups.
14131 2017-06-21 Andrew Pinski <apinski@cavium.com>
14133 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14134 Increment Arith_shift and Arith_shift_reg by 1.
14135 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14137 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14138 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14139 (aarch64_strip_extend): Add new argument and test for it.
14140 (aarch64_cheap_mult_shift_p): New function.
14141 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14142 add a cost if it is true.
14143 Update calls to aarch64_strip_extend.
14144 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14146 2017-06-21 Andrew Pinski <apinski@cavium.com>
14148 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14150 (thunderxt88): Likewise.
14151 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14152 (thunderx_prefetch_tune): New variable.
14153 (thunderx2t99_prefetch_tune): Update for the correct values.
14154 (thunderxt88_tunings): New variable.
14155 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14156 generic_prefetch_tune.
14157 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14159 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14161 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14162 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14163 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14164 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14165 (aarch64_atomic_cas<mode>, GPI): Likewise.
14167 2017-06-21 Martin Liska <mliska@suse.cz>
14169 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14170 statements on cold and hot labels.
14171 * predict.c (tree_estimate_probability_bb): Remove the
14172 prediction from this place.
14174 2017-06-21 Martin Liska <mliska@suse.cz>
14176 PR tree-optimization/79489
14177 * gimplify.c (maybe_add_early_return_predict_stmt): New
14179 (gimplify_return_expr): Call the function.
14180 * predict.c (tree_estimate_probability_bb): Remove handling
14182 * predict.def: Update comment about early return predictor.
14183 * gimple-predict.h (is_gimple_predict): New function.
14184 * predict.def: Change default value of early return to 66.
14185 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14187 * passes.def: Put pass_strip_predict_hints to the beginning of
14190 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14192 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14193 FUNCTION_DECL declarations.
14194 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14196 (dwaf2out_decl): Likewise.
14197 * godump.c (go_early_global_decl): Skip call to the real debug hook
14198 for FUNCTION_DECL declarations.
14199 * passes.c (rest_of_decl_compilation): Skip call to the
14200 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14201 -fdump-go-spec is passed.
14203 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14205 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14206 Reorder for compactness.
14207 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14208 (def_builtin_pure, def_builtin_pure2): New functions.
14209 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14211 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14213 * match.pd (nop_convert): New predicate.
14214 ((A +- CST1) +- CST2): Allow some NOP conversions.
14216 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14219 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14220 with ctors/dtors if GOVD_SHARED is set.
14222 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14224 * config/aarch64/aarch64.md (movti_aarch64):
14225 Emit mov rather than orr.
14226 (movtf_aarch64): Likewise.
14227 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14228 Emit mov rather than orr.
14230 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14232 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14233 Swap alternatives, make integer dup more expensive.
14235 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14237 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14238 Return true for non-tls symbols.
14240 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14242 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14243 (cortex-a75): Likewise.
14244 (cortex-a75.cortex-a55): Likewise.
14245 * config/aarch64/aarch64-tune.md: Regenerate.
14246 * doc/invoke.texi (-mtune): Document new values for -mtune.
14248 2017-06-21 Tom de Vries <tom@codesourcery.com>
14250 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14251 stack_size feature.
14252 (Effective-Target Keywords, Other attributes): Suggest using
14253 dg-add-options stack_size feature to get stack limit in stack_size
14254 effective target documentation.
14256 2017-06-21 Julian Brown <julian@codesourcery.com>
14257 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14259 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14260 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14261 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14263 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14264 attribute type list for neon_multiply.
14265 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14266 attribute type list for neon_multiply.
14267 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14268 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14269 attribute type list for neon_multiply.
14270 * config/arm/types.md (crypto_pmull): Add.
14271 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14272 attribute type list.
14274 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14276 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14279 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14281 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14282 to make sure not to dereference a NULL cost_classes_ptr pointer.
14284 2017-06-20 Carl Love <cel@us.ibm.com>
14286 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14287 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14288 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14289 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14290 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14291 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14292 VMULOSW): New enum "unspec" values.
14293 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14294 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14295 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14296 altivec_vmulosw): New patterns.
14297 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14298 VMULOSW): Add definitions.
14300 2017-06-20 Julia Koval <julia.koval@intel.com>
14302 * config/i386/i386.c: Fix rounding expand for new pattern.
14303 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14305 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14307 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14308 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14310 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14312 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14315 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14317 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14318 architecture, then by implementer ID.
14319 * config/aarch64/aarch64-tune.md: Regenerate.
14321 2017-06-20 Richard Biener <rguenther@suse.de>
14323 PR middle-end/81097
14324 * fold-const.c (split_tree): Fold to type before negating.
14326 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14328 * diagnostic-show-locus.c
14329 (selftest::test_fixit_deletion_affecting_newline): New function.
14330 (selftest::diagnostic_show_locus_c_tests): Call it.
14332 2017-06-20 Andreas Schwab <schwab@suse.de>
14335 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14338 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14340 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14341 __ARM_FEATURE_COPROC according to support.
14343 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14345 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14346 Rewritten to avoid overflow for > 32-bit pointers.
14349 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14350 by removing enum keyword.
14351 (ubsan_type_descriptor): Likewise. Formatting fix.
14354 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14355 splitter): Require TARGET_SSE2 in the condition.
14357 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14360 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14361 for doing vector set of SFmode on ISA 3.0.
14362 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14363 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14365 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14366 SFmode value into a V4SF variable that was extracted from another
14367 V4SF variable without converting the element to double precision
14368 and back to single precision vector format.
14369 (vsx_insert_extract_v4sf_p9_2): Likewise.
14371 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14373 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14374 in UWHI to avoid undefined overflow.
14377 * ubsan.h (enum ubsan_encode_value_phase): New.
14378 (ubsan_encode_value): Change second argument to
14379 enum ubsan_encode_value_phase with default value of
14380 UBSAN_ENCODE_VALUE_GENERIC.
14381 * ubsan.c (ubsan_encode_value): Change second argument to
14382 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14383 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14384 create_tmp_var_raw instead of create_tmp_var and use a
14386 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14387 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14388 ubsan_encode_value callers.
14391 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14392 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14393 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14395 2017-06-19 Richard Biener <rguenther@suse.de>
14397 PR middle-end/81118
14398 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14399 estimates if we changed anything.
14401 2017-06-19 Richard Biener <rguenther@suse.de>
14403 PR tree-optimization/80887
14404 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14405 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14406 simplified lookups, then reset mprts_hook.
14407 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14409 (try_to_simplify): Likewise.
14411 2017-06-19 Martin Liska <mliska@suse.cz>
14414 * gimplify.c (gimplify_switch_expr):
14415 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14417 2017-06-19 Martin Liska <mliska@suse.cz>
14419 * doc/install.texi: Document that PGO runs in 4 stages.
14421 2017-06-19 Martin Liska <mliska@suse.cz>
14424 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14425 to dispatcher function name.
14426 * multiple_target.c (replace_function_decl): New function.
14427 (create_dispatcher_calls): Redirect both edges and references.
14429 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14431 * profile-count.c (profile_count::dump): Dump quality.
14432 (profile_count::differs_from_p): Update for unsigned val.
14433 * profile-count.h (profile_count_quality): New enum.
14434 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14436 2017-06-19 Richard Biener <rguenther@suse.de>
14438 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14439 struct function as arg.
14440 (estimate_numbers_of_iterations): Export overload with loop arg.
14441 (free_numbers_of_iterations_estimates_loop): Use an overload of
14442 free_numbers_of_iterations_estimates instead.
14443 * tree-cfg.c (remove_bb): Adjust.
14444 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14445 * tree-parloops.c (gen_parallel_loop): Likewise.
14446 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14448 (tree_unroll_loops_completely): Likewise.
14449 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14450 Use an overload instead and export.
14451 (estimated_loop_iterations): Adjust.
14452 (max_loop_iterations): Likewise.
14453 (likely_max_loop_iterations): Likewise.
14454 (estimate_numbers_of_iterations): Take struct function as arg
14456 (loop_exits_before_overflow): Adjust.
14457 (free_numbers_of_iterations_estimates_loop): Use an overload.
14458 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14459 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14461 2017-06-19 Richard Biener <rguenther@suse.de>
14464 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14465 RANGE_EXPR conservatively.
14467 2017-06-16 Carl Love <cel@us.ibm.com>
14469 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14470 definitions for vec_float, vec_float2, vec_floato,
14471 vec_floate built-ins.
14472 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14473 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14475 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14476 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14477 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14478 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14479 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14480 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14481 vec_floato): Add builtin defines.
14482 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14483 Update the built-in documentation file for the new built-in
14486 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14488 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14489 (mthumb): Mark as the negative of -marm.
14491 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14493 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14495 (ARM Options, -mtune): Document that this accepts the same
14496 extension options as -mcpu.
14497 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14499 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14501 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14502 permitted extensions.
14504 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14506 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14507 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14508 (armv8-m.main): Add option +nodsp.
14509 * config/arm/arm-cpu-cdata.h: Regenerated.
14511 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14513 * config/arm/t-fuchsia: New file.
14514 * config.gcc (arm*-*-fuchsia*): Use it.
14516 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14518 * config/arm/t-symbian: Rewrite for new option infrastructure.
14520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14522 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14523 (MULTILIB_REQUIRED): Likewise.
14525 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14527 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14528 (MULTILIB_RESUE): Likewise.
14529 (MULTILIB_MATCHES): Likewise.
14530 (MULTLIB_REQUIRED): Likewise.
14532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14534 * config/arm/t-rtems: Rewrite for new option framework.
14536 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14538 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14539 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14540 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14541 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14542 * config/arm/t-multilib: ... here.
14543 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14544 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14545 armv7-a and armv8*-a when A-profile libraries have not been built.
14546 * config/arm/t-rmprofile: Rewrite.
14548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14550 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14551 with a backslash. Remove the backslash after substituting unescaped
14553 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14557 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14558 the driver to pass through the default setting of -mfloat-abi.
14559 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14561 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14562 (all_feat_combs): New rule.
14563 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14565 * config/arm/t-aprofile: Rewrite.
14567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14569 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14570 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14571 fpu is not specified by the user/command-line.
14572 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14573 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14574 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14575 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14576 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14577 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14581 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14582 * config/arm/t-arm-elf: Rewritten.
14584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14586 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14587 have some floating-point instructions.
14588 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14589 (TARGET_MAYBE_HARD_FLOAT): New macro.
14590 * config/arm/arm-builtins.c (arm_init_builtins): Use
14591 TARGET_MAYBE_HARD_FLOAT.
14592 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14596 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14597 (configargs.h): Include it.
14598 (arm_print_hint_for_fpu_option): New function.
14599 (arm_parse_fpu_option): New function.
14600 (candidate_extension): New class.
14601 (arm_canon_for_multilib): New function.
14602 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14603 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14604 (ARCH_CANONICAL_SPECS): New macro.
14605 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14609 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14610 are set after handling multilib fragments. Set target_cpu_default2
14613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14615 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14617 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14618 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14620 (check_arch): Likewise.
14621 * config/arm/arm.c (arm_configure_build_target): Handle
14622 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14623 options in the default.
14625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14627 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14628 when an option is an alias of another.
14629 * config/arm/parsecpu.awk (optalias): New parser token.
14630 (gen_comm_data): Mark non-alias options as such. Emit entries
14631 for extension aliases.
14632 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14633 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14634 (armv6kz, armv6zk, armv6t2): Likewise.
14635 (armv7): Make vfpv3-d16 an alias.
14636 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14638 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14639 Sort in canonical order.
14640 (armv8-a): Sort in canonical order.
14641 (armv8.1-a, armv8.2-a): Likewise.
14642 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14644 (cortex-a9): Sort in canonical order.
14645 * config/arm/arm.c (selftests.h): Include it.
14646 (arm_test_cpu_arch_data): New function.
14647 (arm_run_self_tests): New function.
14648 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14649 (targetm): Move declaration to the end of the file.
14650 * arm-cpu-cdata.h: Regenerated.
14652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14654 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14655 call to target_mode_check describing the type of option passed.
14656 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14657 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14658 arm_parse_cpu_option_name to match parameters against list of
14660 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14661 arm_arch_core_flags data structure.
14662 * config/arm/arm-cpu_cdata.h: Regenerated.
14664 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14666 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14668 (arm_print_hint_for_cpu_option): Likewise.
14669 (arm_print_hint_for_arch_option): Likewise.
14670 (arm_parse_cpu_option_name): Likewise.
14671 (arm_parse_arch_option_name): Likewise.
14672 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14673 of entries in the all_fpus list.
14674 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14675 (arm_parse_cpu_option_name): Declare.
14676 (arm_parse_arch_option_name): Declare.
14677 (arm_parse_option_features): Declare.
14678 (arm_intialize_isa): Declare.
14679 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14681 (gen_comm_data): ... here. Make definitions non-static.
14682 * config/arm/arm-cpu-data.h: Regenerated.
14683 * config/arm/arm-cpu-cdata.h: Regenerated.
14685 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14687 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14688 (cpu_arch_extension): New structure.
14689 (cpu_arch_option, arch_option, cpu_option): New structures.
14690 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14691 architecture types.
14692 (gen_data): Generate new format data tables.
14693 * config/arm/arm.c (cpu_tune): New structure.
14694 (cpu_option, processors): Delete.
14695 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14696 (arm_print_hint_for_cpu_option): ... this and ...
14697 (arm_print_hint_for_arch_option): ... this.
14698 (arm_parse_arch_cpu_name): Delete. Replace with ...
14699 (arm_parse_cpu_option_name): ... this and ...
14700 (arm_parse_arch_option_name): ... this.
14701 (arm_unrecognized_feature): Change type of target parameter to
14703 (arm_parse_arch_cpu_features): Delete. Replace with ...
14704 (arm_parse_option_features): ... this.
14705 (arm_configure_build_target): Rework to use new configuration data
14707 (arm_print_tune_info): Rework for new configuration data tables.
14708 * config/arm/arm-cpu-data.h: Regenerated.
14709 * config/arm/arm-cpu.h: Regenerated.
14711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14713 * Makefile.in (OBJS): Move sbitmap.o from here ...
14714 (OBJS-libcommon): ... to here.
14716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14718 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14719 (ISA_ALL_CRYPTO): New macro.
14720 (ISA_ALL_SIMD): New macro
14721 (ISA_ALL_FP): New macro.
14722 * config/arm/arm.c (fpu_bitlist): Update initializer.
14723 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14725 (arm9e): Add fpu. Add option for nofp
14726 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14727 (arm926ej-s, arm1026ej-s): Likewise.
14728 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14729 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14730 neon-fp16, neon-vfpv4, nofp and nosimd.
14731 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14732 (cortex-a8): Add fpu. Add option for nofp.
14733 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14734 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14735 (cortex-r4f): Add fpu.
14736 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14737 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14739 (cortex-r8): Likewise.
14740 (cortex-m4): Add fpu. Add option for nofp.
14741 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14742 (cortex-a17.cortex-a7): Likewise.
14743 (cortex-a32): Add fpu. Add options for crypto and nofp.
14744 (cortex-a35, cortex-a53): Likewise.
14745 (cortex-a57): Add fpu. Add option for crypto.
14746 (cortex-a72, cortex-a73): Likewise.
14747 (exynos-m1): Likewise.
14748 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14749 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14750 (cortex-m33): Add fpu. Add option for nofp.
14751 * config/arm/arm-cpu-cdata.h: Regenerated
14752 * config/arm/arm-cpu-data.h: Regenerated.
14754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14756 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14757 (armv5te, armv5tej): Likewise.
14758 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14759 (armv7): Add options fp and vfpv3-d16.
14760 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14761 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14763 (armv7ve): Likewise.
14764 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14765 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14766 (armv8-a): Add nocrypto option.
14767 (armv8.1-a, armv8.2-a): Likewise.
14768 (armv8-m.main): add options fp, fp.dp and nofp.
14770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14772 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14774 (armv8-a+crc): Delete.
14775 (armv8.1-a): Add options simd, crypto and nofp.
14776 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14777 (armv8.2-a+fp16): Delete.
14778 (armv8-m.main): Add option dsp.
14779 (armv8-m.main+dsp): Delete.
14780 (cortex-a8): Add fpu. Add nofp option.
14781 (cortex-a9): Add fpu. Add nofp and nosimd options.
14782 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14783 link to main cpu and architecture data structures.
14784 (gen_comm_data): Only put isa attributes from the main architecture
14786 (option): New statement for architecture and CPU entries.
14787 * arm.c (struct cpu_option): New structure.
14788 (struct processors): Add entry for options.
14789 (arm_unrecognized_feature): New function.
14790 (arm_parse_arch_cpu_name): Ignore any characters after the first
14792 (arm_parse_arch_cpu_feature): New function.
14793 (arm_configure_build_target): Separate out any CPU and architecture
14794 features and parse separately. Don't error out if -mfpu=auto is
14795 used with only an architecture string.
14796 (arm_print_asm_arch_directives): New function.
14797 (arm_file_start): Call it.
14798 * config/arm/arm-cpu-cdata.h: Regenerated.
14799 * config/arm/arm-cpu-data.h: Likewise.
14800 * config/arm/arm-tables.opt: Likewise.
14802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14804 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14805 assembler when it is not -mfpu=auto.
14807 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14809 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14810 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14811 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14812 (ASM_CPU_SPEC): Rewrite.
14813 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14814 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14815 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14816 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14817 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14818 copied string is NUL-terminated. Also strip any characters prefixed
14820 (arm_rewrite_selected_arch): New function.
14821 (arm_rewrite_march): New function.
14823 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14825 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14826 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14827 (march, mcpu, mtune): Convert to string-based options.
14828 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14829 (arm_parse_arch_cpu_name): New function.
14830 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14831 identify selected architecture or CPU.
14832 (arm_option_save): New function.
14833 (TARGET_OPTION_SAVE): Redefine.
14834 (arm_option_restore): Restore string options.
14835 (arm_option_print): Print string options.
14837 2017-06-16 Martin Sebor <msebor@redhat.com>
14839 PR tree-optimization/80933
14840 PR tree-optimization/80934
14841 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14842 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14843 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14844 (gimple_fold_builtin): Call them.
14846 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14848 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14849 as unlikely; update profile.
14851 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14853 * predict.c (force_edge_cold): Handle declaring edges impossible
14856 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14858 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14860 (try_unroll_loop_completely): Fix reporting.
14862 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14864 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14866 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14869 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14870 if given a non-constant argument for an intrinsic which requires a
14873 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14875 * profile.c (compare_freqs): New function.
14876 (branch_prob): Sort edge list.
14877 (find_spanning_tree): Assume that the list is priority sorted.
14879 2017-06-16 Richard Biener <rguenther@suse.de>
14881 PR tree-optimization/81090
14882 * passes.def (pass_record_bounds): Remove.
14883 * tree-pass.h (make_pass_record_bounds): Likewise.
14884 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14885 make_pass_record_bounds): Likewise.
14886 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14887 not free niter estimates at the beginning but at the end.
14888 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14890 2017-06-16 Richard Biener <rguenther@suse.de>
14892 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14893 initializer to workaround ICE in host GCC 4.8.
14895 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14897 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14899 (clone_inlined_nodes): Update.
14901 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14903 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14904 prefetch settings, and enable prefetching by default at -O3.
14906 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14908 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14909 Set flag_prefetch_loop_arrays according to tuning data.
14911 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14913 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14914 New tune structure.
14915 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14916 [Unrelated to main purpose of the patch] Place the pointer field last
14917 to enable type checking errors when tune structure are wrongly merged.
14918 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14919 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14920 (thunderx2t99_prefetch_tune): New tune constants.
14921 (tune_params *_tunings): Update all tunings (no functional change).
14922 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14923 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14924 from tunings structures.
14926 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14929 * ubsan.c (instrument_null): Add T argument, use it instead
14930 of computing it based on IS_LHS.
14931 (instrument_object_size): Likewise.
14932 (pass_ubsan::execute): Adjust instrument_null and
14933 instrument_object_size callers to pass gimple_get_lhs or
14934 gimple_assign_rhs1 result to it. Use instrument_null instead of
14935 calling get_base_address and instrument_mem_ref. Handle
14936 aggregate call arguments for object-size sanitization.
14938 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14940 PR tree-optimization/81089
14941 * tree-vrp.c (is_masked_range_test): Validate operands of
14944 2017-06-15 Martin Sebor <msebor@redhat.com>
14947 * dumpfile.c (dump_register): Avoid calling memset to initialize
14948 a class with a default ctor.
14949 * gcc.c (struct compiler): Remove const qualification.
14950 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14951 * hash-table.h: Ditto.
14952 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14954 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14955 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14957 * params.h (struct param_info): Make struct members non-const.
14958 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14959 with default initialization.
14960 * vec.h (vec_copy_construct, vec_default_construct): New helper
14962 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14963 with vec_copy_construct.
14964 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14965 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14966 * doc/invoke.texi (-Wclass-memaccess): Document.
14968 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14970 * emit-rtl.h (is_leaf): Update comment about local
14971 register allocator.
14973 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14976 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14977 for a variable to have a section before checking if the section has a
14979 Set section to.persistent if persistent attribute is set.
14980 Warn if .persistent attribute is used on an automatic variable.
14982 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14984 PR rtl-optimization/80474
14985 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14987 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14989 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14992 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14994 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14995 (casesi): Emit bounds checking as RTL.
14996 (casesi_internal_mips16_<mode>): Remove bounds checking.
14998 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15000 * config/xtensa/xtensa.c (xtensa_option_override): Append
15001 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15002 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15003 xtensa_doloop_hooks): Define unconditionally.
15004 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15006 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15007 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15008 for it in xtensa_option_override.
15009 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15010 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15012 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15014 * doc/cppopts.texi: Document '-' special value to -MF.
15016 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15018 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15019 (cortex_a53_fconst): Likewise.
15020 (cortex_a53_fpmul): Likewise.
15021 (cortex_a53_f_load_64): Likewise.
15022 (cortex_a53_f_load_many): Likewise.
15023 (cortex_a53_advsimd_alu): Likewise.
15024 (cortex_a53_advsimd_alu_q): Likewise.
15025 (cortex_a53_advsimd_mul): Likewise.
15026 (cortex_a53_advsimd_mul_q): Likewise.
15027 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15028 Add missing fmul, r2f_cvt and fconst cases.
15030 2017-06-14 Richard Biener <rguenther@suse.de>
15032 PR middle-end/81088
15033 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15035 (fold_binary_loc): When associating do not treat pre-existing
15036 TREE_OVERFLOW on literal constants as a reason to allow
15037 TREE_OVERFLOW on associated literal constants.
15039 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15041 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15042 (MASK_FEATURES): New macro.
15043 * config/sparc/sparc.c (sparc_option_override): Remove the special
15044 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15046 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15048 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15049 a division of 0 if non-call exceptions are enabled.
15051 2017-06-14 Andrew Pinski <apinski@cavium.com>
15052 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15055 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15056 Improve vector initialization code gen for only variable case.
15058 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15060 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15062 2017-06-14 Richard Biener <rguenther@suse.de>
15064 PR tree-optimization/81083
15065 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15068 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15070 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15071 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15072 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15073 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15074 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15075 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15077 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15079 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15080 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15082 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15084 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15086 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15088 * config/rs6000/t-rtems: Don't handle SPE.
15090 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15092 * config/rs6000/t-linux: Don't handle SPE.
15094 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15096 * config/rs6000/eabispe.h: Delete file.
15098 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15100 * config/rs6000/t-spe: Delete file.
15102 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15104 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15105 (rs6000_legitimate_offset_address_p): Return false for anything in
15106 V2SImode or V2SFmode.
15108 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15110 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15111 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15112 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15114 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15115 (rs6000_legitimate_offset_address_p): Ditto.
15116 (rs6000_emit_move): Ditto.
15117 (rs6000_init_builtins): Remove V4HI_type_node.
15119 2017-06-13 Martin Liska <mliska@suse.cz>
15122 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15123 (gate_asan): Likewise.
15124 * asan.h (asan_no_sanitize_address_p): Remove the function.
15125 (sanitize_flags_p): New function.
15126 * builtins.def: Fix coding style.
15127 * common.opt: Use renamed enum value.
15128 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15129 * doc/extend.texi: Document no_sanitize attribute.
15130 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15131 to SANITIZE_UNDEFINED_NONDEFAULT.
15132 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15133 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15134 Use sanitize_flags_p.
15135 * gimplify.c (gimplify_function_tree): Likewise.
15136 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15137 * opts.c (parse_no_sanitize_attribute): New function.
15138 (common_handle_option): Use renamed enum value.
15139 * opts.h (parse_no_sanitize_attribute): Declare.
15140 * tree.c (sanitize_flags_p): New function.
15141 * tree.h: Declared here.
15142 * tsan.c: Use sanitize_flags_p.
15143 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15144 (instrument_mem_ref): Likewise.
15145 (instrument_bool_enum_load): Likewise.
15146 (do_ubsan_in_current_function): Remove the function.
15147 (pass_ubsan::execute): Use sanitize_flags_p.
15148 * ubsan.h: Remove do_ubsan_in_current_function
15149 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15150 (dump_function_to_file): Use it here.
15152 2017-06-13 Martin Jambor <mjambor@suse.cz>
15154 PR tree-optimization/80803
15155 PR tree-optimization/81063
15156 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15157 (propagate_subaccesses_across_link): Enqueue subtree whenever
15158 necessary instead of relying on the caller.
15160 2017-06-13 Martin Jambor <mjambor@suse.cz>
15162 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15163 that have a first_link.
15164 (sort_and_splice_var_accesses): Do not check first_link before
15166 (subtree_mark_written_and_enqueue): Likewise.
15167 (propagate_all_subaccesses): Likewise and do not stop at first
15168 parent with a first_link.
15170 2017-06-13 Martin Jambor <mjambor@suse.cz>
15172 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15175 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15177 * match.pd: New pattern.
15179 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15181 * tree-vrp.c (is_masked_range_test): New function.
15182 (register_edge_assert_for): Determine ranges for
15185 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15187 PR tree-optimization/67328
15188 * fold-const.c (maskable_range_p): New function.
15189 (build_range_check): Generate bittests if possible.
15191 2017-06-13 Martin Liska <mliska@suse.cz>
15193 * gimple-pretty-print.c (dump_probability): Add new argument.
15194 (dump_edge_probability): Dump both probability and count.
15195 (dump_gimple_label): Likewise.
15196 (dump_gimple_bb_header): Likewise.
15198 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15201 * config/avr/avr-devices.c: Fix indentation.
15202 * config/avr/gen-avr-mmcu-specs.c: Dito.
15204 2017-06-13 Richard Biener <rguenther@suse.de>
15206 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15207 instead get vector type from stmt_info.
15208 (vectorizable_reduction): Adjust. Remove dead code.
15210 2017-06-13 Richard Biener <rguenther@suse.de>
15212 PR middle-end/81065
15213 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15214 case of C * (x * C2 + C3).
15215 (fold_addr_of_array_ref_difference): Properly fold index difference.
15217 2017-06-12 David S. Miller <davem@davemloft.net>
15220 * config/sparc/sparc.md (return expander): Emit frame blockage if
15221 function uses alloca.
15223 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15225 * combine.c (make_field_assignment): Check len rather than the mode
15226 precision when calling force_to_mode.
15228 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15230 Support multilibs and devices that see flash in RAM address range.
15233 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15234 (avr_mcu_t) <flash_pm_offset>: New field.
15235 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15236 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15237 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15238 (AVR_TINY_PM_OFFSET): Remove macro.
15239 * config/avr/avr.opt (-mshort-calls): New option.
15240 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15241 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15242 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15243 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15244 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15245 instead of avr_arch->have_jmp_call.
15246 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15247 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15248 avr_arch->flash_pm_offset to define.
15249 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15250 new field flash_pm_offset. Add entry for avrxmega3.
15251 (avr_texinfo): Add entry for avrxmega3.
15252 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15253 attiny212, attiny214,
15254 attiny412, attiny414, attiny416, attiny417,
15255 attiny814, attiny816, attiny817,
15256 attiny1614, attiny1616, attiny1617,
15257 attiny3214, attiny3216, attiny3217.
15258 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15259 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15260 (avr_print_operand_address) [AVR_TINY]: Same.
15261 (avr_asm_init_sections) <readonly_data_section>: Only patch
15262 callback if avr_arch->flash_pm_offset = 0.
15263 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15264 for rodata if avr_arch->flash_pm_offset != 0.
15265 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15266 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15267 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15268 (m_options): Append opt_rcall.
15269 (m_dirnames): Append dir_rcall.
15270 * config/avr/t-multilib: Regenerate.
15272 * configure.ac [target=avr]: Check whether avrxmega3 default
15273 linker description file works as needed.
15274 * configure: Regenerate.
15275 * doc/avr-mmcu.texi: Regenerate.
15276 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15277 <__AVR_ARCH__>: Document avrxmega3 and 103.
15278 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15279 <__AVR_SHORT_CALLS__>: Document it.
15280 <__AVR_PM_BASE_ADDRESS__>: Document it.
15281 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15282 (AVR Variable Attributes) <progmem>: Document this is
15283 not needed for avrxmega3.
15284 (AVR Named Address Spaces) <__flash>: Dito.
15286 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15288 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15290 2017-06-12 Doug Rupp <rupp@adacore.com>
15292 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15293 Append vxworks-stdint.h to the tm_file list.
15294 * config/vxworks-stdint.h: New file.
15296 2017-06-12 Martin Liska <mliska@suse.cz>
15298 PR tree-optimization/81041
15299 * tree-profile.c (gimple_gen_ic_func_profiler):
15300 Create an extra BB in profile-generate
15301 (gimple_gen_time_profiler): Likewise.
15303 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15305 PR tree-optimization/81003
15306 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15307 (update_range_test): Use it instead of force_gimple_operand_gsi.
15309 2017-06-12 Richard Biener <rguenther@suse.de>
15311 PR tree-optimization/81053
15312 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15313 with backedge value not defined in loop. Simplify def stmt
15316 2017-06-11 Tom de Vries <tom@codesourcery.com>
15319 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15321 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15322 nvptx_cannot_force_const_mem.
15324 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15326 * opts.c (finish_options): Move test for flag_split_stack after
15327 it has been initialized.
15329 2017-06-11 Jason Merrill <jason@redhat.com>
15331 * tree.h (id_equal): New.
15332 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15333 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15334 instead of strcmp of IDENTIFIER_POINTER.
15336 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15338 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15339 (mark_all_inlined_calls_cdtor): Fix formating.
15340 (inline_transform): Rescale profile before inlining.
15342 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15344 * cgraph.h (cgraph_edge::clone): Update prototype.
15345 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15346 (cgraph_node::create_clone): Update.
15347 (cgraph_node::create_version_clone): Update.
15348 * tree-inline.c (copy_bb): Update.
15349 (expand_call_inline): Update.
15351 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15353 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15354 factored out from ...
15355 (rs6000_emit_prologue): ... here.
15357 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15359 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15360 factored out from ...
15361 (rs6000_emit_prologue): ... here.
15363 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15365 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15367 (handle_missing_profiles): Fix computation of tp_first_run.
15368 (counts_to_freqs): Do not touch freqs when count is 0.
15370 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15372 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15375 2017-06-10 Tom de Vries <tom@codesourcery.com>
15377 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15378 attributes): Document signal effective target.
15380 2017-06-10 Tom de Vries <tom@codesourcery.com>
15382 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15383 Document effective target stack_size.
15385 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15387 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15388 to the edit_context if they can be auto-applied.
15390 2017-06-9 Ian Lance Taylor <iant@golang.org>
15392 * opts.c (finish_options): If -fsplit-stack, disable implicit
15393 -forder-blocks-and-partition.
15394 * doc/invoke.texi (Optimize Options): Document that when using
15395 -fsplit-stack -forder-blocks-and-partition is not implicitly
15398 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15400 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15401 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15402 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15403 * builtins.def (abort, trap, unreachable): Declare cold.
15404 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15405 * tree-core.h (ECF_COLD): New.
15406 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15407 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15409 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15411 * predict.c (unlikely_executed_stmt_p): Cleanup.
15413 2017-06-09 Richard Biener <rguenther@suse.de>
15415 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15416 model if the ref is always written to.
15418 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15420 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15422 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15424 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15427 2017-06-09 Tom de Vries <tom@codesourcery.com>
15430 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15431 "target cannot support label values" when encountering LABEL_REF.
15433 2017-06-09 Martin Liska <mliska@suse.cz>
15435 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15436 (gimple_gen_ic_func_profiler): Emit direct comparison
15437 of __gcov_indirect_call_callee with NULL.
15438 (gimple_gen_time_profiler): Change probability from
15439 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15441 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15443 * profile.c (edge_gcov_counts): Turn to pointer.
15444 (compute_branch_probabilities, compute_branch_probabilities): Update.
15445 (branch_prob): Do not clear edge_gcov_count.
15446 * profile.h (edge_gcov_counts): Turn to pointer.
15447 (edge_gcov_count): Update.
15449 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15451 * gimple.h (gimple_check_failed): Mark cold.
15453 2017-06-09 Richard Biener <rguenther@suse.de>
15455 PR tree-optimization/66623
15456 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15457 refactor check_reduction into two parts, properly computing
15458 whether we have to check reduction validity for outer loop
15461 2017-06-09 Richard Biener <rguenther@suse.de>
15463 PR tree-optimization/79483
15464 * graphite-scop-detection.c (order): New global.
15465 (get_order): Compute bb to order mapping that satisfies code
15466 generation constraints.
15467 (cmp_pbbs): New helper.
15468 (build_scops): Start domwalk at entry block, sort generated
15471 2017-06-09 Richard Biener <rguenther@suse.de>
15473 PR middle-end/81007
15474 * ipa-polymorphic-call.c
15475 (ipa_polymorphic_call_context::restrict_to_inner_class):
15476 Skip FIELD_DECLs with error_mark_node type.
15477 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15480 2017-06-09 Martin Liska <mliska@suse.cz>
15482 * predict.c (struct branch_predictor): New struct.
15483 (test_prediction_value_range): New test.
15484 (predict_c_tests): New function.
15485 * selftest-run-tests.c (selftest::run_tests): Run the function.
15486 * selftest.h: Declare new tests.
15488 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15491 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15492 gen_add3_insn did not fail.
15493 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15494 r0, construct that number in a temporary reg and add that reg to r0.
15495 If asked to put the result in r0 as well, fail.
15497 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15499 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15500 for early expansion of vec_eqv.
15502 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15504 PR middle-end/81005
15505 * ubsan.c (instrument_null): Avoid pointless code temporary.
15506 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15508 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15512 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15515 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15518 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15520 * predict.c (unlikely_executed_edge_p): Move ahead.
15521 (probably_never_executed_edge_p): Use it.
15523 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15525 PR middle-end/79988
15526 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15527 gimple_call_builtin_p call.
15529 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15531 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15532 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15533 rtl_check_failed_type2, rtl_check_failed_code1,
15534 rtl_check_failed_code2, rtl_check_failed_code_mode,
15535 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15536 rtvec_check_failed_bounds, rtl_check_failed_flag,
15537 _fatal_insn_not_found, _fatal_insn): Likewise.
15538 * tree.h (tree_contains_struct_check_failed,
15539 tree_check_failed, tree_not_check_failed,
15540 tree_class_check_failed, tree_range_check_failed,
15541 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15542 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15543 tree_operand_check_failed, omp_clause_check_failed,
15544 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15547 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15549 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15550 flag_branch_probabilities.
15551 * ipa-inline.c (edge_badness): Likewise.
15552 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15553 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15554 * predict.c (maybe_hot_frequency_p): Likewise.
15555 (probably_never_executed): Likewise.
15556 * sched-ebb.c (schedule_ebbs): Likewise.
15557 * sched-rgn.c (find_single_block_region): Likewise.
15558 * tracer.c (tail_duplicate): Likewise.
15560 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15562 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15563 longer requires x_flag_profile_use.
15565 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15567 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15568 instead of flag_reorder_blocks_and_partition.
15569 * dbxout.c (dbxout_function_end): Likewise.
15570 * dwarf2out.c (gen_subprogram_die): Likewise.
15571 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15572 * hw-doloop.c (reorg_loops): Likewise.
15573 * varasm.c (assemble_start_function,
15574 assemble_end_function): Likewise.
15575 (decide_function_section): Do not check for
15576 flag_reorder_blocks_and_partition.
15578 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15580 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15582 (chkp_get_hard_register_fake_addr_expr): Ditto.
15583 (chkp_build_addr_expr): Add check for hard reg case.
15584 (chkp_parse_array_and_component_ref): Ditto.
15585 (chkp_find_bounds_1): Ditto.
15586 (chkp_process_stmt): Don't generate bounds store for
15589 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15591 * predict.c (maybe_hot_bb_p): Do not check profile status.
15592 (maybe_hot_edge_p): Likewise.
15593 (probably_never_executed): Check for zero counts even if profile
15595 (unlikely_executed_edge_p): New function.
15596 (unlikely_executed_stmt_p): New function.
15597 (unlikely_executed_bb_p): New function.
15598 (set_even_probabilities): Use unlikely predicates.
15599 (combine_predictions_for_bb): Likewise.
15600 (predict_paths_for_bb): Likewise.
15601 (predict_paths_leading_to_edge): Likewise.
15602 (determine_unlikely_bbs): New function.
15603 (estimate_bb_frequencies): Use it.
15604 (compute_function_frequency): Use zero counts even if profile is
15606 * profile-count.h: Fix typo.
15608 2017-08-08 Julia Koval <julia.koval@intel.com>
15610 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15611 _mm512_mask_cvtsepi16_storeu_epi8,
15612 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15613 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15614 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15615 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15616 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15617 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15618 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15619 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15620 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15621 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15622 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15623 __builtin_ia32_pmovuswb256mem_mask,
15624 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15625 __builtin_ia32_pmovwb512mem_mask): New builtins.
15627 2017-08-08 Julia Koval <julia.koval@intel.com>
15629 PR target/73350,80862
15630 * config/i386/subst.md (round): Fix round pattern.
15631 * config/i386/i386.c (ix86_erase_embedded_rounding):
15632 Fix erasing rounding for the fixed pattern.
15634 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15636 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15638 2017-06-08 Martin Liska <mliska@suse.cz>
15640 PR gcov-profile/80911
15641 * gcov.c (block_info::block_info): New constructor.
15643 2017-06-07 Carl Love <cel@us.ibm.com>
15645 * config/rs6000/rs6000-c: The return type of the following
15646 built-in functions was implemented as int not long long. Fix sign
15647 of return value for the unsigned version of vec_mulo and vec_mule.
15648 vector unsigned long long vec_bperm (vector unsigned long long,
15649 vector unsigned char)
15650 vector signed long long vec_mule (vector signed int,
15652 vector unsigned long long vec_mule (vector unsigned int,
15653 vector unsigned int)
15654 vector signed long long vec_mulo (vector signed int,
15656 vector unsigned long long vec_mulo (vector unsigned int,
15657 vector unsigned int)
15658 * doc/extend.texi: Fix the documentation for the built-in
15661 2017-06-07 Carl Love <cel@us.ibm.com>
15664 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15667 2017-06-07 Carl Love <cel@us.ibm.com>
15669 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15670 support, Generate doublehv for signed int/float for BE case only.
15672 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15674 * doc/invoke.texi (mcx16): Rewrite.
15676 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15678 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15679 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15680 *mov<mode>_softfloat, and an anonymous splitter): Use
15681 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15683 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15685 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15687 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15688 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15689 (rs6000_debug_reg_global): Adjust.
15690 (rs6000_init_hard_regno_mode_ok): Adjust.
15691 (rs6000_dbx_register_number): Adjust.
15692 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15693 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15694 Remove SPE_ACC and SPEFSCR.
15695 (REG_ALLOC_ORDER): Ditto.
15696 (FRAME_POINTER_REGNUM): Change to 111.
15697 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15698 (REG_CLASS_NAMES): Ditto.
15699 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15700 (REGISTER_NAMES): Ditto.
15701 (ADDITIONAL_REG_NAMES): Ditto.
15702 (rs6000_reg_names): Ditto.
15703 * config/rs6000/rs6000.md: Renumber some register number
15706 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15708 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15710 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15711 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15713 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15714 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15715 Delete the SPE high registers.
15716 (REG_ALLOC_ORDER): Ditto.
15717 (enum reg_class): Remove SPE_HIGH_REGS.
15718 (REG_CLASS_NAMES): Ditto.
15719 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15720 (REGISTER_NAMES): Ditto.
15721 (rs6000_reg_names): Ditto.
15722 * doc/tm.texi.in: Remove SPE as example.
15723 * doc/tm.texi: Regenerate.
15725 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15727 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15728 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15729 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15730 * config/rs6000/rs6000.md (type): Remove "brinc".
15732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15734 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15735 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15736 * config/rs6000/linuxspe.h: Delete file.
15737 * config/rs6000/rs6000.md: Don't include spe.md.
15738 * config/rs6000/spe.h: Delete file.
15739 * config/rs6000/spe.md: Delete file.
15740 * config/rs6000/t-rs6000: Remove spe.md.
15742 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15744 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15745 (reg_or_none500mem_operand): Delete.
15746 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15747 instead of reg_or_none500mem_operand.
15749 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15751 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15752 handling of SPE flags.
15753 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15755 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15757 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15759 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15760 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15761 paired_divv2sf3): Similar.
15762 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15763 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15764 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15766 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15767 Rename the paired_* instruction patterns.
15768 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15770 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15771 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15772 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15773 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15774 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15775 PAIRED_VECTOR_MODE.
15776 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15777 (spe_func_has_64bit_regs_p): Delete.
15778 (spe_expand_predicate_builtin): Delete.
15779 (spe_expand_evsel_builtin): Delete.
15780 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15781 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15782 (invalid_e500_subreg): Delete.
15783 (rs6000_legitimize_address): Always force_reg op2 as well, for
15784 paired single memory accesses.
15785 (rs6000_member_type_forces_blk): Delete.
15786 (rs6000_spe_function_arg): Delete.
15787 (rs6000_expand_unop_builtin): Delete SPE handling.
15788 (rs6000_expand_binop_builtin): Ditto.
15789 (spe_expand_stv_builtin): Delete.
15790 (bdesc_2arg_spe): Delete.
15791 (spe_expand_builtin): Delete.
15792 (spe_expand_predicate_builtin): Delete.
15793 (spe_expand_evsel_builtin): Delete.
15794 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15795 (spe_init_builtins): Delete.
15796 (spe_func_has_64bit_regs_p): Delete.
15797 (savres_routine_name): Delete "info" parameter. Adjust callers.
15798 (rs6000_emit_stack_reset): Ditto.
15799 (rs6000_dwarf_register_span): Delete.
15800 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15801 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15802 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15804 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15806 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15807 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15808 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15809 mulv2sf3, divv2sf3): Delete expanders.
15811 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15813 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15815 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15817 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15818 * config/rs6000/rs6000.c: Ditto.
15820 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15822 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15823 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15824 comparison_operator.
15826 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15828 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15829 * config/rs6000/rs6000.opt: Ditto.
15830 * config/rs6000/t-rtems: Ditto.
15832 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15834 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15835 TARGET_E500_SINGLE by 0, simplify.
15836 * config/rs6000/rs6000.c: Ditto.
15837 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15838 (spe_build_register_parallel): Delete.
15839 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15840 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15841 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15842 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15843 (E500_CONVERT): Delete.
15844 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15848 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15849 * config/rs6000/dfp.md: Ditto.
15850 (negdd2, *negdd2_fpr): Merge.
15851 (absdd2, *absdd2_fpr): Merge.
15852 (negtd2, *negtd2_fpr): Merge.
15853 (abstd2, *abstd2_fpr): Merge.
15854 * config/rs6000/e500.h: Delete file.
15855 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15856 TARGET_FPRS by 1 and simplify.
15857 * config/rs6000/rs6000-c.c: Ditto.
15858 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15859 TARGET_DF_SPE by 0.
15860 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15862 * config/rs6000/rs6000.md: Ditto.
15863 (floatdidf2, *floatdidf2_fpr): Merge.
15864 (move_from_CR_gt_bit): Delete.
15865 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15866 (E500_CR_IOR_COMPARE): Delete.
15867 (All patterns that require !TARGET_FPRS): Delete.
15868 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15870 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15872 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15874 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15876 * graphds.c (add_edge): Intitialize edge's attached data.
15877 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15878 pointer parameter. Call pointed function on each edge during
15879 graph traversing. Skip traversing the edge when the function
15881 (graphds_dfs, graphds_scc): Ditto.
15882 (for_each_edge): New parameter. Pass the new parameter to callback
15884 * graphds.h (skip_edge_callback): New function pointer type.
15885 (graphds_dfs, graphds_scc): New function pointer parameter.
15886 (graphds_edge_callback, for_each_edge): New parameter.
15888 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15890 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15891 out code checking if runtime alias check is possible to below ...
15892 Call the new function.
15893 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15894 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15896 2017-06-07 Marek Polacek <polacek@redhat.com>
15899 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15900 TYPE_OVERFLOW_WRAPS check.
15902 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15904 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15905 if versioning is required.
15906 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15907 peeling with the check for versioning.
15909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15911 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15912 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15913 Set true to new parameter if new ssa variable is defined.
15914 (vect_gen_vector_loop_niters): Refactor. Set range information
15915 for the new vector loop bound variable.
15916 (vect_do_peeling): Ditto.
15918 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15920 * tree-affine.c (ssa.h): Include header file.
15921 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15922 has wrapping overflow behavior.
15924 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15926 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15928 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15930 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15931 (tree_to_aff_combination): ... here.
15933 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15935 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15936 reg_pressure model function.
15937 (ivopts_global_cost_for_size): Delete.
15938 (determine_set_costs, iv_ca_recount_cost): Call new model function
15939 ivopts_estimate_reg_pressure.
15941 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15943 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15944 expensive than udiv. Remove floating point cases from mod.
15946 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15948 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15949 Increase idiv cost.
15951 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15953 * config/aarch64/aarch64.md
15954 (copysignsf3): Fix mask generation.
15956 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15958 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15960 (class dump_manager): Add register_dumps method.
15961 * dumpfile.c: Include langhooks.h.
15962 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15963 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15964 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15965 (dump_manager::dump_register): Start with 512 entries instead of 32.
15966 (dump_manager::register_dumps): New method.
15967 * toplev.c (general_init): Instead of invoking register_dumps
15968 langhook, invoke register_dumps method on the dump manager.
15969 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15972 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15974 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15975 Say that # requires an associated define_split to exist, and that
15976 the define_split must be suitable for use after register allocation.
15978 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15980 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15981 (compute_outgoing_frequencies): Also initialize zero counts.
15982 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15983 around loops; preserve more of profile when nothing changes.
15985 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15987 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15989 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15990 * config/arm/arm-cpu-cdata.h: Regenerate.
15991 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15992 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15993 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15994 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15995 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15997 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15999 * config/arm/t-rmprofile: Likewise.
16000 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16002 2017-06-06 David S. Miller <davem@davemloft.net>
16005 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16006 blockage if function uses alloca.
16008 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16010 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16011 New "uid" fields to hold pretty-print IDs of group and ref.
16012 Memory references are now identified as <group_id>:<ref_id>
16013 instead of using [random] addresses.
16014 (dump_mem_details): Simplify, no functional change.
16015 (dump_mem_ref): Simplify and make output more concise.
16016 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16017 (find_or_create_group): Initialize group uid.
16018 (record_ref): Initialize ref uid. Improve debug output.
16019 (prune_group_by_reuse, should_issue_prefetch_p,)
16020 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16021 (mark_nontemporal_store, determine_loop_nest_reuse):
16022 Improve debug output.
16024 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16026 * dbgcnt.def (prefetch): New debug counter.
16027 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16028 (schedule_prefetches): Stop issueing prefetches if debug counter
16031 2017-06-06 Tom de Vries <tom@codesourcery.com>
16033 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16034 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16036 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16038 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16039 Use aarch64_reg_or_zero predicate for operand 4.
16040 (aarch64_compare_and_swap<mode> define_insn_and_split):
16041 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16042 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16044 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16046 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16047 (arm_compute_save_core_reg_mask): This.
16048 (thumb1_compute_save_reg_mask): Rename into ...
16049 (thumb1_compute_save_core_reg_mask): This.
16050 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16051 (arm_compute_frame_layout): Likewise.
16053 2017-06-06 Richard Biener <rguenther@suse.de>
16055 PR tree-optimization/80974
16056 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16057 keep or clear leaders SSA info.
16059 2017-06-06 Tom de Vries <tom@codesourcery.com>
16061 * config/nvptx/nvptx.c (split_mode_p): New function.
16062 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16064 2017-06-06 Tom de Vries <tom@codesourcery.com>
16066 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16068 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16071 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16074 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16076 * shrink-wrap.c (handle_simple_exit): Update profile.
16077 (try_shrink_wrapping): Upate profile.
16079 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16081 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16082 (tree_guess_outgoing_edge_probabilities): New.
16083 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16084 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16086 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16088 * ipa-split.c (split_function): Initialize return bb profile.
16090 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16092 * profile.c (compute_branch_probabilities): Also initialize
16093 EXIT_BLOCK profile.
16095 2017-06-06 Richard Biener <rguenther@suse.de>
16097 PR tree-optimization/80928
16098 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16099 (vect_analyze_loop_operations): Properly guard analysis for
16101 (vect_transform_loop): Likewise.
16102 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16103 (vect_model_induction_cost): Do not cost for pure SLP.
16104 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16105 of induction in inner loop vectorization.
16106 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16107 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16108 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16110 (vect_analyze_slp_cost_1): Cost induction.
16111 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16112 (vect_get_slp_vect_defs): Likewise.
16113 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16114 (vect_transform_stmt): Handle SLP reductions.
16115 * tree-vectorizer.h (vectorizable_induction): Adjust.
16117 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16119 * config/rs6000/rs6000.c (make_resolver_func): Update
16120 init_lowered_empty_function call.
16122 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16124 * doc/invoke.texi: Document the -fprofile-abs-path option.
16125 * common.opt (fprofile-abs-path): New option.
16126 * gcov-io.h (gcov_write_filename): Declare.
16127 * gcov-io.c (gcov_write_filename): New function.
16128 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16129 * profile.c (output_location): Likewise.
16131 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16133 * shring-wrap.c: Revert accidental commit.
16135 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16137 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16139 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16141 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16143 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16144 profile in callgraph edge.
16145 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16146 (apply_scale): Likewise.
16147 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16148 Also scale profile when inlining function with zero profile.
16149 (initialize_cfun): Update exit block profile even when it is zero.
16150 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16151 when profile is read.
16153 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16155 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16156 (CLONE_*): New constants to define the processors we can generate
16157 code for with the target_clone attribute.
16158 (rs6000_clone_map): New array to identify which clone processors
16159 the current program is running on.
16160 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16161 target_clone attribute.
16162 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16163 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16164 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16165 (cpu_expand_builtin): Add support for target_clone attribute.
16166 (rs6000_valid_attribute_p): Allow "default" attribute.
16167 (get_decl_name): New debug function to simplify printing the
16168 current function name in debugging statements.
16169 (rs6000_clone_priority): New functions to support the target_clone
16170 attribute, and be able to generate code to switch between ISA 2.05
16171 through ISA 3.0 (power6 through power9).
16172 (rs6000_compare_version_priority): Likewise.
16173 (rs6000_get_function_versions_dispatcher): Likewise.
16174 (make_resolver_func): Likewise.
16175 (add_condition_to_bb): Likewise.
16176 (dispatch_function_versions): Likewise.
16177 (rs6000_generate_version_dispatcher_body): Likewise.
16178 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16179 (fusion_gpr_load_p): Fix a spacing issue.
16180 * doc/extend.texi (Common Function Attributes): Document that the
16181 PowerPC supports the target_clone attribute.
16183 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16185 * config/arm/arm.h: explain F symbol found in description of ARM
16186 register allocation in its legend.
16188 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16190 * config/mips/frame-header-opt.c: Include profile-count.h.
16191 * config/riscv/riscv.c: Include profile-count.h
16193 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16195 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16197 (sm_set_flag_if_changed): Add bbs field.
16198 (execute_sm_if_changed_flag_set): Pass BBS.
16199 (execute_sm): Update.
16201 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16203 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16206 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16208 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16209 (peephole2): New peephole2 to emit the above.
16210 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16212 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16214 * config/aarch64/aarch64.c (define_peephole2 above
16215 *sub_<shift>_<mode>): New peephole.
16217 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16219 * config/i386/i386.c (make_resolver_func): Update.
16220 * Makefile.in: Add profile-count.h and profile-count.o
16221 * auto-profile.c (afdo_indirect_call): Update to new API.
16222 (afdo_set_bb_count): Update.
16223 (afdo_propagate_edge): Update.
16224 (afdo_propagate_circuit): Update.
16225 (afdo_calculate_branch_prob): Update.
16226 (afdo_annotate_cfg): Update.
16227 * basic-block.h: Include profile-count.h
16228 (struct edge_def): Turn count to profile_count.
16229 (struct basic_block_def): Likewie.
16230 (REG_BR_PROB_BASE): Move to profile-count.h
16231 (RDIV): Move to profile-count.h
16232 * bb-reorder.c (max_entry_count): Turn to profile_count.
16233 (find_traces): Update.
16234 (rotate_loop):Update.
16235 (connect_traces):Update.
16236 (sanitize_hot_paths):Update.
16237 * bt-load.c (migrate_btr_defs): Update.
16238 * cfg.c (RDIV): Remove.
16239 (init_flow): Use alloc_block.
16240 (alloc_block): Uninitialize count.
16241 (unchecked_make_edge): Uninitialize count.
16242 (check_bb_profile): Update.
16243 (dump_edge_info): Update.
16244 (dump_bb_info): Update.
16245 (update_bb_profile_for_threading): Update.
16246 (scale_bbs_frequencies_int): Update.
16247 (scale_bbs_frequencies_gcov_type): Update.
16248 (scale_bbs_frequencies_profile_count): New.
16249 * cfg.h (update_bb_profile_for_threading): Update.
16250 (scale_bbs_frequencies_profile_count): Declare.
16251 * cfgbuild.c (compute_outgoing_frequencies): Update.
16252 (find_many_sub_basic_blocks): Update.
16253 * cfgcleanup.c (try_forward_edges): Update.
16254 (try_crossjump_to_edge): Update.
16255 * cfgexpand.c (expand_gimple_tailcall): Update.
16256 (construct_exit_block): Update.
16257 * cfghooks.c (verify_flow_info): Update.
16258 (dump_bb_for_graph): Update.
16259 (split_edge): Update.
16260 (make_forwarder_block): Update.
16261 (duplicate_block): Update.
16262 (account_profile_record): Update.
16263 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16264 (get_estimated_loop_iterations): Update.
16265 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16266 (single_likely_exit): Update.
16267 * cfgloopmanip.c (scale_loop_profile): Update.
16269 (set_zero_probability): Update.
16270 (lv_adjust_loop_entry_edge): Update.
16271 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16272 (purge_dead_edges): Update.
16273 (rtl_account_profile_record): Update.
16274 * cgraph.c (cgraph_node::create): Uninitialize count.
16275 (symbol_table::create_edge): Uninitialize count.
16276 (cgraph_update_edges_for_call_stmt_node): Update.
16277 (cgraph_edge::dump_edge_flags): Update.
16278 (cgraph_node::dump): Update.
16279 (cgraph_edge::maybe_hot_p): Update.
16280 * cgraph.h: Include profile-count.h
16281 (create_clone), create_edge, create_indirect_edge): Update.
16282 (cgraph_node): Turn count to profile_count.
16283 (cgraph_edge0: Likewise.
16284 (make_speculative, clone): Update.
16285 (create_edge): Update.
16286 (init_lowered_empty_function): Update.
16287 * cgraphclones.c (cgraph_edge::clone): Update.
16288 (duplicate_thunk_for_node): Update.
16289 (cgraph_node::create_clone): Update.
16290 * cgraphunit.c (cgraph_node::analyze): Update.
16291 (cgraph_node::expand_thunk): Update.
16292 * final.c (dump_basic_block_info): Update.
16293 * gimple-streamer-in.c (input_bb): Update.
16294 * gimple-streamer-out.c (output_bb): Update.
16295 * graphite.c (print_global_statistics): Update.
16296 (print_graphite_scop_statistics): Update.
16297 * hsa-brig.c: Include basic-block.h.
16298 * hsa-dump.c: Include basic-block.h.
16299 * hsa-gen.c (T sum_slice): Update.
16300 (convert_switch_statements):Update.
16301 * hsa-regalloc.c: Include basic-block.h.
16302 * ipa-chkp.c (chkp_produce_thunks): Update.
16303 * ipa-cp.c (struct caller_statistics): Update.
16304 (init_caller_stats): Update.
16305 (gather_caller_stats): Update.
16306 (ipcp_cloning_candidate_p): Update.
16307 (good_cloning_opportunity_p): Update.
16308 (get_info_about_necessary_edges): Update.
16309 (dump_profile_updates): Update.
16310 (update_profiling_info): Update.
16311 (update_specialized_profile): Update.
16312 (perhaps_add_new_callers): Update.
16313 (decide_about_value): Update.
16314 (ipa_cp_c_finalize): Update.
16315 * ipa-devirt.c (struct odr_type_warn_count): Update.
16316 (struct decl_warn_count): Update.
16317 (struct final_warning_record): Update.
16318 (possible_polymorphic_call_targets): Update.
16319 (ipa_devirt): Update.
16320 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16321 * ipa-icf.c (sem_function::merge): Update.
16322 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16323 * ipa-inline.c (compute_uninlined_call_time): Update.
16324 (compute_inlined_call_time): Update.
16325 (want_inline_small_function_p): Update.
16326 (want_inline_self_recursive_call_p): Update.
16327 (edge_badness): Update.
16328 (lookup_recursive_calls): Update.
16329 (recursive_inlining): Update.
16330 (inline_small_functions): Update.
16331 (dump_overall_stats): Update.
16332 (dump_inline_stats): Update.
16333 * ipa-profile.c (ipa_profile_generate_summary): Update.
16334 (ipa_propagate_frequency): Update.
16335 (ipa_profile): Update.
16336 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16337 * ipa-utils.c (ipa_merge_profiles): Update.
16338 * loop-doloop.c (doloop_modify): Update.
16339 * loop-unroll.c (report_unroll): Update.
16340 (unroll_loop_runtime_iterations): Update.
16341 * lto-cgraph.c (lto_output_edge): Update.
16342 (lto_output_node): Update.
16343 (input_node): Update.
16344 (input_edge): Update.
16345 (merge_profile_summaries): Update.
16346 * lto-streamer-in.c (input_cfg): Update.
16347 * lto-streamer-out.c (output_cfg): Update.
16348 * mcf.c (create_fixup_graph): Update.
16349 (adjust_cfg_counts): Update.
16350 (sum_edge_counts): Update.
16351 * modulo-sched.c (sms_schedule): Update.
16352 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16353 * predict.c (maybe_hot_count_p): Update.
16354 (probably_never_executed): Update.
16355 (dump_prediction): Update.
16356 (combine_predictions_for_bb): Update.
16357 (propagate_freq): Update.
16358 (handle_missing_profiles): Update.
16359 (counts_to_freqs): Update.
16360 (rebuild_frequencies): Update.
16361 (force_edge_cold): Update.
16362 * predict.h: Include profile-count.h
16363 (maybe_hot_count_p, counts_to_freqs): UPdate.
16364 * print-rtl-function.c: Do not include cfg.h
16365 * print-rtl.c: Include basic-block.h
16366 * profile-count.c: New file.
16367 * profile-count.h: New file.
16368 * profile.c (is_edge_inconsistent): Update.
16369 (correct_negative_edge_counts): Update.
16370 (is_inconsistent): Update.
16371 (set_bb_counts): Update.
16372 (read_profile_edge_counts): Update.
16373 (compute_frequency_overlap): Update.
16374 (compute_branch_probabilities): Update; Initialize and deinitialize
16376 (branch_prob): Update.
16377 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16378 (edge_gcov_count): New.
16379 (bb_gcov_count): New.
16380 * shrink-wrap.c (try_shrink_wrapping): Update.
16381 * tracer.c (better_p): Update.
16382 * trans-mem.c (expand_transaction): Update.
16383 (ipa_tm_insert_irr_call): Update.
16384 (ipa_tm_insert_gettmclone_call): Update.
16385 * tree-call-cdce.c: Update.
16386 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16387 (gimple_duplicate_sese_tail): Update.
16388 (gimple_account_profile_record): Update.
16389 (execute_fixup_cfg): Update.
16390 * tree-inline.c (copy_bb): Update.
16391 (copy_edges_for_bb): Update.
16392 (initialize_cfun): Update.
16393 (freqs_to_counts): Update.
16394 (copy_cfg_body): Update.
16395 (expand_call_inline): Update.
16396 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16397 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16398 (try_unroll_loop_completely): Update.
16399 (try_peel_loop): Update.
16400 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16401 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16402 * tree-ssa-loop-split.c (connect_loops): Update.
16403 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16404 * tree-ssa-reassoc.c (branch_fixup): Update.
16405 * tree-ssa-tail-merge.c (replace_block_by): Update.
16406 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16407 (compute_path_counts): Update.
16408 (update_profile): Update.
16409 (recompute_probabilities): Update.
16410 (update_joiner_offpath_counts): Update.
16411 (estimated_freqs_path): Update.
16412 (freqs_to_counts_path): Update.
16413 (clear_counts_path): Update.
16414 (ssa_fix_duplicate_block_edges): Update.
16415 (duplicate_thread_path): Update.
16416 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16417 (struct switch_conv_info): Update.
16418 * tree-tailcall.c (decrease_profile): Update.
16419 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16420 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16421 * value-prof.c (check_counter): Update.
16422 (gimple_divmod_fixed_value): Update.
16423 (gimple_mod_pow2): Update.
16424 (gimple_mod_subtract): Update.
16425 (gimple_ic_transform): Update.
16426 (gimple_stringop_fixed_value): Update.
16427 * value-prof.h (gimple_ic): Update.
16429 2017-06-02 Carl Love <cel@us.ibm.com>
16431 * config/rs6000/rs6000-c: Add support for built-in functions
16432 vector double vec_doublee (vector signed int);
16433 vector double vec_doublee (vector unsigned int);
16434 vector double vec_doublee (vector float);
16435 vector double vec_doubleh (vector signed int);
16436 vector double vec_doubleh (vector unsigned int);
16437 vector double vec_doubleh (vector float);
16438 vector double vec_doublel (vector signed int);
16439 vector double vec_doublel (vector unsigned int);
16440 vector double vec_doublel (vector float);
16441 vector double vec_doubleo (vector signed int);
16442 vector double vec_doubleo (vector unsigned int);
16443 vector double vec_doubleo (vector float);.
16444 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16445 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16447 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16448 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16449 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16451 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16452 vec_doublel, vec_doubleh.
16453 * doc/extend.texi: Update the built-in documentation file for the
16454 new built-in functions.
16456 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16459 * ipa-inline-analysis.c (free_growth_caches): Set
16460 edge_removal_hook_holder to NULL after removing it.
16462 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16464 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16465 comparision with zero.
16467 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16469 for early expansion of vec_min and vec_max builtins.
16470 (builtin_function_type): Add min/max unsigned variants to those
16471 identified as having unsigned arguments.
16473 2017-06-02 Olivier Hainque <hainque@adacore.com>
16475 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16477 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16479 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16480 Use VALL_F16 iterator rather than VALL.
16482 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16484 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16485 Emit CBNZ inside loop when doing a strong exchange and comparing
16486 against zero. Generate the CC flags after the loop.
16488 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16490 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16491 (dl_section_ref): New.
16492 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16493 On AIX, append an expression to subtract the size of the
16494 section length to dl_section_ref.
16496 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16498 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16499 for early expansion of vector absolute builtins.
16501 2017-06-02 Richard Biener <rguenther@suse.de>
16503 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16504 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16506 2017-06-02 Richard Biener <rguenther@suse.de>
16508 PR tree-optimization/80948
16509 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16510 stmt order as well.
16512 2017-06-02 Richard Biener <rguenther@suse.de>
16514 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16516 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16517 for inductions as relevant.
16519 2017-06-02 Richard Biener <rguenther@suse.de>
16521 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16522 (vectorizable_induction): ... this. Remove dead code.
16524 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16526 * builtins. (expand_builtin_alloca): Remove second parameter and
16527 infer its value from the first parameter instead.
16528 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16530 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16532 PR rtl-optimization/80903
16533 * loop-doloop.c (add_test): Unshare sequence.
16535 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16537 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16539 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16541 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16543 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16544 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16545 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16546 (xlogue_layout::get_stub_rtx): Make static.
16547 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16548 (xlogue_layout::compute_stub_managed_regs): Rename to...
16549 (xlogue_layout::count_stub_managed_regs): ...this.
16550 (xlogue_layout::is_stub_managed_reg): New function.
16551 (xlogue_layout::m_stub_names): Rename to...
16552 (xlogue_layout::s_stub_names): ...this, make static.
16553 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16554 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16555 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16556 xlogue_layout::s_stub_names): Instantiate statics.
16557 (stub_managed_regs): Remove.
16558 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16559 (disable_call_ms2sysv_xlogues): Rename to...
16560 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16561 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16563 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16564 change after reload_completed.
16565 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16567 (ix86_expand_prologue): Likewise.
16568 (ix86_expand_epilogue): Likewise.
16569 (ix86_expand_split_stack_prologue): Likewise.
16570 (ix86_compute_frame_layout): Remove frame parameter ...
16571 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16572 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16574 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16575 (ix86_frame): Move from here ...
16576 * config/i386/i386.h (ix86_frame): ... to here.
16577 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16578 complete ix86_frame data structure instead. Remove some_ld_name.
16580 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16582 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16583 symbols that hold a DECL_VALUE_EXPR.
16585 2017-06-01 Martin Jambor <mjambor@suse.cz>
16587 PR tree-optimization/80898
16588 * tree-sra.c (process_subtree_disqualification): Removed.
16589 (disqualify_candidate): Do not acll
16590 process_subtree_disqualification.
16591 (subtree_mark_written_and_enqueue): New function.
16592 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16593 RHS has been disqualified and re-queue LHS if necessary. Apart
16594 from that, ignore disqualified RHS.
16596 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16598 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16599 address fetch for z10 or later.
16601 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16603 * config/arc/arc.md (tst_movb): Add guard when splitting.
16605 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16607 * config/arc/arc.c (arc_can_eliminate): Test against
16608 arc_frame_pointer_needed.
16610 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16612 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16613 to prevent store reordering.
16614 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16615 (type): Add block type.
16616 (stack_tie): Define special instruction to be used in
16619 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16621 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16622 constraint. It is not valid for the pattern.
16623 (noncommutative_binary_comparison): Likewise.
16625 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16627 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16630 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16632 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16633 be used by the reg-alloc.
16635 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16637 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16638 reg-alloc when having mul64 or mul32x16 instructions.
16639 (mulsidi3): Likewise.
16640 (umulsidi3): Likewise.
16641 (mulsi32x16): New pattern.
16642 (mulsi64): Likewise.
16643 (mulsidi64): Likewise.
16644 (umulsidi64): Likewise.
16645 (MUL32x16_REG): Define.
16646 (mul64_600): Use MUL32x16_REG.
16647 (mac64_600): Likewise.
16648 (umul64_600): Likewise.
16649 (umac64_600): Likewise.
16651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16653 * config/arc/arc.md (mulsi3_700): Make it commutative.
16655 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16657 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16659 (*sign_extendsidi2_insn): Likewise for movstosw.
16661 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16663 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16664 the type of the input discriminant value. Convert the
16665 discriminant value of signedness vary.
16667 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16669 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16670 Add to -Wall section.
16672 2017-06-01 Richard Biener <rguenther@suse.de>
16674 PR middle-end/66313
16675 * fold-const.c (fold_plusminus_mult_expr): If the factored
16676 factor may be zero use a wrapping type for the inner operation.
16677 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16678 and handle moved defs.
16679 (process_assignment): Properly guard the unary op case. Return a
16680 tri-state indicating that moving the stmt before the call may allow
16681 to continue. Pass through to_move.
16682 (find_tail_calls): Handle moving unrelated defs before
16685 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16688 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16689 splitter result in the canonical way.
16691 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16693 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16694 also for 32bit target. Update insn attributes.
16695 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16697 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16699 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16700 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16701 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16702 (_mm_maskz_min_ss): New intrinsics.
16704 2017-05-31 Martin Liska <mliska@suse.cz>
16706 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16707 Change comment style to one we normally use.
16708 (vectorizable_reduction): Likewise.
16709 (vectorizable_induction): Likewise.
16710 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16711 (vectorizable_call): Likewise.
16712 (vectorizable_simd_clone_call): Likewise.
16713 (vectorizable_conversion): Likewise.
16714 (vectorizable_assignment): Likewise.
16715 (vectorizable_shift): Likewise.
16716 (vectorizable_operation): Likewise.
16717 (vectorizable_store): Likewise.
16718 (vectorizable_load): Likewise.
16719 * tree-vectorizer.h: Likewise.
16721 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16723 * passes.c (emergency_dump_function): New.
16724 * tree-pass.h (emergency_dump_function): Declare.
16725 * plugin.c (plugins_internal_error_function): Remove.
16726 * plugin.h (plugins_internal_error_function): Remove declaration.
16727 * toplev.c (internal_error_function): New static function. Use it...
16728 (general_init): ...here.
16730 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16732 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16733 (arc_rtx_costs): Add costs for new patterns.
16734 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16735 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16737 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16739 * tree-ssa-strlen.c (get_next_strinfo): New function.
16740 (get_stridx_plus_constant): Use it.
16741 (zero_length_string): Likewise.
16742 (adjust_related_strinfos): Likewise.
16743 (adjust_last_stmt): Likewise.
16745 2017-05-31 Richard Biener <rguenther@suse.de>
16748 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16749 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16751 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16753 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16754 loop_vinfo argument and use of dependence distance vectors.
16755 Check instead whether the two references differ only in their
16756 initial value and assume that they have the same alignment if the
16757 difference is a multiple of the vector alignment.
16758 (vect_analyze_data_refs_alignment): Update call accordingly.
16760 2017-05-31 Martin Liska <mliska@suse.cz>
16763 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16765 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16767 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16768 (create_intersect_range_checks): Move from ...
16769 * tree-data-ref.c (create_intersect_range_checks_index)
16770 (create_intersect_range_checks): ... to here.
16771 (create_runtime_alias_checks): New function factored from ...
16772 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16773 here. Call above function.
16774 * tree-data-ref.h (create_runtime_alias_checks): New function.
16776 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16778 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16779 segment length for dr_b and compute it in wide_int.
16781 2017-05-31 Richard Biener <rguenther@suse.de>
16783 PR tree-optimization/80906
16784 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16785 and pass through iv_map.
16786 (copy_bb_and_scalar_dependences): Adjust.
16787 (translate_pending_phi_nodes): Likewise.
16788 (copy_loop_close_phi_args): Handle code-generating IVs instead
16791 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16793 * diagnostic-color.c (color_dict): Add "type-diff".
16794 (parse_gcc_colors): Update comment.
16795 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16796 -fdiagnostics-show-template-tree and -fno-elide-type.
16797 (GCC_COLORS): Add type-diff to example.
16799 (-fdiagnostics-show-template-tree): New.
16800 (-fno-elide-type): New.
16801 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16802 the pp_format_decoder callback. Call any m_format_postprocessor's
16804 (pretty_printer::pretty_printer): Initialize
16805 m_format_postprocessor.
16806 (pretty_printer::~pretty_printer): Delete any
16807 m_format_postprocessor.
16808 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16809 (class format_postprocessor): New class.
16810 (struct pretty_printer::format_decoder): Document the new parameters.
16811 (struct pretty_printer::m_format_postprocessor): New field.
16812 * tree-diagnostic.c (default_tree_printer): Update for new
16813 bool and const char ** params.
16814 * tree-diagnostic.h (default_tree_printer): Likewise.
16816 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16818 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16819 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16820 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16821 rs6000_gen_cell_microcode code.
16822 (rs6000_final_prescan_insn): Delete.
16823 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16824 "warn-cell-microcode" entries.
16825 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16826 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16827 throughout. Change cc_reg_not_micro_cr0_operand to
16828 cc_reg_not_cr0_operand throughout.
16829 (*extendhi<mode>2_noload): Delete.
16830 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16831 (mwarn-cell-microcode): Delete.
16832 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16833 -mgen-cell-microcode and -mwarn-cell-microcode.
16835 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16838 * config/i386/constraints.md (Yd): New constraint.
16840 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16841 and (?Yd, r) alternatives. Update insn attributes.
16842 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16843 and (?*Yd, r) alternatives. Update insn attributes.
16844 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16846 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16848 * gimplify.c (gimplify_modify_expr): Don't create a
16849 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16852 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16854 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16856 2017-05-30 Richard Biener <rguenther@suse.de>
16858 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16859 and reduc_def fields.
16860 (STMT_VINFO_REDUC_TYPE): New define.
16861 (STMT_VINFO_REDUC_DEF): Likewise.
16862 (vect_force_simple_reduction): Adjust prototype.
16863 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16864 (vect_is_simple_reduction): Remove check_reduction argument.
16865 (vect_force_simple_reduction): Adjust and set
16866 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16867 (vectorizable_reduction): Do not re-do reduction analysis
16868 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16869 * tree-parloops.c (gather_scalar_reductions): Adjust.
16871 2017-05-30 Richard Biener <rguenther@suse.de>
16873 PR middle-end/80901
16874 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16877 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16879 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16880 Introduce unknown_misalignment parameter and remove vf.
16881 (vect_peeling_hash_get_lowest_cost):
16882 Pass unknown_misalignment parameter.
16883 (vect_enhance_data_refs_alignment):
16884 Fix unsupportable data ref treatment.
16886 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16888 * tree-vect-data-refs.c (vect_get_data_access_cost):
16889 Workaround for SLP handling.
16890 (vect_enhance_data_refs_alignment):
16891 Compute costs for doing no peeling at all, compare to the best
16892 peeling costs so far and avoid peeling if cheaper.
16894 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16896 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16897 Return peeling info and set costs to zero for unlimited cost
16899 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16900 with unknown misalignment. Compute and costs for unknown
16901 misalignment, compare them to the costs for known misalignment
16902 and choose the cheapest for peeling.
16904 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16906 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16907 (vect_get_peeling_costs_all_drs): Create function.
16908 (vect_peeling_hash_get_lowest_cost):
16909 Use vect_get_peeling_costs_all_drs.
16910 (vect_peeling_supportable): Create function.
16911 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16913 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16915 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16916 DR_HAS_NEGATIVE_STEP.
16917 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16918 (vect_enhance_data_refs_alignment): Use.
16919 (vect_duplicate_ssa_name_ptr_info): Use.
16920 * tree-vectorizer.h (dr_misalignment): Use.
16921 (known_alignment_for_access_p): Use.
16923 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16926 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16928 (has_section_name): New function.
16930 2017-05-30 Martin Liska <mliska@suse.cz>
16933 * auto-profile.c (get_function_decl_from_block): Fix
16936 2017-05-30 Richard Biener <rguenther@suse.de>
16938 PR middle-end/80876
16939 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16941 2017-05-30 Martin Liska <mliska@suse.cz>
16943 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16944 * dumpfile.h (struct dump_file_info): Remove ctors.
16946 2017-05-30 Martin Liska <mliska@suse.cz>
16948 * predict.def: Fix GNU coding style.
16950 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16952 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16953 Mark 'to' argument with ATTRIBUTE_UNUSED.
16955 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16957 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16958 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16959 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16962 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16964 * doc/install.texi (Options specification): Restore entry of
16965 --enable-sjlj-exceptions.
16967 2017-05-27 Michael Eager <eager@eagercon.com>
16970 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16972 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16974 * config/microblaze/microblaze.h
16975 (FIXED_REGISTERS): Update in macro.
16976 (CALL_USED_REGISTERS): Update in macro.
16978 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16980 * doc/install.texi: Add links to macOS binary distributions.
16982 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16986 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16988 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16990 2017-05-26 Martin Liska <mliska@suse.cz>
16992 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16994 2017-05-26 Martin Liska <mliska@suse.cz>
16996 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16997 always leading ';; '.
16998 (dump_bb_info): Likewise.
16999 (brief_dump_cfg): Likewise.
17000 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17001 * dumpfile.c: Remove usage of TDF_VERBOSE.
17002 * dumpfile.h (enum dump_kind): Likewise.
17003 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17004 * print-tree.c (debug_verbose): Remove.
17005 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17006 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17007 * tree-diagnostic.c (default_tree_printer): Replace
17008 TDF_DIAGNOSTIC with TDF_SLIM.
17010 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17012 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17013 in parameter loop, rather than loop_vinfo.
17014 (create_intersect_range_checks): Ditto.
17015 (vect_create_cond_for_alias_checks): Update call to above functions.
17017 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17019 PR tree-optimization/80815
17020 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17021 for merging runtime alias checks. Handle negative DR_STEPs.
17023 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17025 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17027 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17028 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17029 out code pruning runtime alias checks.
17030 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17031 factored out from above.
17032 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17034 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17036 (prune_runtime_alias_test_list): New decalaration.
17038 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17040 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17041 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17042 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17043 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17044 (operator==, comp_dr_with_seg_len_pair): Ditto.
17045 (vect_prune_runtime_alias_test_list): Ditto.
17047 2017-05-26 Martin Liska <mliska@suse.cz>
17050 * params.def: Bound partial-inlining-entry-probability param.
17052 2017-05-26 Marek Polacek <polacek@redhat.com>
17055 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17058 2017-05-26 Richard Biener <rguenther@suse.de>
17060 PR tree-optimization/80842
17061 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17064 2017-05-26 Richard Biener <rguenther@suse.de>
17066 PR tree-optimization/80844
17067 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17069 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17071 * doc/md.texi (Machine Constraints): Update x86 family
17072 machine constraints section to match 'config/i386/constraints.md'.
17074 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17076 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17078 2017-05-25 Nathan Sidwell <nathan@acm.org>
17080 * doc/invoke.texi (--enable-languages): Update documentation.
17082 2017-05-25 Martin Liska <mliska@suse.cz>
17084 * dumpfile.c: Add TDF_FOLDING.
17085 * dumpfile.h (enum dump_kind): Likewise.
17086 * genmatch.c (dt_simplify::gen_1): Use it.
17088 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17090 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17092 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17094 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17095 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17097 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17099 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17100 * match.pd (X == C): Rewrite it here.
17101 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17102 with_certain_nonzero_bits2): New predicates.
17103 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17105 2017-05-24 Nathan Sidwell <nathan@acm.org>
17107 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17110 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17113 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17115 * config/powerpcspe: New port. Files are copied from the rs6000
17116 port, with "rs6000" in filenames replaced by "powerpcspe".
17118 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17120 PR rtl-optimization/80754
17121 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17123 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17125 * config/sparc/sparc.md (length): Return the correct value for -mflat
17126 sibcalls to match output_sibcall.
17128 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17132 * config/rs6000/rs6000.c (struct machine_function): Add new field
17134 (rs6000_get_separate_components): Init that field, use it.
17135 (rs6000_components_for_bb): Use the field.
17137 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17139 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17141 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17143 PR middle-end/80823
17144 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17146 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17149 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17150 address against address_operand predicate.
17151 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17153 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17155 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17156 parameters passed indirectly.
17158 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17160 * config/i386/i386.md (*movdi_internal): Remove SSE4
17161 alternative 18 (?r, *v). Update insn attributes.
17162 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17163 Update insn attributes.
17164 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17165 Update insn attributes.
17166 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17167 alternative 1 (r, v). Remove isa attribute.
17168 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17169 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17170 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17172 2017-05-23 Tom de Vries <tom@codesourcery.com>
17174 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17177 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17179 * cgraphunit.c (symbol_table::process_new_functions): Update.
17180 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17181 (inline_generate_summary): Rename to ...
17182 (ipa_fn_summary_generate): ... this one.
17183 (inline_read_summary): Rename to ...
17184 (ipa_fn_summary_read): ... this one.
17185 (inline_write_summary): Rename to ...
17186 (ipa_fn_summary_write): ... this one.
17187 (inline_free_summary): Rename to ...
17188 (ipa_free_fn_summary): ... this one.
17189 (pass_data_local_fn_summary, pass_local_fn_summary,
17190 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17191 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17192 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17193 make_pass_ipa_fn_summary): New.
17194 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17195 inline_write_summary, inline_free_summary): Remove.
17196 (ipa_free_fn_summary) : New.
17197 * ipa-inline.c (ipa_inline): Update.
17198 (pass_ipa_inline): Do not generate summaries.
17199 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17201 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17202 and add pass_ipa_fn_summary.
17203 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17205 (make_pass_inline_parameters): Remove.
17207 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17209 * omp-low.c (struct omp_context): Remove "default_kind" member.
17212 * omp-offload.c (execute_oacc_device_lower): Remove the
17213 parallelism dimensions function attributes for unparallelized
17214 OpenACC kernels constructs.
17216 2017-05-23 Martin Liska <mliska@suse.cz>
17218 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17220 (cgraph_edge::make_speculative): Likewise.
17221 (cgraph_edge::resolve_speculation): Likewise.
17222 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17223 (cgraph_node::dump): Likewise.
17224 * cgraph.h: Likewise.
17225 * cgraphunit.c (analyze_functions): Likewise.
17226 (symbol_table::compile): Likewise.
17227 * ipa-cp.c (print_all_lattices): Likewise.
17228 (determine_versionability): Likewise.
17229 (initialize_node_lattices): Likewise.
17230 (ipcp_verify_propagated_values): Likewise.
17231 (estimate_local_effects): Likewise.
17232 (update_profiling_info): Likewise.
17233 (create_specialized_node): Likewise.
17234 (perhaps_add_new_callers): Likewise.
17235 (decide_about_value): Likewise.
17236 (decide_whether_version_node): Likewise.
17237 (identify_dead_nodes): Likewise.
17238 (ipcp_store_bits_results): Likewise.
17239 * ipa-devirt.c (dump_targets): Likewise.
17240 (ipa_devirt): Likewise.
17241 * ipa-icf.c (sem_item::dump): Likewise.
17242 (sem_function::equals): Likewise.
17243 (sem_variable::equals): Likewise.
17244 (sem_item_optimizer::read_section): Likewise.
17245 (sem_item_optimizer::execute): Likewise.
17246 (congruence_class::dump): Likewise.
17247 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17248 (dump_inline_summary): Likewise.
17249 (estimate_node_size_and_time): Likewise.
17250 (inline_analyze_function): Likewise.
17251 * ipa-inline-transform.c (inline_call): Likewise.
17252 * ipa-inline.c (report_inline_failed_reason): Likewise.
17253 (want_early_inline_function_p): Likewise.
17254 (edge_badness): Likewise.
17255 (update_edge_key): Likewise.
17256 (inline_small_functions): Likewise.
17257 * ipa-profile.c (ipa_profile): Likewise.
17258 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17259 (ipa_make_edge_direct_to_target): Likewise.
17260 (remove_described_reference): Likewise.
17261 (ipa_impossible_devirt_target): Likewise.
17262 (propagate_controlled_uses): Likewise.
17263 (ipa_print_node_params): Likewise.
17264 (ipcp_transform_function): Likewise.
17265 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17266 (propagate_pure_const): Likewise.
17267 * ipa-reference.c (generate_summary): Likewise.
17268 (read_write_all_from_decl): Likewise.
17269 (propagate): Likewise.
17270 (ipa_reference_read_optimization_summary): Likewise.
17271 * ipa-utils.c (ipa_merge_profiles): Likewise.
17272 * ipa.c (walk_polymorphic_call_targets): Likewise.
17273 (symbol_table::remove_unreachable_nodes): Likewise.
17274 (ipa_single_use): Likewise.
17275 * passes.c (execute_todo): Likewise.
17276 * predict.c (drop_profile): Likewise.
17277 * symtab.c (symtab_node::get_dump_name): New function.
17278 (symtab_node::dump_name): Likewise.
17279 (symtab_node::dump_asm_name): Likewise.
17280 (symtab_node::dump_references): Likewise.
17281 (symtab_node::dump_referring): Likewise.
17282 (symtab_node::dump_base): Likewise.
17283 (symtab_node::debug_symtab): Likewise.
17284 * tree-sra.c (convert_callers_for_node): Likewise.
17285 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17286 * value-prof.c (init_node_map): Likewise.
17288 2017-05-23 Martin Liska <mliska@suse.cz>
17290 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17291 and symtab_node::debug_symtab to symbol_table::debug.
17292 * cgraphunit.c (analyze_functions): Use the renamed function.
17293 (symbol_table::compile): Likewise.
17294 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17295 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17296 * passes.c (execute_todo): Likewise.
17297 * symtab.c (symbol_table::dump): New function.
17298 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17300 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17302 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17303 that nonconst implies exec.
17305 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17307 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17308 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17309 (inline_edge_summary_vec): Turn into ...
17310 (ipa_call_summaries): ... this one.
17311 (redirect_to_unreachable, edge_set_predicate,
17312 evaluate_properties_for_edge, inline_summary_alloc,
17313 reset_ipa_call_summary, reset_inline_summary,
17314 inline_summary_t::duplicate): Update.
17315 (inline_edge_duplication_hook): Turn to ...
17316 (ipa_call_summary_t::duplicate): ... this one.
17317 (inline_edge_removal_hook): Turn to ...
17318 (ipa_call_summary_t::remove): ... this one.
17319 (dump_inline_edge_summary): Turn to ...
17320 (dump_ipa_call_summary): ... this one.
17321 (estimate_function_body_sizes): Update.
17322 (inline_update_callee_summaries): Update.
17323 (remap_edge_change_prob): Update.
17324 (remap_edge_summaries): Update.
17325 (inline_merge_summary): Update.
17326 (do_estimate_edge_time): Update.
17327 (inline_generate_summary): Update.
17328 (inline_read_section): Update.
17329 (inline_read_summary): Update.
17330 (inline_free_summary): Update.
17331 * ipa-inline.c (can_inline_edge_p): Update.
17332 (compute_inlined_call_time): Update.
17333 (want_inline_small_function_p): Update.
17334 (edge_badness): Update.
17335 (early_inliner): Update.
17336 * ipa-inline.h (inline_edge_summary): Turn to ...
17337 (ipa_call_summary): ... this one.
17338 (ipa_call_summary_t): New class.
17339 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17340 (ipa_call_summaries): New.
17341 (inline_edge_summary): Remove.
17342 (estimate_edge_growth): Update.
17343 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17344 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17345 * ipa-split.c (execute_split_functions): Update.
17346 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17348 2017-05-23 Tom de Vries <tom@codesourcery.com>
17350 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17351 attributes): Document rdrand effective target.
17353 2017-05-23 Tom de Vries <tom@codesourcery.com>
17355 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17356 attributes): Sort alphabetically.
17358 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17360 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17362 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17365 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17366 V2DF/V2DI splat into two separate patterns, one that handles
17367 registers, and the other that only handles memory. Drop support
17368 for splatting from a GPR on ISA 2.07 and then splitting the
17369 splat into direct move and splat.
17370 (vsx_splat_<mode>_reg): Likewise.
17371 (vsx_splat_<mode>_mem): Likewise.
17373 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17375 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17377 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17379 PR middle-end/80809
17380 * omp-low.c (finish_taskreg_remap): New function.
17381 (finish_taskreg_scan): If unit size of ctx->record_type
17382 is non-constant, unshare the size expression and replace
17383 decls in it with possible outer var refs.
17385 PR middle-end/80809
17386 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17387 GOVD_SHARED rather than GOVD_PRIVATE with it.
17388 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17389 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17391 PR middle-end/80853
17392 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17393 as last argument to build_outer_var_ref for pointer bases of array
17394 section reductions.
17396 2017-05-19 Martin Sebor <msebor@redhat.com>
17398 * print-tree.c (print_node): Print DECL_READ_P flag.
17400 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17402 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17403 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17404 * cgraph.c: Likewise.
17405 * cgraphunit.c: Likewise.
17406 * gengtype.c: Likewise.
17407 * ipa-cp.c: Likewise.
17408 * ipa-devirt.c: Likewise.
17409 * ipa-icf.c: Likewise.
17410 * ipa-predicate.c: Likewise.
17411 * ipa-profile.c: Likewise.
17412 * ipa-prop.c: Likewise.
17413 * ipa-split.c: Likewise.
17415 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17416 edge_predicate_pool, dump_inline_hints,
17417 inline_summary::account_size_time, redirect_to_unreachable,
17418 edge_set_predicate, set_hint_predicate,
17419 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17420 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17421 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17422 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17423 ipa_call_summary_t::remove, initialize_growth_caches,
17424 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17425 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17426 mark_modified, unmodified_parm_1, unmodified_parm,
17427 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17428 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17429 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17430 will_be_nonconstant_predicate, record_modified_bb_info,
17431 get_minimal_bb, record_modified, param_change_prob,
17432 phi_result_unknown_predicate, predicate_for_phi_result,
17433 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17434 estimate_function_body_sizes, compute_inline_parameters,
17435 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17436 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17437 inline_update_callee_summaries, remap_edge_change_prob,
17438 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17439 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17440 inline_analyze_function, inline_summary_t::insert,
17441 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17442 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17443 inline_free_summary): Move to ipa-fnsummary.h
17444 (predicate_t): Remove.
17445 * ipa-fnsummary.c: New file.
17446 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17447 (enum inline_hints_vals, inline_hints, agg_position_info,
17448 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17449 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17450 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17451 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17452 inline_read_summary, inline_write_summary, inline_free_summary,
17453 inline_analyze_function, initialize_inline_failed,
17454 inline_merge_summary, inline_update_overall_summary,
17455 compute_inline_parameters): Move to ipa-fnsummary.h
17456 * ipa-fnsummary.h: New file.
17457 * ipa-inline-transform.h: Include ipa-inline.h.
17458 * ipa-inline.c: LIkewise.
17460 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17462 * ipa-inline.c (edge_badness): Use inlined_time instead of
17463 inline_summaries->get.
17465 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17467 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17469 2017-05-22 Nathan Sidwell <nathan@acm.org>
17471 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17472 (fdump-lang): Document 'raw' option.
17473 * dumpfile.h (TDI_tu): Delete.
17474 * dumpfile.c (dump_files): Remove translation-unit.
17475 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17477 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17479 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17480 command option from $(AWK) call.
17481 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17482 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17483 [FORMAT]: Remove handling of variable.
17484 * config/avr/t-multilib: Regenerate.
17486 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17488 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17490 (dump_inline_summary): Do not print self_time.
17491 (estimate_function_body_sizes): Do not set self_time.
17492 (compute_inline_parameters): Likewise.
17493 (inline_read_section, inline_write_summary): Do not stream self_time.
17494 * ipa-inline.h (inline_summary): Drop self_time.
17496 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17498 * ipa-inline-analysis.c (account_size_time): Rename to ...
17499 (inline_summary::account_size_time): ... this one.
17500 (reset_ipa_call_summary): Turn to ...
17501 (ipa_call_summary::reset): ... this one.
17502 (reset_inline_summary): Turn to ...
17503 (inline_summary::reset): ... this one.
17504 (inline_summary_t::remove): Update.
17505 (inline_summary_t::duplicate): Update.
17506 (ipa_call_summary_t::remove): Update.
17507 (dump_inline_summary): Update.
17508 (estimate_function_body_sizes): Update.
17509 (compute_inline_parameters): Update.
17510 (estimate_node_size_and_time): Update.
17511 (inline_merge_summary): Update.
17512 (inline_update_overall_summary): Update.
17513 (inline_read_section): Update.
17514 (inline_write_summary): Update.
17515 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17516 add account_size_time and reset member functions.
17517 (ipa_call_summary): Add reset function.
17518 * ipa-predicate.h (predicate::operator &): Constify.
17520 2017-05-22 Richard Biener <rguenther@suse.de>
17522 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17524 2017-05-19 Jason Merrill <jason@redhat.com>
17526 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17528 2017-05-19 Marek Polacek <polacek@redhat.com>
17531 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17532 TYPE_OVERFLOW_WRAPS checks.
17534 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17536 * tree-core.h (enum omp_clause_default_kind): Add
17537 "OMP_CLAUSE_DEFAULT_PRESENT".
17538 * tree-pretty-print.c (dump_omp_clause): Handle it.
17539 * gimplify.c (enum gimplify_omp_var_data): Add
17540 "GOVD_MAP_FORCE_PRESENT".
17541 (gimplify_adjust_omp_clauses_1): Map it to
17542 "GOMP_MAP_FORCE_PRESENT".
17543 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17545 * gimplify.c (oacc_default_clause): Clarify.
17547 2017-05-19 Nathan Sidwell <nathan@acm.org>
17549 LANG_HOOK_REGISTER_DUMPS
17550 * toplev.c (general_init): Call register dump lang hook.
17551 * doc/invoke.texi: Document -fdump-lang option family.
17552 * dumpfile.c (dump_files): Remove class dump here.
17553 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17554 * dumpfile.h (tree_dump_index): Remove TDI_class.
17555 * langhooks-def.h (lhd_register_dumps): Declare.
17556 (LANG_HOOKS_REGISTER_DUMPS): Define.
17557 (LANG_HOOKS_INITIALIZER): Add it.
17558 * langhooks.c (lhd_register_dumps): Define.
17559 * langhooks.h (struct lang_hooks): Add register_dumps.
17561 2017-05-19 Nathan Sidwell <nathan@acm.org>
17563 * context.h (context::set_passes): New.
17564 * context.c (context::context): Do not create pass manager.
17565 * toplev.c (general_init): Create pass manager here.
17567 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17569 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17570 use this splitter if two add or or instructions would also work for
17571 the constant we want to generate.
17573 2017-05-19 Richard Biener <rguenther@suse.de>
17576 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17577 predicate evaluation.
17579 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17581 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17583 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17586 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17588 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17589 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17590 (inline_edge_summary_vec): Turn into ...
17591 (ipa_call_summaries): ... this one.
17592 (redirect_to_unreachable, edge_set_predicate,
17593 evaluate_properties_for_edge, inline_summary_alloc,
17594 reset_ipa_call_summary, reset_inline_summary,
17595 inline_summary_t::duplicate): Update.
17596 (inline_edge_duplication_hook): Turn to ...
17597 (ipa_call_summary_t::duplicate): ... this one.
17598 (inline_edge_removal_hook): Turn to ...
17599 (ipa_call_summary_t::remove): ... this one.
17600 (dump_inline_edge_summary): Turn to ...
17601 (dump_ipa_call_summary): ... this one.
17602 (estimate_function_body_sizes): Update.
17603 (inline_update_callee_summaries): Update.
17604 (remap_edge_change_prob): Update.
17605 (remap_edge_summaries): Update.
17606 (inline_merge_summary): Update.
17607 (do_estimate_edge_time): Update.
17608 (inline_generate_summary): Update.
17609 (inline_read_section): Update.
17610 (inline_read_summary): Update.
17611 (inline_free_summary): Update.
17612 * ipa-inline.c (can_inline_edge_p): Update.
17613 (compute_inlined_call_time): Update.
17614 (want_inline_small_function_p): Update.
17615 (edge_badness): Update.
17616 (early_inliner): Update.
17617 * ipa-inline.h (inline_edge_summary): Turn to ...
17618 (ipa_call_summary): ... this one.
17619 (ipa_call_summary_t): New class.
17620 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17621 (ipa_call_summaries): New.
17622 (inline_edge_summary): Remove.
17623 (estimate_edge_growth): Update.
17624 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17625 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17626 * ipa-split.c (execute_split_functions): Update.
17627 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17629 2017-05-19 Richard Biener <rguenther@suse.de>
17631 PR middle-end/80764
17632 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17634 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17636 * config/rs6000/rs6000.c (struct machine_function): Add field
17637 fpr_is_wrapped_separately.
17638 (rs6000_get_separate_components): Use 64 components. Handle the
17639 new FPR components.
17640 (rs6000_components_for_bb): Handle the FPR components.
17641 (rs6000_emit_prologue_components): Handle the FPR components.
17642 (rs6000_emit_epilogue_components): Handle the FPR components.
17643 (rs6000_set_handled_components): Handle the FPR components.
17644 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17645 that are already separately shrink-wrapped.
17646 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17647 that are already separately shrink-wrapped.
17649 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17652 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17655 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17656 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17657 cases where the register allocator uses a move and an offsettable
17658 memory operation to/from a FPR register on ISA 2.06/2.07.
17659 (define_peephole2 for Altivec d-form store): Likewise.
17661 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17664 * config/i386/mmx.md (*mov<mode>_internal): Enable
17665 alternatives 11, 12, 13 and 14 also for 32bit targets.
17666 Remove alternatives 15, 16, 17 and 18.
17667 * config/i386/sse.md (vec_concatv2di): Change
17668 alternative (!x, *y) to (x, ?!*Yn).
17670 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17672 * dumpfile.h (enum dump_kind): Remove stray comma.
17674 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17676 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17677 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17678 predicate::num_conditions
17679 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17680 (CHANGED): turn into predicate::changed.
17681 (agg_position_info): Move to ipa-predicate.h
17682 (add_condition, predicate::add_clause, predicate::operator &=,
17683 predicate::or_with, predicate::evaluate, predicate::probability,
17684 dump_condition, dump_clause, predicate::dump,
17685 predicate::remap_after_duplication, predicate::remap_after_inlining,
17686 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17687 (evaluate_conditions_for_known_args): Update.
17688 (set_cond_stmt_execution_predicate): Update.
17689 * ipa-inline.h: Include ipa-predicate.h
17690 (condition, inline_param_summary, conditions, agg_position_info,
17691 predicate): Move to ipa-predicate.h
17692 * ipa-predicate.c: New file.
17693 * ipa-predicate.h: New file.
17695 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17697 * final.c (leaf_function_p): Check we are not in a sequence.
17699 2017-05-18 Martin Liska <mliska@suse.cz>
17701 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17702 * dumpfile.c (dump_register): Use new enum dump_kind.
17703 (get_dump_file_name): Likewise.
17704 (dump_enable_all): Likewise.
17705 (dump_switch_p_1): Likewise.
17706 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17707 * dumpfile.h (enum dump_kind): New enum type.
17708 (struct dump_file_info): Create constructor and
17709 format fields and comments.
17710 * passes.c (pass_manager::register_one_dump_file):
17712 * statistics.c (statistics_early_init): Likewise.
17713 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17714 TDF_TREE with TDF_SLIM.
17715 (gather_memory_references_ref): Likewise.
17717 2017-05-18 Martin Liska <mliska@suse.cz>
17719 * vec.h (struct vnull): Use it.
17721 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17723 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17724 (true_predicate, false_predicate, true_predicate_p,
17725 false_predicate_p): Remove.
17726 (single_cond_predicate, not_inlined_predicate): Turn to member function
17728 (add_condition): Update.
17729 (add_clause): Turn to...
17730 (predicate::add_clause): ... this one; update; allow passing NULL
17732 (and_predicates): Turn to ...
17733 (predicate::operator &=): ... this one.
17734 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17735 (or_predicates): Turn to ...
17736 (predicate::or_with): ... this one.
17737 (evaluate_predicate): Turn to ...
17738 (predicate::evaluate): ... this one.
17739 (predicate_probability): Turn to ...
17740 (predicate::probability): ... this one.
17741 (dump_condition): Update.
17742 (dump_predicate): Turn to ...
17743 (predicate::dump): ... this one.
17744 (account_size_time): Update.
17745 (edge_set_predicate): Update.
17746 (set_hint_predicate): UPdate.
17747 (evaluate_conditions_for_known_args): Update.
17748 (evaluate_properties_for_edge): Update.
17749 (remap_predicate_after_duplication): Turn to...
17750 (predicate::remap_after_duplication): ... this one.
17751 (remap_hint_predicate_after_duplication): Update.
17752 (inline_summary_t::duplicate): UPdate.
17753 (dump_inline_edge_summary): Update.
17754 (dump_inline_summary): Update.
17755 (set_cond_stmt_execution_predicate): Update.
17756 (set_switch_stmt_execution_predicate): Update.
17757 (compute_bb_predicates): Update.
17758 (will_be_nonconstant_expr_predicate): Update.
17759 (will_be_nonconstant_predicate): Update.
17760 (phi_result_unknown_predicate): Update.
17761 (predicate_for_phi_result): Update.
17762 (array_index_predicate): Update.
17763 (estimate_function_body_sizes): Update.
17764 (estimate_node_size_and_time): Update.
17765 (estimate_ipcp_clone_size_and_time): Update.
17766 (remap_predicate): Rename to ...
17767 (predicate::remap_after_inlining): ... this one.
17768 (remap_hint_predicate): Update.
17769 (inline_merge_summary): Update.
17770 (inline_update_overall_summary): Update.
17771 (estimate_size_after_inlining): Update.
17772 (read_predicate): Rename to ...
17773 (predicate::stream_in): ... this one.
17774 (read_inline_edge_summary): Update.
17775 (write_predicate): Rename to ...
17776 (predicate::stream_out): ... this one.
17777 (write_inline_edge_summary): Update.
17778 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17779 (clause_t): Turn to uint32_t
17780 (predicate): Turn to class; implement constructor and operators
17782 (size_time_entry): Update.
17783 (inline_summary): Update.
17784 (inline_edge_summary): Update.
17786 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17788 * fold-const.c (fold_binary_loc): Move transformation...
17789 * match.pd (C - X CMP X): ... here.
17791 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17793 * config/sparc/sparc.c (sparc_option_override): Set function
17794 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17795 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17797 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17799 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17801 2017-05-18 Marek Polacek <polacek@redhat.com>
17804 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17805 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17806 gimple_assign_load_p.
17808 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17810 PR middle-end/80692
17811 * real.c (do_compare): Give decimal_do_compare preference over
17812 comparing just the signs.
17814 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17816 * doc/md.texi (Canonicalization of Instructions): Describe the
17817 canonical form of instructions that inherently set a condition
17820 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17822 PR middle-end/80775
17823 * tree-cfg.c: Move deletion of unreachable case statements to after
17824 the merging of consecutive case labels.
17826 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17828 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17829 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17830 restoring of callee-saved registers.
17832 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17834 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17835 * config/visium/visium.c (single_set_and_flags): Likewise.
17836 * config/visium/visium.md (Substitutions): Likewise.
17838 2017-05-17 Martin Liska <mliska@suse.cz>
17840 * cfg.c: Introduce dump_flags_t type and
17841 use it instead of int type.
17843 * cfghooks.c: Likewise.
17844 * cfghooks.h (struct cfg_hooks): Likewise.
17845 * cfgrtl.c: Likewise.
17846 * cfgrtl.h: Likewise.
17847 * cgraph.c (cgraph_node::get_body): Likewise.
17848 * coretypes.h: Likewise.
17849 * domwalk.c: Likewise.
17850 * domwalk.h: Likewise.
17851 * dumpfile.c (struct dump_option_value_info): Likewise.
17852 (dump_enable_all): Likewise.
17853 (dump_switch_p_1): Likewise.
17854 (opt_info_switch_p): Likewise.
17855 * dumpfile.h (enum tree_dump_index): Likewise.
17856 (struct dump_file_info): Likewise.
17857 * genemit.c: Likewise.
17858 * generic-match-head.c: Likewise.
17859 * gengtype.c (open_base_files): Likewise.
17860 * gimple-pretty-print.c: Likewise.
17861 * gimple-pretty-print.h: Likewise.
17862 * graph.c (print_graph_cfg): Likewise.
17863 * graphite-scop-detection.c (dot_all_sese): Likewise.
17864 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17865 * loop-unroll.c (report_unroll): Likewise.
17866 * passes.c (pass_manager::register_one_dump_file): Likewise.
17867 * print-tree.c: Likewise.
17868 * statistics.c: Likewise.
17869 * tree-cfg.c: Likewise.
17870 * tree-cfg.h: Likewise.
17871 * tree-dfa.c: Likewise.
17872 * tree-dfa.h: Likewise.
17873 * tree-dump.c (dump_function): Likewise.
17874 * tree-dump.h (struct dump_info): Likewise.
17875 * tree-pretty-print.c: Likewise.
17876 * tree-pretty-print.h: Likewise.
17877 * tree-ssa-live.c: Likewise.
17878 * tree-ssa-live.h: Likewise.
17879 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17880 * tree-vect-loop.c: Likewise.
17881 * tree-vect-slp.c: Likewise.
17883 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17884 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17886 PR tree-optimization/80457
17887 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17888 of all arguments to a statement as scalar_to_vec operations.
17889 (vectorizable_call): Adjust call to vect_model_simple_cost for
17891 (vectorizable_conversion): Likewise.
17892 (vectorizable_assignment): Likewise.
17893 (vectorizable_shift): Likewise.
17894 (vectorizable_operation): Likewise.
17895 (vectorizable_comparison): Likewise.
17896 (vect_is_simple_cond): Record the def types for operands.
17897 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17898 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17899 for statement argument count.
17901 2017-05-16 Carl Love <cel@us.ibm.com>
17903 * config/rs6000/rs6000-c: Add support for built-in functions
17904 vector unsigned long long vec_bperm (vector unsigned long long,
17905 vector unsigned char)
17906 vector signed long long vec_mule (vector signed int,
17908 vector unsigned long long vec_mule (vector unsigned int,
17909 vector unsigned int)
17910 vector signed long long vec_mulo (vector signed int,
17912 vector unsigned long long vec_mulo (vector unsigned int,
17913 vector unsigned int)
17914 vector signed char vec_sldw (vector signed char,
17915 vector signed char,
17917 vector unsigned char vec_sldw (vector unsigned char,
17918 vector unsigned char,
17920 vector signed short vec_sldw (vector signed short,
17921 vector signed short,
17923 vector unsigned short vec_sldw (vector unsigned short,
17924 vector unsigned short,
17926 vector signed int vec_sldw (vector signed int,
17929 vector unsigned int vec_sldw (vector unsigned int,
17930 vector unsigned int,
17932 vector signed long long vec_sldw (vector signed long long,
17933 vector signed long long,
17935 vector unsigned long long vec_sldw (vector unsigned long long,
17936 vector unsigned long long,
17938 * config/rs6000/rs6000-c: Add support for built-in functions
17939 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17940 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17941 * doc/extend.texi: Update the built-in documentation for the
17942 new built-in functions.
17944 2017-05-16 Marek Polacek <polacek@redhat.com>
17948 * tree.c (save_expr): Don't fold the expression.
17950 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17952 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17953 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17954 and (?*y,m). Update insn attributes.
17956 2017-05-16 Martin Liska <mliska@suse.cz>
17958 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17959 flags argument of print_gimple_stmt, print_gimple_expr,
17960 print_generic_stmt and print_generic_expr.
17961 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17962 * coretypes.h: Likewise.
17963 * except.c (dump_eh_tree): Likewise.
17964 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17965 * gimple-pretty-print.h: Likewise.
17966 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17967 (backprop::push_to_worklist): Likewise.
17968 (backprop::pop_from_worklist): Likewise.
17969 (backprop::process_use): Likewise.
17970 (backprop::intersect_uses): Likewise.
17971 (note_replacement): Likewise.
17972 * gimple-ssa-store-merging.c
17973 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17974 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17975 (pass_store_merging::execute): Likewise.
17976 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17977 (ssa_base_cand_dump_callback): Likewise.
17978 (dump_incr_vec): Likewise.
17979 (replace_refs): Likewise.
17980 (replace_mult_candidate): Likewise.
17981 (create_add_on_incoming_edge): Likewise.
17982 (create_phi_basis): Likewise.
17983 (insert_initializers): Likewise.
17984 (all_phi_incrs_profitable): Likewise.
17985 (introduce_cast_before_cand): Likewise.
17986 (replace_one_candidate): Likewise.
17987 * gimplify.c (gimplify_expr): Likewise.
17988 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17989 (set_rename): Likewise.
17990 (rename_uses): Likewise.
17991 (copy_loop_phi_nodes): Likewise.
17992 (add_close_phis_to_merge_points): Likewise.
17993 (copy_loop_close_phi_args): Likewise.
17994 (copy_cond_phi_args): Likewise.
17995 (graphite_copy_stmts_from_block): Likewise.
17996 (translate_pending_phi_nodes): Likewise.
17997 * graphite-poly.c (print_pdr): Likewise.
17998 (dump_gbb_cases): Likewise.
17999 (dump_gbb_conditions): Likewise.
18000 (print_scop_params): Likewise.
18001 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18002 (build_cross_bb_scalars_use): Likewise.
18003 (gather_bbs::before_dom_children): Likewise.
18004 * hsa-dump.c (dump_hsa_immed): Likewise.
18005 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18006 (get_replacement_map): Likewise.
18007 * ipa-inline-analysis.c (dump_condition): Likewise.
18008 (estimate_function_body_sizes): Likewise.
18009 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18010 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18011 * ipa-prop.c (ipa_dump_param): Likewise.
18012 (ipa_print_node_jump_functions_for_edge): Likewise.
18013 (ipa_modify_call_arguments): Likewise.
18014 (ipa_modify_expr): Likewise.
18015 (ipa_dump_param_adjustments): Likewise.
18016 (ipa_dump_agg_replacement_values): Likewise.
18017 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18018 * ipa-pure-const.c (check_stmt): Likewise.
18019 (pass_nothrow::execute): Likewise.
18020 * ipa-split.c (execute_split_functions): Likewise.
18021 * omp-offload.c (dump_oacc_loop_part): Likewise.
18022 (dump_oacc_loop): Likewise.
18023 * trans-mem.c (tm_log_emit): Likewise.
18024 (tm_memopt_accumulate_memops): Likewise.
18025 (dump_tm_memopt_set): Likewise.
18026 (dump_tm_memopt_transform): Likewise.
18027 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18028 (print_loop): Likewise.
18029 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18030 (chkp_gather_checks_info): Likewise.
18031 (chkp_get_check_result): Likewise.
18032 (chkp_remove_check_if_pass): Likewise.
18033 (chkp_use_outer_bounds_if_possible): Likewise.
18034 (chkp_reduce_bounds_lifetime): Likewise.
18035 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18036 (chkp_mark_completed_bounds): Likewise.
18037 (chkp_register_incomplete_bounds): Likewise.
18038 (chkp_mark_invalid_bounds): Likewise.
18039 (chkp_maybe_copy_and_register_bounds): Likewise.
18040 (chkp_build_returned_bound): Likewise.
18041 (chkp_get_bound_for_parm): Likewise.
18042 (chkp_build_bndldx): Likewise.
18043 (chkp_get_bounds_by_definition): Likewise.
18044 (chkp_generate_extern_var_bounds): Likewise.
18045 (chkp_get_bounds_for_decl_addr): Likewise.
18046 * tree-chrec.c (chrec_apply): Likewise.
18047 * tree-data-ref.c (dump_data_reference): Likewise.
18048 (dump_subscript): Likewise.
18049 (dump_data_dependence_relation): Likewise.
18050 (analyze_overlapping_iterations): Likewise.
18051 * tree-inline.c (expand_call_inline): Likewise.
18052 (tree_function_versioning): Likewise.
18053 * tree-into-ssa.c (dump_defs_stack): Likewise.
18054 (dump_currdefs): Likewise.
18055 (dump_names_replaced_by): Likewise.
18056 (dump_update_ssa): Likewise.
18057 (update_ssa): Likewise.
18058 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18059 * tree-parloops.c (build_new_reduction): Likewise.
18060 (try_create_reduction_list): Likewise.
18061 (ref_conflicts_with_region): Likewise.
18062 (oacc_entry_exit_ok_1): Likewise.
18063 (oacc_entry_exit_single_gang): Likewise.
18064 * tree-pretty-print.h: Likewise.
18065 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18066 (get_scalar_evolution): Likewise.
18067 (add_to_evolution): Likewise.
18068 (get_loop_exit_condition): Likewise.
18069 (analyze_evolution_in_loop): Likewise.
18070 (analyze_initial_condition): Likewise.
18071 (analyze_scalar_evolution): Likewise.
18072 (instantiate_scev): Likewise.
18073 (number_of_latch_executions): Likewise.
18074 (gather_chrec_stats): Likewise.
18075 (final_value_replacement_loop): Likewise.
18076 (scev_const_prop): Likewise.
18077 * tree-sra.c (dump_access): Likewise.
18078 (disqualify_candidate): Likewise.
18079 (create_access): Likewise.
18080 (reject): Likewise.
18081 (maybe_add_sra_candidate): Likewise.
18082 (create_access_replacement): Likewise.
18083 (analyze_access_subtree): Likewise.
18084 (analyze_all_variable_accesses): Likewise.
18085 (sra_modify_assign): Likewise.
18086 (initialize_constant_pool_replacements): Likewise.
18087 (find_param_candidates): Likewise.
18088 (decide_one_param_reduction): Likewise.
18089 (replace_removed_params_ssa_names): Likewise.
18090 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18091 * tree-ssa-copy.c (dump_copy_of): Likewise.
18092 (copy_prop_visit_cond_stmt): Likewise.
18093 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18094 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18095 (record_equivalences_from_stmt): Likewise.
18096 * tree-ssa-dse.c (compute_trims): Likewise.
18097 (delete_dead_call): Likewise.
18098 (delete_dead_assignment): Likewise.
18099 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18100 (forward_propagate_into_cond): Likewise.
18101 (pass_forwprop::execute): Likewise.
18102 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18103 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18105 (move_computations_worker): Likewise.
18106 (execute_sm): Likewise.
18107 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18108 (remove_exits_and_undefined_stmts): Likewise.
18109 (remove_redundant_iv_tests): Likewise.
18110 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18111 (adjust_iv_update_pos): Likewise.
18112 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18113 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18114 (value_replacement): Likewise.
18115 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18116 * tree-ssa-pre.c (print_pre_expr): Likewise.
18117 (get_representative_for): Likewise.
18118 (create_expression_by_pieces): Likewise.
18119 (insert_into_preds_of_block): Likewise.
18120 (eliminate_insert): Likewise.
18121 (eliminate_dom_walker::before_dom_children): Likewise.
18122 (eliminate): Likewise.
18123 (remove_dead_inserted_code): Likewise.
18124 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18125 * tree-ssa-reassoc.c (get_rank): Likewise.
18126 (eliminate_duplicate_pair): Likewise.
18127 (eliminate_plus_minus_pair): Likewise.
18128 (eliminate_not_pairs): Likewise.
18129 (undistribute_ops_list): Likewise.
18130 (eliminate_redundant_comparison): Likewise.
18131 (update_range_test): Likewise.
18132 (optimize_range_tests_var_bound): Likewise.
18133 (optimize_vec_cond_expr): Likewise.
18134 (rewrite_expr_tree): Likewise.
18135 (rewrite_expr_tree_parallel): Likewise.
18136 (linearize_expr): Likewise.
18137 (break_up_subtract): Likewise.
18138 (linearize_expr_tree): Likewise.
18139 (attempt_builtin_powi): Likewise.
18140 (attempt_builtin_copysign): Likewise.
18141 (transform_stmt_to_copy): Likewise.
18142 (transform_stmt_to_multiply): Likewise.
18143 (dump_ops_vector): Likewise.
18144 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18145 (print_scc): Likewise.
18146 (set_ssa_val_to): Likewise.
18147 (visit_reference_op_store): Likewise.
18148 (visit_use): Likewise.
18149 (sccvn_dom_walker::before_dom_children): Likewise.
18150 (run_scc_vn): Likewise.
18151 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18153 (expr_hash_elt::print): Likewise.
18154 (const_and_copies::pop_to_marker): Likewise.
18155 (const_and_copies::record_const_or_copy_raw): Likewise.
18156 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18157 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18158 (dump_predicates): Likewise.
18159 (find_uninit_use): Likewise.
18160 (warn_uninitialized_phi): Likewise.
18161 (pass_late_warn_uninitialized::execute): Likewise.
18162 * tree-ssa.c (verify_vssa): Likewise.
18163 (verify_ssa): Likewise.
18164 (maybe_optimize_var): Likewise.
18165 * tree-vrp.c (dump_value_range): Likewise.
18166 (dump_all_value_ranges): Likewise.
18167 (dump_asserts_for): Likewise.
18168 (register_edge_assert_for_2): Likewise.
18169 (vrp_visit_cond_stmt): Likewise.
18170 (vrp_visit_switch_stmt): Likewise.
18171 (vrp_visit_stmt): Likewise.
18172 (vrp_visit_phi_node): Likewise.
18173 (simplify_cond_using_ranges_1): Likewise.
18174 (fold_predicate_in): Likewise.
18175 (evrp_dom_walker::before_dom_children): Likewise.
18176 (evrp_dom_walker::push_value_range): Likewise.
18177 (evrp_dom_walker::pop_value_range): Likewise.
18178 (execute_early_vrp): Likewise.
18180 2017-05-16 Richard Biener <rguenther@suse.de>
18182 * dwarf2out.c (loc_list_from_tree_1): Do not create
18183 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18185 2017-05-16 Richard Biener <rguenther@suse.de>
18187 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18189 (note_variable_value_in_expr): If we resolved the decl ref
18190 do not push to the stack.
18192 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18194 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18195 operations in fast-math mode.
18196 (vaddq_f16): Likewise.
18197 (vmul_f16): Likewise.
18198 (vmulq_f16): Likewise.
18199 (vsub_f16): Likewise.
18200 (vsubq_f16): Likewise.
18201 * config/arm/neon.md (add<mode>3): New.
18203 (fma:<VH:mode>3): New. Also remove outdated comment.
18206 2017-05-16 Martin Liska <mliska@suse.cz>
18210 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18211 (odr_types_equivalent_p): Likewise.
18213 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18215 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18217 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18220 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18221 non-interunit SSE move alternatives with '?'.
18222 (zero-extendsidi peephole2): New peephole to skip intermediate
18223 general register in SSE zero-extend sequence.
18225 2017-05-15 Jeff Law <law@redhat.com>
18227 * reorg.c (relax_delay_slots): Create a new variable to hold
18228 the temporary target rather than clobbering TARGET_LABEL.
18230 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18231 missing argument to extract_bit_field call.
18232 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18234 2017-05-15 Martin Liska <mliska@suse.cz>
18237 * gcc.c (process_command): Do not allow empty argument of -o option.
18239 2017-05-15 Renlin Li <renlin.li@arm.com>
18241 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18242 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18243 * config/aarch64/constraints.md (Usf): Add long call check.
18244 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18245 (call_value): Likewise.
18246 (sibcall): Likewise.
18247 (sibcall_value): Likewise.
18249 (call_value_insn): New.
18250 (sibcall_insn): Update rtx pattern.
18251 (sibcall_value_insn): Likewise.
18252 (call_internal): Remove.
18253 (call_value_internal): Likewise.
18254 (sibcall_internal): Likewise.
18255 (sibcall_value_internal): Likewise.
18256 (call_reg): Likewise.
18257 (call_symbol): Likewise.
18258 (call_value_reg): Likewise.
18259 (call_value_symbol): Likewise.
18261 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18264 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18266 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18268 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18269 compatible with CCGOCmode and with CCZmode.
18271 2017-05-14 Martin Sebor <msebor@redhat.com>
18273 PR middle-end/77671
18274 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18275 (gimple_fold_builtin_snprintf): Same.
18276 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18277 (gimple_fold_builtin_snprintf): Same.
18278 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18279 of character types.
18280 (is_call_safe): New function.
18281 (try_substitute_return_value): Call it.
18282 (try_simplify_call): New function.
18283 (pass_sprintf_length::handle_gimple_call): Call it.
18285 2017-05-14 Martin Sebor <msebor@redhat.com>
18287 PR middle-end/80669
18288 * builtins.c (expand_builtin_stpncpy): Simplify.
18290 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18292 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18293 * config/i386/i386.h
18294 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18295 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18296 (struct machine_function): Add new members call_ms2sysv,
18297 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18298 (struct machine_frame_state): New fields sp_realigned and
18299 sp_realigned_offset.
18300 * config/i386/i386.c
18301 (enum xlogue_stub): New enum.
18302 (enum xlogue_stub_sets): New enum.
18303 (class xlogue_layout): New class.
18304 (struct ix86_frame): New fields stack_realign_allocate_offset,
18305 stack_realign_offset and outlined_save_offset. Modify comments to
18306 detail stack layout when using out-of-line stubs.
18307 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18308 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18309 -mcall-ms2sysv-xlogues.
18310 (stub_managed_regs): New static variable.
18311 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18312 registers managed by out-of-line stub.
18313 (disable_call_ms2sysv_xlogues): New function.
18314 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18315 m->call_ms2sysv when appropriate and compute frame layout for
18317 (sp_valid_at, fp_valid_at): New inline functions.
18318 (choose_basereg): New function.
18319 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18321 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18322 Use align parameter of choose_baseaddr to generated aligned SSE movs
18324 (pro_epilogue_adjust_stack): Modify to track
18325 machine_frame_state::sp_realigned.
18326 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18327 (ix86_nsaved_sseregs): Likewise.
18328 (ix86_emit_save_regs): Likewise.
18329 (ix86_emit_save_regs_using_mov): Likewise.
18330 (ix86_emit_save_sse_regs_using_mov): Likewise.
18331 (get_scratch_register_on_entry): Likewise.
18332 (gen_frame_set): New function.
18333 (gen_frame_load): Likewise.
18334 (gen_frame_store): Likewise.
18335 (emit_outlined_ms2sysv_save): Likewise.
18336 (emit_outlined_ms2sysv_restore): Likewise.
18337 (ix86_expand_prologue): Modify stack re-alignment code and call
18338 emit_outlined_ms2sysv_save when appropriate.
18339 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18340 parameter rtx_insn *insn, which allows the function to be used to only
18341 generate the notes.
18342 (ix86_expand_epilogue): Modify validity checks of frame and stack
18343 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18344 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18345 * config/i386/predicates.md
18346 (save_multiple): New predicate.
18347 (restore_multiple): Likewise.
18348 * config/i386/sse.md
18349 (save_multiple<mode>): New pattern.
18350 (save_multiple_realign<mode>): Likewise.
18351 (restore_multiple<mode>): Likewise.
18352 (restore_multiple_and_return<mode>): Likewise.
18353 (restore_multiple_leave_return<mode>): Likewise.
18354 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18356 2017-05-14 Julia Koval <julia.koval@intel.com>
18358 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18359 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18360 (__builtin_ia32_xsetbv): New builtins.
18361 * config/i386/i386.c (ix86_expand_special_args_builtin):
18363 (ix86_expand_builtin): Special expand for new intrinsics.
18364 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18365 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18366 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18368 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18370 * cfganal.c (inverted_post_order_compute): Change argument type
18372 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18373 * df-core.c (rest_of_handle_df_initialize): Adjust.
18374 (rest_of_handle_df_finish): Likewise.
18375 (df_analyze_1): Likewise.
18376 (df_analyze): Likewise.
18377 (loop_inverted_post_order_compute): Change argument to be a vec *.
18378 (df_analyze_loop): Adjust.
18379 (df_get_n_blocks): Likewise.
18380 (df_get_postorder): Likewise.
18381 * df.h (struct df_d): Change field to be a vec.
18382 * lcm.c (compute_laterin): Adjust.
18383 (compute_available): Likewise.
18384 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18385 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18386 * tree-ssa-pre.c (compute_antic): Likewise.
18388 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18390 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18391 (depth_first_search::depth_first_search): Change structure init
18392 function to this constructor.
18393 (depth_first_search::add_bb): Rename function to this member.
18394 (depth_first_search::execute): Likewise.
18395 (flow_dfs_compute_reverse_finish): Adjust.
18397 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18399 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18400 (longest_simple_path): Likewise.
18401 * shrink-wrap.c (spread_components): Likewise.
18402 (disqualify_problematic_components): Likewise.
18403 (emit_common_heads_for_components): Likewise.
18404 (emit_common_tails_for_components): Likewise.
18405 (insert_prologue_epilogue_for_components): Likewise.
18407 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18409 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18412 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18414 * df-core.c (df_set_blocks): Start using auto_bitmap.
18415 (df_compact_blocks): Likewise.
18416 * df-problems.c (df_rd_confluence_n): Likewise.
18417 * df-scan.c (df_insn_rescan_all): Likewise.
18418 (df_process_deferred_rescans): Likewise.
18419 (df_update_entry_block_defs): Likewise.
18420 (df_update_exit_block_uses): Likewise.
18421 (df_entry_block_bitmap_verify): Likewise.
18422 (df_exit_block_bitmap_verify): Likewise.
18423 (df_scan_verify): Likewise.
18424 * lra-constraints.c (lra_constraints): Likewise.
18425 (undo_optional_reloads): Likewise.
18426 (lra_undo_inheritance): Likewise.
18427 * lra-remat.c (calculate_gen_cands): Likewise.
18428 (do_remat): Likewise.
18429 * lra-spills.c (assign_spill_hard_regs): Likewise.
18430 (spill_pseudos): Likewise.
18431 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18432 (bitmap_set_subtract_values): Likewise.
18434 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18436 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18437 management with auto_bitmap.
18438 (fix_inter_tick): Likewise.
18439 (fix_recovery_deps): Likewise.
18440 * ira.c (add_store_equivs): Likewise.
18441 (find_moveable_pseudos): Likewise.
18442 (split_live_ranges_for_shrink_wrap): Likewise.
18443 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18444 (rtx_reuse_manager::seen_def_p): Likewise.
18445 (rtx_reuse_manager::set_seen_def): Likewise.
18446 * print-rtl.h (class rtx_reuse_manager): Likewise.
18448 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18450 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18452 (migrate_btr_def): Likewise.
18453 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18454 * df-core.c (loop_post_order_compute): Likewise.
18455 (loop_inverted_post_order_compute): Likewise.
18456 * hsa-common.h: Likewise.
18457 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18458 * init-regs.c (initialize_uninitialized_regs): Likewise.
18459 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18460 (inline_small_functions): Likewise.
18461 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18462 * ira.c (combine_and_move_insns): Likewise.
18463 (build_insn_chain): Likewise.
18464 * loop-invariant.c (find_invariants): Likewise.
18465 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18466 * predict.c (tree_predict_by_opcode): Likewise.
18467 (predict_paths_leading_to): Likewise.
18468 (predict_paths_leading_to_edge): Likewise.
18469 (estimate_loops_at_level): Likewise.
18470 (estimate_loops): Likewise.
18471 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18472 (spread_components): Likewise.
18473 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18474 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18475 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18476 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18477 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18478 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18479 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18480 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18481 (mark_threaded_blocks): Likewise.
18482 (thread_through_all_blocks): Likewise.
18483 * tree-ssa.c (verify_ssa): Likewise.
18484 (execute_update_addresses_taken): Likewise.
18485 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18487 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18489 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18491 (post_order_compute): Likewise.
18492 (inverted_post_order_compute): Likewise.
18493 (pre_and_rev_post_order_compute_fn): Likewise.
18495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18497 * genrecog.c (int_set::int_set): Explicitly construct our
18498 auto_vec base class.
18499 * vec.h (auto_vec::auto_vec): New constructor.
18501 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18503 * bitmap.h (class auto_bitmap): New constructor taking
18504 bitmap_obstack * argument.
18506 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18508 * bitmap.h (class auto_bitmap): Change type of m_bits to
18509 bitmap_head, and adjust ctor / dtor and member operators.
18511 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18513 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18514 when returned register mode doesn't match original mode.
18516 2017-05-12 Jeff Law <law@redhat.com>
18517 Jakub Jelinek <jakub@redhat.com>
18519 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18520 we look for cc setter after the compare-elim changes.
18521 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18522 within the vector to match what compare-elim now expects.
18523 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18524 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18526 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18527 after the compare-elim changes.
18528 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18529 the vector to match what compare-elim now expects.
18530 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18531 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18532 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18533 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18534 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18536 * config/visium/visium.c (single_set_and_flags): Fix where
18537 we look for cc setter after the compare-elim changes.
18538 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18539 with the vector to match what compare-elim now expects.
18540 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18541 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18542 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18543 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18544 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18545 (neg<mode>2_insn_set_overflow): Likewise.
18547 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18549 PR middle-end/79794
18550 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18551 maybe_expand_insn call, set ops[0].target. If still set after call,
18552 set alt_rtl. Add extra arg to recursive calls.
18553 (extract_bit_field): Add alt_rtl argument. Pass to
18555 * expmed.h (extract_bit_field): Fix prototype.
18556 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18557 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18558 to extract_bit_field_calls.
18559 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18560 Pass alt_rtl to extract_bit_field calls.
18561 * calls.c (store_unaligned_arguments_into_psuedos)
18562 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18563 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18565 * optabs.h (struct expand_operand): Add target bitfield.
18567 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18569 * compare-elim.c (try_eliminate_compare): Canonicalize
18570 operation with embedded compare to
18571 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18572 (set (reg) (operation)].
18574 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18576 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18579 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18580 cost of adding a carry flag for ADC instruction.
18581 [case MINUS]: Ignore the cost of subtracting a carry flag
18582 for SBB instruction.
18584 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18586 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18588 * config/rs6000/bmiintrin.h: New file.
18589 * config/rs6000/bmi2intrin.h: New file.
18590 * config/rs6000/x86intrin.h: New file.
18592 2017-05-12 Jeff Law <law@redhat.com>
18594 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18597 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18599 PR middle-end/80707
18600 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18602 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18604 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18605 early expansion of vector divide builtins.
18606 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18607 builtins identified as having unsigned arguments.
18609 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18611 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18612 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18613 expansion of vector logical operations (and, andc, or, xor,
18616 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18618 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18619 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18621 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18624 early GIMPLE expansion of vector multiplies.
18626 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18628 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18629 TARGET_HAVE_MOVT conditional.
18630 (movt splitter): Likewise.
18632 2017-05-12 Richard Biener <rguenther@suse.de>
18634 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18635 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18636 Fold all stmts not inplace.
18638 2017-05-12 Richard Biener <rguenther@suse.de>
18640 PR tree-optimization/80713
18641 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18642 inserted_exprs bit for not removed stmts.
18644 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18646 PR middle-end/69921
18647 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18648 parallelized" attribute for parallelized OpenACC kernels.
18649 * omp-offload.c (execute_oacc_device_lower): Use it.
18651 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18652 Set "oacc kernels" attribute.
18653 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18654 parameter. Adjust all users.
18655 (oacc_fn_attrib_kernels_p): Remove function.
18656 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18657 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18658 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18659 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18660 assert "oacc kernels" attribute is set.
18662 2017-05-11 Carl Love <cel@us.ibm.com>
18664 * config/rs6000/rs6000-c: Add support for built-in functions
18665 vector unsigned char vec_popcnt (vector signed char)
18666 vector unsigned char vec_popcnt (vector unsigned char)
18667 vector unsigned short vec_popcnt (vector signed short)
18668 vector unsigned short vec_popcnt (vector unsigned short)
18669 vector unsigned int vec_popcnt (vector signed int)
18670 vector unsigned int vec_popcnt (vector unsigned int)
18671 vector unsigned long long vec_popcnt (vector signed long long)
18672 vector unsigned long long vec_popcnt (vector unsigned long long)
18673 vector signed long long vec_slo (vector signed long long,
18674 vector signed char)
18675 vector signed long long vec_slo (vector signed long long,
18676 vector unsigned char)
18677 vector unsigned long long vec_slo (vector unsigned long long,
18678 vector signed char)
18679 vector unsigned long long vec_slo (vector unsigned long long,
18680 vector unsigned char)
18681 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18682 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18683 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18684 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18685 * doc/extend.texi: Update the built-in documentation file for the
18686 new built-in functions.
18688 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18690 * attribs.h (sorted_attr_string): Move machine independent
18691 functions for target clone support from the i386 port to common
18692 code. Rename ix86_function_versions to common_function_versions.
18693 Rename make_name to make_unique_name.
18694 (common_function_versions): Likewise.
18695 (make_unique_name): Likewise.
18696 (make_dispatcher_decl): Likewise.
18697 (is_function_default_version): Likewise.
18698 * attribs.c (attr_strcmp): Likewise.
18699 (sorted_attr_string): Likewise.
18700 (common_function_versions): Likewise.
18701 (make_unique_name): Likewise.
18702 (make_dispatcher_decl): Likewise.
18703 (is_function_default_version): Likewise.
18704 * config/i386/i386.c (attr_strcmp): Likewise.
18705 (sorted_attr_string): Likewise.
18706 (ix86_function_versions): Likewise.
18707 (make_name): Likewise.
18708 (make_dispatcher_decl): Likewise.
18709 (is_function_default_version): Likewise.
18710 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18712 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18715 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18716 Account for direct move costs for vec_construct of integer
18719 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18722 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18723 (UNSPEC_STX_ATOMIC): Ditto.
18724 (loaddi_via_sse): New insn.
18725 (storedi_via_sse): Ditto.
18726 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18727 Update corresponding peephole2 patterns.
18728 (atomic_storedi_fpu): Ditto.
18730 2017-05-11 Julia Koval <julia.koval@intel.com>
18732 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18733 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18735 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18736 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18737 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18739 2017-05-11 Nathan Sidwell <nathan@acm.org>
18741 * graphite-poly.c: Include dumpfile.h.
18743 * dumpfle.h (dump_function): Declare here ...
18744 * tree-dump.h (dump_function): ... not here.
18745 * dumpfile.c: #include tree-cfg.h.
18746 (dump_function): Move here from ...
18747 * tree-dump.c (dump_function): ... here.
18748 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18749 * graphite-poly.c: Don't include tree-dump.h.
18750 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18751 * print-tree.c: Likewise.
18752 * stor-layout.c: Likewise.
18753 * tree-nested.c: Likewise.
18755 * dumpfile.c (dump_start): Use TDF_FLAGS.
18756 (dump_enable_all): Fix TDF_KIND check thinko.
18758 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18761 array entries to represent two legal parameterizations of the
18762 overloaded __builtin_cmpb function, as represented by the
18763 P6_OV_BUILTIN_CMPB constant.
18764 (altivec_resolve_overloaded_builtin): Add special case handling
18765 for the __builtin_cmpb function, as represented by the
18766 P6_OV_BUILTIN_CMPB constant.
18767 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18768 (BU_P6_64BIT_2): New macro.
18769 (BU_P6_OVERLOAD_2): New macro
18770 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18771 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18772 (CMPB): Add overload support to represent both 32-bit and 64-bit
18773 compare-bytes function.
18774 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18775 support for TARGET_CMPB.
18776 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18777 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18778 documentation of the __builtin_cmpb overloaded built-in function.
18780 2017-05-11 Richard Biener <rguenther@suse.de>
18782 PR tree-optimization/80705
18783 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18784 bases are not vectorizable.
18786 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18788 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18789 when counting register pressure.
18791 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18793 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18794 (iv_ca_more_deps): Renamed to ...
18795 (iv_ca_compare_deps): ... this.
18796 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18798 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18800 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18802 (determine_group_iv_costs): ... here.
18803 (find_inv_vars_cb): Record inv var if it's not recorded before.
18805 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18807 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18808 (get_shiftadd_cost): Ditto.
18810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18812 * tree-ssa-address.c: Include header file.
18813 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18815 (add_to_parts): Refactor.
18816 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18817 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18820 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18822 PR tree-optimization/53090
18823 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18825 (extract_cond_operands): Detect condition with IV on both sides
18826 and return COMP_IV_EXPR_2.
18827 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18828 (rewrite_use_compare): Simplify by removing call to function
18829 extract_cond_operands.
18831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18833 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18834 (extract_cond_operands): Detect condition comparing against non-
18835 invariant bound and return appropriate enum value.
18836 (find_interesting_uses_cond): Update use of extract_cond_operands.
18837 Handle its return value accordingly.
18838 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18842 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18843 nonlinear iv_use computation in loop invariant sensitive way.
18845 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18847 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18848 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18850 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18852 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18853 (dump_cand): Support iv_cand.inv_exprs.
18854 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18856 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18859 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18861 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18863 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18864 as local function. Include necessary header files.
18865 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18869 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18873 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18874 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18875 RSHIFT_EXPR and BIT_NOT_EXPR.
18877 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18879 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18880 (adjust_setup_cost): New parameter supporting round up adjustment.
18881 (struct address_cost_data): Delete.
18882 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18883 (split_address_cost, ptr_difference_cost): Delete.
18884 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18885 (struct ainc_cost_data): New struct.
18886 (get_address_cost_ainc): New function.
18887 (get_address_cost, get_computation_cost): Reimplement.
18888 (determine_group_iv_cost_address): Record inv_expr for all uses of
18890 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18891 (iv_ca_has_deps): Reimplemented to ...
18892 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18894 (iv_ca_extend): Call iv_ca_more_deps.
18896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18898 * tree-ssa-address.c (struct mem_address): Move to header file.
18899 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18900 * tree-ssa-address.h (struct mem_address): Move from C file.
18901 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18905 * tree-affine.h (aff_combination_type): New interface.
18906 (aff_combination_zero_p): Remove static.
18907 (aff_combination_const_p): New interface.
18908 (aff_combination_singleton_var_p): New interfaces.
18910 2017-05-11 Richard Biener <rguenther@suse.de>
18912 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18913 Skip unreachable blocks and destinations.
18914 (eliminate): Move stmt removal and fixup ...
18915 (fini_eliminate): ... here. Skip inserted exprs.
18916 (pass_pre::execute): Move fini_pre after fini_eliminate.
18917 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18918 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18919 PRE to get rid of dead code that has invalid SSA form and
18920 split critical edges again.
18922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18924 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18926 2017-05-11 Richard Biener <rguenther@suse.de>
18928 * passes.c (execute_function_todo): Verify loops if they are
18929 said to be up-to-date.
18930 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18931 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18933 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18936 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18937 handle calling assemble_external ourself.
18940 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18941 modes with zero size. Enhance comment.
18943 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18945 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18946 built-ins for vec_xl and vec_xst with short and char pointer
18949 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18951 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18952 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18953 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18954 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18955 (_mm_maskz_min_round_ss): New intrinsics.
18956 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18957 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18958 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18959 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18960 (__builtin_ia32_minss_mask_round): New builtins.
18961 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18963 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18965 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18966 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18968 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18971 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18973 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18974 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18975 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18976 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18977 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18978 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18979 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18980 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18981 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18982 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18983 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18984 (__builtin_ia32_mulss_mask_round): New builtins.
18985 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18986 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18987 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18989 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18990 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18992 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18995 2017-05-10 Julia Koval <julia.koval@intel.com>
18997 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18998 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18999 (_mm256_setr_m128i): New intrinsics.
19001 2017-05-10 Julia Koval <julia.koval@intel.com>
19003 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19004 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19005 (_mm_maskz_rcp14_ss): New intrinsics.
19006 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19007 (__builtin_ia32_rcp14ss_mask): New builtins.
19008 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19010 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19012 PR tree-optimization/51513
19013 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19014 (assert_unreachable_fallthru_edge_p): Use it.
19015 (group_case_labels_stmt): Likewise.
19016 * tree-cfg.h: Prototype it.
19017 * stmt.c: Include cfghooks.h and tree-cfg.h.
19018 (emit_case_dispatch_table) <gap_label>: New local variable.
19019 Use it to fill dispatch table gaps.
19020 Test for default_label before updating probabilities.
19021 (expand_case) <default_label>: Remove unneeded initialization.
19022 Test for unreachable default case statement and remove its edge.
19023 Set default_label accordingly.
19024 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19026 2017-05-10 Carl Love <cel@us.ibm.com>
19028 * config/rs6000/rs6000-c: Add support for built-in functions
19029 vector signed char vec_neg (vector signed char)
19030 vector signed short int vec_neg (vector short int)
19031 vector signed int vec_neg (vector signed int)
19032 vector signed long long vec_neg (vector signed long long)
19033 vector float vec_neg (vector float)
19034 vector double vec_neg (vector double)
19035 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19037 * config/rs6000/altivec.h: Add define for vec_neg
19038 * doc/extend.texi: Update the built-in documentation for the
19039 new built-in functions.
19041 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19043 PR tree-optimization/77644
19044 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19046 2017-05-10 Nathan Sidwell <nathan@acm.org>
19048 * dumpfile.h (TDI_lang_all): New.
19049 (TDF_KIND): New. Renumber others
19050 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19052 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19054 (get_dump_file_name): Adjust suffix generation.
19055 (dump_enable_all): Use TDF_KIND.
19056 * doc/invoke.texi (-fdump-lang-all): Document.
19058 * dumpfile.h: Tabify.
19060 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19063 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19064 Move member access before delete.
19066 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19068 * tree-inline.c (expand_call_inline): Split block at stmt
19071 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19074 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19075 are now unused after splitting mov{sf,sd}_hardfloat.
19076 (f32_lr2): Likewise.
19077 (f32_lm): Likewise.
19078 (f32_lm2): Likewise.
19079 (f32_li): Likewise.
19080 (f32_li2): Likewise.
19081 (f32_lv): Likewise.
19082 (f32_sr): Likewise.
19083 (f32_sr2): Likewise.
19084 (f32_sm): Likewise.
19085 (f32_sm2): Likewise.
19086 (f32_si): Likewise.
19087 (f32_si2): Likewise.
19088 (f32_sv): Likewise.
19089 (f32_dm): Likewise.
19090 (f32_vsx): Likewise.
19091 (f32_av): Likewise.
19092 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19093 For movsf, order stores so the VSX stores occur before the GPR
19094 store which encourages the register allocator to use a traditional
19095 FPR instead of a GPR. For movsd, order the stores so that the GPR
19096 store comes before the VSX stores to allow the power6 to work.
19097 This is due to the power6 not having a 32-bit integer store
19098 instruction from a FPR.
19099 (movsf_hardfloat): Likewise.
19100 (movsd_hardfloat): Likewise.
19102 2017-05-09 Martin Sebor <msebor@redhat.com>
19104 PR translation/80280
19105 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19108 PR translation/80280
19109 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19110 data member added in r247778.
19111 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19113 2017-05-09 Nathan Sidwell <nathan@acm.org>
19115 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19117 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19120 2017-05-09 Marek Polacek <polacek@redhat.com>
19122 * doc/invoke.texi: Fix typo.
19124 2017-05-09 Richard Biener <rguenther@suse.de>
19126 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19127 (vrp_val_is_min): Likewise.
19128 (set_value_range_to_value): Likewise.
19129 (set_value_range_to_nonnegative): Likewise.
19130 (gimple_assign_nonzero_p): Likewise.
19131 (gimple_stmt_nonzero_p): Likewise.
19132 (vrp_int_const_binop): Likewise. Remove unreachable case.
19133 (adjust_range_with_scev): Adjust comments.
19134 (compare_range_with_value): Likewise.
19135 (extract_range_from_phi_node): Likewise.
19136 (test_for_singularity): Likewise.
19138 2017-05-09 Richard Biener <rguenther@suse.de>
19140 * tree-vrp.c (get_single_symbol): Add assert that we don't
19141 get overflowed constants as invariant part.
19142 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19143 checks. Use wi::cmp instead of recursing for integer constants.
19144 (compare_values): Just ignore whether we assumed undefined
19145 overflow instead of failing the compare.
19146 (extract_range_for_var_from_comparison_expr): Add comment before the
19147 TREE_NO_WARNING sets.
19148 (test_for_singularity): Likewise.
19149 (extract_range_from_comparison): Do not disable optimization
19150 when we assumed undefined overflow.
19151 (extract_range_basic): Remove init of unused var.
19153 2017-05-09 Richard Biener <rguenther@suse.de>
19155 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19156 (extract_range_from_multiplicative_op_1): Adjust.
19157 (extract_range_from_binary_expr_1): Use int_const_binop.
19159 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19162 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19163 rs6000_store_data_bypass_p in seven define_bypass directives and
19164 in several comments.
19165 * config/rs6000/rs6000-protos.h: Add prototype for
19166 rs6000_store_data_bypass_p function.
19167 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19168 function implements slightly different (rs6000-specific) semantics
19169 than store_data_bypass_p, returning false rather than aborting
19170 with assertion error when arguments do not satisfy the
19171 requirements of store data bypass.
19172 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19173 rs6000_store_data_bypass_p.
19175 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19177 * config/xtensa/xtensa-protos.h
19178 (xtensa_initial_elimination_offset): New declaration.
19179 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19180 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19181 macro definition, add case for FRAME_POINTER_REGNUM when
19182 FRAME_GROWS_DOWNWARD.
19183 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19184 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19185 xtensa_initial_elimination_offset.
19187 2017-05-08 Nathan Sidwell <nathan@acm.org>
19189 * doc/invoke.texi: Alphabetize -fdump options.
19191 2017-05-08 Martin Sebor <msebor@redhat.com>
19193 PR translation/80280
19194 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19196 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19198 * target.def (compute_frame_layout): New optional target hook.
19199 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19200 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19201 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19203 * reload1.c (verify_initial_elim_offsets): Likewise.
19204 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19205 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19206 (arm_get_frame_offsets): Split up into this ...
19207 (arm_compute_frame_layout): ... and this function.
19209 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19211 * config/aarch64/constraints.md (Usa): New constraint.
19212 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19214 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19216 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19217 with_multilib_list after it has been checked.
19219 2017-05-08 Richard Biener <rguenther@suse.de>
19221 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19222 (bitmap_set_subtract_values): Likewise.
19224 2017-05-08 Richard Biener <rguenther@suse.de>
19226 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19227 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19229 (gimple_stmt_nonzero_warnv_p): Rename to ...
19230 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19232 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19233 (extract_range_basic): Adjust, do not disable propagation on
19234 strict overflow sensitive simplification.
19235 (vrp_visit_cond_stmt): Likewise.
19237 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19239 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19240 body size unconditionally.
19242 2017-05-07 Jeff Law <law@redhat.com>
19245 2017-05-06 Jeff Law <law@redhat.com>
19246 PR tree-optimization/78496
19247 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19250 PR tree-optimization/78496
19251 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19252 (simplify_stmt_using_ranges): Call it.
19253 (vrp_dom_walker::before_dom_children): Extract equivalences
19254 from an ASSERT_EXPR with an equality comparison against a
19257 2017-05-06 Jeff Law <law@redhat.com>
19259 PR tree-optimization/78496
19260 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19263 PR tree-optimization/78496
19264 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19265 (simplify_stmt_using_ranges): Call it.
19266 (vrp_dom_walker::before_dom_children): Extract equivalences
19267 from an ASSERT_EXPR with an equality comparison against a
19270 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19272 * lra-constraints.c (lra_copy_reg_equiv): New function.
19273 (split_reg): Use it to copy equivalence information from the
19274 original register to the spill register.
19276 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19278 PR rtl-optimization/75964
19279 * simplify-rtx.c (simplify_const_relational_operation): Remove
19280 invalid handling of comparisons of integer ABS.
19282 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19284 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19285 initialize to zero.
19286 (init_regs): Remove declaration.
19287 (function_arg_advance_32): Initialize error_p as boolean variable.
19289 2017-05-05 Nathan Sidwell <nathan@acm.org>
19291 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19292 lines. Use for (;;).
19294 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19296 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19297 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19298 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19299 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19300 VF=2 that require versioning.
19302 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19304 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19307 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19309 * diagnostic.h (diagnostic_override_option_index): Convert from
19310 macro to inline function.
19312 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19314 * diagnostic.c (last_module_changed_p): New function.
19315 (set_last_module): New function.
19316 (diagnostic_report_current_module): Convert macro usage to
19317 the above functions.
19318 * diagnostic.h (diagnostic_context::last_module): Strengthen
19319 from const line_map * to const line_map_ordinary *.
19320 (diagnostic_last_module_changed): Delete macro.
19321 (diagnostic_set_last_module): Delete macro.
19323 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19325 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19326 with diagnostic_report_diagnostic.
19327 (diagnostic_n_impl_richloc): Likewise.
19328 * diagnostic.h (report_diagnostic): Delete macro.
19329 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19330 with diagnostic_report_diagnostic.
19331 * substring-locations.c (format_warning_va): Likewise.
19333 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19335 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19336 save/restor of format_spec. Move option-printing code to...
19337 (print_option_information): ...this new function, and
19338 reimplement by simply printing to the pretty_printer,
19339 rather than appending to the format string.
19341 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19343 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19344 handling logic into...
19345 (update_effective_level_from_pragmas): ...this new function.
19347 2017-05-04 Andrew Waterman <andrew@sifive.com>
19349 * config/riscv/riscv.opt (mstrict-align): New option.
19350 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19351 (SLOW_UNALIGNED_ACCESS): Define.
19352 (riscv_slow_unaligned_access): Declare.
19353 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19355 (riscv_slow_unaligned_access): New variable.
19356 (rocket_tune_info): Set slow_unaligned_access to true.
19357 (optimize_size_tune_info): Set slow_unaligned_access to false.
19358 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19359 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19360 (riscv_option_override): Set riscv_slow_unaligned_access.
19361 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19363 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19365 * config/riscv/riscv.md: Unify indentation.
19367 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19372 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19374 (extendsi<mode>2): Add support for doing sign extension via
19375 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19376 don't have ISA 3.0 instructions.
19377 (extendsi<mode>2 splitter): Likewise.
19378 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19379 generate the normal insns since SImode can now go in vector
19380 registers. Disallow the special UNSPECs needed for previous
19381 machines to hide SImode being used. Add new insns
19382 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19383 (fix_trunc<mode>si2_stfiwx): Likewise.
19384 (fix_trunc<mode>si2_internal): Likewise.
19385 (fixuns_trunc<mode>si2): Likewise.
19386 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19387 (fctiw<u>z_<mode>_smallint): Likewise.
19388 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19389 of floating point to 32-bit integer from doing a direct move to
19390 the GPR registers to do a store.
19391 (fctiwz_<mode>): Break long line.
19393 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19395 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19396 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19397 (addr_list, addr_offset_valid_p): New.
19398 (split_address_groups): Check offset validity with above function.
19399 (gt-tree-ssa-loop-ivopts.h): Include header file.
19401 2017-05-05 Nathan Sidwell <nathan@acm.org>
19403 * config.gcc (arm*-*-*): Add missing 'fi'.
19405 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19407 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19408 included in -fopt-info does not matter.
19409 * doc/optinfo.texi (-fopt-info): Fix description of default
19410 behavour. Explicitly say order of options included in -fopt-info
19413 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19415 * config.gcc: Allow combinations of aprofile and rmprofile values for
19416 --with-multilib-list.
19417 * config/arm/t-multilib: New file.
19418 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19419 variables. Remove setting of ISA and floating-point ABI in
19420 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19421 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19422 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19423 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19425 * config/arm/t-rmprofile: Likewise except for the matches changes.
19426 * doc/install.texi (--with-multilib-list): Document the combination of
19427 aprofile and rmprofile values and warn about pitfalls in doing that.
19429 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19431 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19432 (movdi_aarch64): Likewise.
19434 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19436 PR tree-optimization/80632
19437 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19439 (build_arrays): Initialize it for virtual phis.
19440 (fix_phi_nodes): Use it for virtual phis.
19442 PR tree-optimization/80558
19443 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19444 [x, y] op z into [x op, y op z] for op & or | if conditions
19447 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19448 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19451 * config/arm/arm.md (use_literal_pool): Remove.
19452 (64-bit immediate split): No longer takes cost into consideration
19453 if arm_disable_literal_pool is enabled.
19454 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19455 used when arm_disable_literal_pool is enabled.
19456 (arm_max_const_double_inline_cost): Remove use of
19457 arm_disable_literal_pool.
19458 (push_minipool_fix): Add assert.
19459 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19460 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19461 (no_literal_pool_sf_immediate): New.
19463 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19465 PR tree-optimization/80613
19466 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19467 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19469 2017-05-05 Richard Biener <rguenther@suse.de>
19471 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19473 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19475 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19476 of this flag from insn conditions due to removal from r247495.
19478 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19480 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19482 (arm_early_store_addr_dep_ptr): Likewise.
19483 * config/arm/aarch-common-protos.h
19484 (arm_early_load_addr_dep_ptr): Add prototype.
19485 (arm_early_store_addr_dep_ptr): Likewise.
19486 * config/arm/cortex-a53.md: Add new bypasses.
19488 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19490 * tree.c (next_type_uid): Change type to unsigned.
19491 (type_hash_canon): Decrement back next_type_uid if
19492 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19493 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19496 2017-05-04 Martin Sebor <msebor@redhat.com>
19498 * builtins.c: Fix a trivial typo in a comment.
19500 PR middle-end/79234
19501 * builtins.c (check_sizes): Adjust to handle reading past the end.
19502 Avoid printing excessive upper bound of ranges. Use %E to print
19503 tree nodes instead of converting them to %wu.
19504 (expand_builtin_memchr): New function.
19505 (compute_dest_size): Rename...
19506 (compute_objsize): ...to this.
19507 (expand_builtin_memcpy): Adjust.
19508 (expand_builtin_mempcpy): Adjust.
19509 (expand_builtin_strcat): Adjust.
19510 (expand_builtin_strcpy): Adjust.
19511 (check_strncat_sizes): Adjust.
19512 (expand_builtin_strncat): Adjust.
19513 (expand_builtin_strncpy): Adjust and simplify.
19514 (expand_builtin_memset): Adjust.
19515 (expand_builtin_bzero): Adjust.
19516 (expand_builtin_memcmp): Adjust.
19517 (expand_builtin): Handle memcmp.
19518 (maybe_emit_chk_warning): Check strncat just once.
19520 2017-05-04 Martin Sebor <msebor@redhat.com>
19522 PR preprocessor/79214
19523 PR middle-end/79222
19524 PR middle-end/79223
19525 * builtins.c (check_sizes): Add inlining context and issue
19526 warnings even when -Wno-system-headers is set.
19527 (check_strncat_sizes): Same.
19528 (expand_builtin_strncat): Same.
19529 (expand_builtin_memmove): New function.
19530 (expand_builtin_stpncpy): Same.
19531 (expand_builtin): Handle memmove and stpncpy.
19533 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19535 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19536 which is not used any more.
19538 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19540 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19542 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19544 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19545 (cortexa53_tunings): Likewise.
19546 (cortexa57_tunings): Likewise.
19547 (cortexa72_tunings): Likewise.
19548 (cortexa73_tunings): Likewise.
19550 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19552 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19553 Set loop alignment to 8.
19555 2017-05-04 Martin Sebor <msebor@redhat.com>
19557 PR translation/80280
19558 * builtins.c (expand_builtin_object_size): Add missing quoting to
19559 %D and like directives.
19560 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19561 (hsa_type_for_tree_type): Same.
19562 (verify_function_arguments): Same.
19563 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19564 * varasm.c (get_section): Same.
19567 2017-05-04 Martin Sebor <msebor@redhat.com>
19569 PR translation/80280
19570 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19572 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19574 * config/aarch64/aarch64.c (generic_addrcost_table):
19575 Change HI/TI mode setting.
19577 2017-05-04 Martin Jambor <mjambor@suse.cz>
19579 PR tree-optimization/80622
19580 * tree-sra.c (comes_initialized_p): New function.
19581 (build_accesses_from_assign): Only set write lazily when
19582 comes_initialized_p is false.
19583 (analyze_access_subtree): Use comes_initialized_p.
19584 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19585 instead of testing for PARM_DECL.
19587 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19589 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19590 constraint on operand 0 to allow more general addressing modes.
19591 Adjust output template.
19592 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19594 * config/aarch64/aarch64-protos.h
19595 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19596 * config/aarch64/constraints.md (Dp): New address constraint.
19597 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19600 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19602 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19603 update use of estimate_ipcp_clone_size_and_time.
19604 (estimate_local_effects): Update use of
19605 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19606 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19607 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19608 Return nonspecialized time.
19610 2017-05-04 Richard Biener <rguenther@suse.de>
19612 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19613 for the last VUSE which def dominates the PHI. Directly call
19615 (get_continuation_for_phi_1): Remove.
19617 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19619 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19620 to explain the use of truncating division. Cap the number of
19621 iterations to the maximum given by nb_iterations_upper_bound,
19624 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19626 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19627 * configure: Regenerate.
19628 * config.in: Regenerate.
19629 * config/i386/driver-mingw32.c: new file.
19630 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19631 * config.host: Link driver-mingw32.o on MinGW host.
19632 * doc/install.texi: Document new --enable-mingw-wildcard configure
19635 2017-05-04 Marek Polacek <polacek@redhat.com>
19637 PR tree-optimization/80612
19638 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19640 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19641 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19643 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19644 (movt splitter): Likewise.
19645 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19646 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19647 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19648 block for Thumb-1 with MOVT.
19649 (thumb2_legitimate_address_p): Move code block ...
19650 (can_avoid_literal_pool_for_label_p): ... into this new function.
19651 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19653 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19654 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19655 "M-profile targets with the MOVT instruction".
19657 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19659 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19660 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19661 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19663 2017-05-04 Martin Liska <mliska@suse.cz>
19665 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19666 variable cond_code.
19668 2017-05-04 Richard Biener <rguenther@suse.de>
19670 * tree.c (array_at_struct_end_p): Handle arrays at struct
19671 end with flexarrays more conservatively. Refactor and treat
19672 arrays of arrays or aggregates more strict. Fix
19673 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19674 * tree.c (array_at_struct_end_p): Adjust prototype.
19675 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19676 * gimple-fold.c (get_range_strlen): Likewise.
19677 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19679 2017-05-04 Richard Biener <rguenther@suse.de>
19681 PR tree-optimization/31130
19682 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19684 (supports_overflow_infinity): Likewise.
19685 (is_negative_overflow_infinity): Likewise.
19686 (is_positive_overflow_infinity): Likewise.
19687 (is_overflow_infinity): Likewise.
19688 (stmt_overflow_infinity): Likewise.
19689 (overflow_infinity_range_p): Likewise.
19690 (usable_range_p): Remove as always returning true.
19691 (make_overflow_infinity): Remove.
19692 (negative_overflow_infinity): Likewise.
19693 (positive_overflow_infinity): Likewise.
19694 (avoid_overflow_infinity): Likewise.
19695 (set_value_range): Adjust accordingly.
19696 (set_value_range_to_nonnegative): Likewise, remove now unused
19697 overflow_infinity arg.
19698 (vrp_operand_equal_p): Adjust.
19699 (update_value_range): Likewise.
19700 (range_int_cst_singleton_p): Likewise.
19701 (operand_less_p): Likewise.
19702 (compare_values_warnv): Likewise.
19703 (extract_range_for_var_from_comparison_expr): Likewise.
19704 (vrp_int_const_binop): Likewise.
19705 (zero_nonzero_bits_from_vr): Likewise.
19706 (extract_range_from_multiplicative_op_1): Likewise.
19707 (extract_range_from_binary_expr_1): Likewise.
19708 (extract_range_from_unary_expr): Likewise.
19709 (extract_range_from_comparison): Likewise.
19710 (extract_range_basic): Likewise.
19711 (adjust_range_with_scev): Likewise.
19712 (compare_ranges): Likewise.
19713 (compare_range_with_value): Likewise.
19714 (dump_value_range): Likewise.
19715 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19717 (simplify_cond_using_ranges): Adjust.
19719 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19721 * brig-builtins.def: Added a builtin for class_f64.
19722 * builtin-types.def: Added a builtin type needed by class_f64.
19724 2017-05-03 Jason Merrill <jason@redhat.com>
19726 * timevar.def: Add TV_CONSTEXPR.
19728 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19730 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19732 2017-05-03 Martin Jambor <mjambor@suse.cz>
19734 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19735 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19736 * ipa-cp.c (ipcp_propagate_stage): Do not call
19737 ipa_update_after_lto_read.
19738 * ipa-inline.c (ipa_inline): Likewise.
19740 2017-05-03 Martin Jambor <mjambor@suse.cz>
19742 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19743 tag. Added a default constructor and a destructor.
19744 (ipa_edge_args_sum_t): New class;
19745 (ipa_edge_args_sum): Declare.
19746 (ipa_edge_args_vector): Remove declaration.
19747 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19748 (ipa_free_edge_args_substructures): Remove declaration.
19749 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19750 (ipa_edge_args_info_available_for_edge_p): Likewise.
19751 * ipa-prop.c (ipa_edge_args_vector): Removed.
19752 (edge_removal_hook_holder): Likewise.
19753 (edge_duplication_hook_holder): Likewise.
19754 (ipa_edge_args_sum): New variable.
19755 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19756 ipa_edge_args_vector.
19757 (ipa_free_edge_args_substructures): Likewise.
19758 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19759 ipa_edge_args_vector.
19760 (ipa_edge_removal_hook): Turned into method
19761 ipa_edge_args_sum_t::remove.
19762 (ipa_edge_duplication_hook): Turned into method
19763 ipa_edge_args_sum_t::duplicate.
19764 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19765 registering edge hooks.
19766 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19767 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19768 ipa_edge_args_sum instead of ipa_edge_args_vector.
19769 * ipa-profile.c (ipa_profile): Likewise.
19771 2017-05-03 Martin Jambor <mjambor@suse.cz>
19773 * symbol-summary.h (function_summary): New method exists.
19774 (function_summary::symtab_removal): Deallocate through release.
19775 (call_summary): New class.
19776 (gt_ggc_mx): New overload.
19777 (gt_pch_nx): Likewise.
19778 (gt_pch_nx): Likewise.
19780 2017-05-03 Jeff Law <law@redhat.com>
19782 PR tree-optimization/78496
19783 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19784 from simplify_cond_using_ranges. Split off code to walk
19785 backwards through casts into ...
19786 (simplify_cond_using_ranges_2): New function.
19787 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19788 (execute_vrp): After identifying jump threads, call
19789 simplify_cond_using_ranges_2.
19791 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19794 * ipa-inline.h (inline_summary): Add ctor.
19795 (create_ggc): Do not use ggc_cleared_alloc.
19797 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19798 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19800 * gcc.c (handle_braces): Support escaping in switch matching
19802 * doc/invoke.texi (Spec Files): Document it.
19803 Remove superfluous @code markup in items.
19805 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19807 * diagnostic-show-locus.c (struct column_range): New struct.
19808 (get_affected_columns): New function.
19809 (get_printed_columns): New function.
19810 (struct correction): New struct.
19811 (correction::ensure_capacity): New function.
19812 (correction::ensure_terminated): New function.
19813 (struct line_corrections): New struct.
19814 (line_corrections::~line_corrections): New dtor.
19815 (line_corrections::add_hint): New function.
19816 (layout::print_trailing_fixits): Reimplement in terms of the new
19818 (selftest::test_overlapped_fixit_printing): New function.
19819 (selftest::diagnostic_show_locus_c_tests): Call it.
19821 2017-05-03 Nathan Sidwell <nathan@acm.org>
19823 Canonicalize canonical type hashing
19824 * tree.h (type_hash_canon_hash): Declare.
19825 * tree.c (type_hash_list, attribute_hash_list): Move into
19826 type_hash_canon_hash.
19827 (build_type_attribute_qual_variant): Break out hash code calc into
19828 type_hash_canon_hash.
19829 (type_hash_canon_hash): New. Generic type hash computation.
19830 (build_range_type_1, build_array_type_1, build_function_type,
19831 build_method_type_directly, build_offset_type, build_complex_type,
19832 make_vector_type): Call it.
19834 2017-05-03 Richard Biener <rguenther@suse.de>
19836 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19837 When all DRs have unknown misaligned do not always peel
19838 when there is a store but apply the same costing model as if
19839 there were only loads.
19841 2017-05-03 Richard Biener <rguenther@suse.de>
19844 PR tree-optimization/80492
19845 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19846 compare_base_decls returning dont-know properly.
19848 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19850 * config/arm/iterators.md (CCSI): New mode iterator.
19851 (arch): New mode attribute.
19852 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19853 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19854 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19855 code iterator for success result mode.
19856 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19857 the corresponding new insn generators.
19859 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19862 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19863 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19865 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19867 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19868 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19869 (DDR_A): Wrap DDR argument in brackets.
19870 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19871 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19872 (DDR_REVERSED_P): Likewise.
19874 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19876 PR tree-optimization/79472
19877 * tree-switch-conversion.c (struct switch_conv_info): Add
19878 contiguous_range and default_case_nonstandard fields.
19879 (collect_switch_conv_info): Compute contiguous_range and
19880 default_case_nonstandard fields, don't clear final_bb if
19881 contiguous_range and only the default case doesn't have the required
19883 (check_all_empty_except_final): Set default_case_nonstandard instead
19884 of failing if contiguous_range and the default case doesn't have empty
19886 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19887 and only the default case doesn't have the required constants. Skip
19889 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19890 if default_case_nonstandard.
19891 (build_constructors): Build constant 1 just once. Assert that default
19892 values aren't inserted in between cases if contiguous_range. Skip
19894 (build_arrays): Skip virtual phis.
19895 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19896 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19897 Handle virtual phis.
19898 (gen_inbound_check): Handle default_case_nonstandard case.
19899 (process_switch): Adjust check_final_bb caller. Call
19900 gather_default_values with the first non-default case instead of
19901 default case if default_case_nonstandard.
19903 2017-05-02 Nathan Sidwell <nathan@acm.org>
19905 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19906 check. Fix formatting.
19908 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19910 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19911 errors when comparing specialized and unspecialized times.
19913 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19915 * diagnostic-show-locus.c
19916 (layout::should_print_annotation_line_p): Make private.
19917 (layout::print_annotation_line): Make private.
19918 (layout::annotation_line_showed_range_p): Make private.
19919 (layout::show_ruler): Make private.
19920 (layout::print_source_line): Make private. Pass in line and
19921 line_width, rather than calling location_get_source_line. Drop
19923 (layout::print_leading_fixits): New method.
19924 (layout::print_any_fixits): Rename to...
19925 (layout::print_trailing_fixits): ...this, and make private.
19926 Don't print newline fixits.
19927 (diagnostic_show_locus): Move logic for printing one row into...
19928 (layout::print_line): ...this new function. Move the
19929 location_get_source_line call and error-handling from
19930 print_source_line to here. Call print_leading_fixits, and rename
19931 print_any_fixits to print_trailing_fixits.
19932 (selftest::test_fixit_insert_containing_newline): Update now that
19933 newlines are partially supported.
19934 (selftest::test_fixit_insert_containing_newline_2): New test.
19935 (selftest::test_fixit_replace_containing_newline): Update comments.
19936 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19937 * edit-context.c (class added_line): New class.
19938 (class edited_line): Describe newline handling in comment.
19939 (edited_line::actually_edited_p): New method.
19940 (edited_line::print_content): Delete redundant decl.
19941 (edited_line::m_predecessors): New field.
19942 (edited_file::print_content): Call edited_line::print_content.
19943 (edited_file::print_diff): Update to support newlines.
19944 (edited_file::print_diff_hunk): Likewise.
19945 (edited_file::print_run_of_changed_lines): New function.
19946 (edited_file::print_diff_line): Convert to...
19947 (print_diff_line): ...this.
19948 (edited_file::get_effective_line_count): New function.
19949 (edited_line::edited_line): Initialize new field m_predecessors.
19950 (edited_line::~edited_line): Clean up m_predecessors.
19951 (edited_line::apply_fixit): Handle newlines.
19952 (edited_line::get_effective_line_count): New function.
19953 (edited_line::print_content): New function.
19954 (edited_line::print_diff_lines): New function.
19955 (selftest::test_applying_fixits_insert_containing_newline): New
19957 (selftest::test_applying_fixits_replace_containing_newline): New
19959 (selftest::insert_line): New function.
19960 (selftest::test_applying_fixits_multiple_lines): Add example of
19962 (selftest::edit_context_c_tests): Call the new tests.
19964 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19966 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19967 parameter cand. Update dump information.
19968 (get_computation_cost): Update uses.
19970 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19972 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19973 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19974 (get_computation_at, rewrite_use_address): Update use of
19975 get_computation_aff.
19977 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19979 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19980 (get_computation): Delete.
19981 (get_computation_cost): Implement like get_computation_cost_at.
19982 Use get_computation_at.
19983 (get_computation_cost_at): Delete.
19984 (rewrite_use_nonlinear_expr): Use get_computation_at.
19985 (rewrite_use_compare, remove_unused_ivs): Ditto.
19987 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19989 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19991 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19993 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19994 (ivopts_global_cost_for_size): Rename parameter and update uses.
19995 (iv_ca_recount_cost): Update uses.
19996 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19997 candidates seperately in n_invs and n_cands.
19998 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20000 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20002 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20003 (find_inv_vars_cb): New.
20004 (find_depends): Renamed to ...
20005 (find_inv_vars): ... this.
20006 (add_candidate_1, force_var_cost): Call find_inv_vars.
20007 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20009 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20011 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20012 inv_vars. Add inv_exprs.
20013 (struct iv_cand): Rename depends_on to inv_vars.
20014 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20015 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20016 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20017 (dump_cand): Dump inv_vars.
20018 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20019 (record_invariant, find_depends, add_candidate_1): Ditto.
20020 (set_group_iv_cost, force_var_cost): Ditto.
20021 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20022 (get_computation_cost_at, get_computation_cost): Ditto.
20023 (determine_group_iv_cost_generic): Ditto.
20024 (determine_group_iv_cost_address): Ditto.
20025 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20026 (determine_group_iv_costs): Ditto.
20027 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20028 (iv_ca_set_remove_invariants): Renamed to ...
20029 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20030 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20031 (iv_ca_set_add_invariants): Renamed to ...
20032 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20033 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20034 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20035 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20036 (create_new_ivs): Remove useless dump.
20038 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20040 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20042 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20043 (iv_ca_set_no_cp, create_new_iv): Ditto.
20045 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20047 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20049 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20051 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20052 function tree_check2.
20054 2017-05-02 Martin Liska <mliska@suse.cz>
20056 * doc/gcov.texi: Add missing preposition.
20057 * gcov.c (function_info::function_info): Properly fill up
20058 all member variables.
20060 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20062 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20064 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20066 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20068 2017-05-02 Martin Liska <mliska@suse.cz>
20071 * lto-streamer-in.c (lto_read_tree_1): Remove
20072 LTO_STREAMER_DEBUG.
20073 * lto-streamer.c (struct tree_hash_entry): Likewise.
20074 (struct tree_entry_hasher): Likewise.
20075 (tree_entry_hasher::hash): Likewise.
20076 (tree_entry_hasher::equal): Likewise.
20077 (lto_streamer_init): Likewise.
20078 (lto_orig_address_map): Likewise.
20079 (lto_orig_address_get): Likewise.
20080 (lto_orig_address_remove): Likewise.
20081 * lto-streamer.h: Likewise.
20082 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20083 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20085 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20087 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20088 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20089 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20090 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20091 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20092 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20093 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20094 (mm_maskz_sub_ss): New intrinsics.
20095 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20096 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20097 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20098 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20099 (__builtin_ia32_subss_mask_round): New builtins.
20100 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20101 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20102 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20104 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20105 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20107 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20110 2017-05-02 Martin Jambor <mjambor@suse.cz>
20112 PR tree-optimization/78687
20113 * tree-sra.c (access): New field parent.
20114 (process_subtree_disqualification): New function.
20115 (disqualify_candidate): Call it.
20116 (build_accesses_from_assign): Reset write flag if creating an
20118 (build_access_subtree): Fill in parent field and also prpagate
20119 down grp_write flag.
20120 (create_artificial_child_access): New parameter set_grp_write, set
20121 grp_write to its value.
20122 (propagate_subaccesses_across_link): Also propagate grp_write flag
20124 (propagate_all_subaccesses): Push the closest parent back to work
20125 queue if add_access_to_work_queue returned true.
20127 2017-05-02 Richard Biener <rguenther@suse.de>
20129 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20130 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20131 -fstrict-overflow documentation.
20132 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20133 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20134 flag_strict_overflow.
20135 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20136 * lto-opts.c (lto_write_options): Do not stream it.
20137 * lto-wrapper.c (merge_and_complain): Do not handle it.
20138 * opts.c (default_options_table): Do not set -fstrict-overflow.
20139 (finish_options): Likewise do not clear it when sanitizing.
20140 * simplify-rtx.c (simplify_const_relational_operation): Do not
20141 test flag_strict_overflow.
20143 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20145 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20146 using enabled attribute.
20147 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20148 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20149 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20150 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20151 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20152 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20153 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20154 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20155 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20156 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20158 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20160 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20162 2017-05-02 Richard Biener <rguenther@suse.de>
20164 PR tree-optimization/80591
20166 2017-04-10 Richard Biener <rguenther@suse.de>
20168 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20171 2017-05-02 Richard Biener <rguenther@suse.de>
20173 PR tree-optimization/80549
20174 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20175 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20176 headers if they do not have a preheader.
20178 2017-05-02 Martin Liska <mliska@suse.cz>
20181 * common.opt: Fix typo.
20182 * doc/invoke.texi: Likewise.
20184 2017-05-01 Jan Beulich <jbeulich@suse.com>
20186 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20187 swapping, add (x,x,m,x,n) alternative.
20189 2017-05-01 Nathan Sidwell <nathan@acm.org>
20191 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20192 unnecessary unadjusted_alignment check.
20194 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20197 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20199 * gimplify.c (gimplify_cilk_detach): New function.
20200 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20201 * tree-core.h: Document EXPR_CILK_SPAWN.
20202 * tree.h (EXPR_CILK_SPAWN): Define.
20204 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20206 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20207 to use new fixit_hint representation, using the "replace" logic.
20208 (get_line_span_for_fixit_hint): Likewise.
20209 (layout::print_any_fixits): Likewise.
20210 (selftest::test_one_liner_many_fixits): Rename to...
20211 (selftest::test_one_liner_many_fixits_1): ...this, and update
20212 comment and expected output to reflect that the multiple fix-it
20213 hints are now consolidated into one insertion.
20214 (selftest::test_one_liner_many_fixits_2): New test.
20215 (selftest::test_diagnostic_show_locus_one_liner): Update for
20217 (selftest::test_fixit_consolidation): Update for fix-it API
20219 * diagnostic.c (print_parseable_fixits): Likewise.
20220 * edit-context.c (edited_line::m_line_events): Convert from
20221 auto_vec <line_event *> to auto_vec <line_event>.
20222 (class line_event): Convert from abstract base class to a concrete
20223 class, taking over the role of replace_event.
20224 (class insert_event): Delete.
20225 (class replace_event): Rename to class line_event. Convert to
20227 (edit_context::add_fixits): Reimplement.
20228 (edit_context::apply_insert): Delete.
20229 (edit_context::apply_replace): Rename to...
20230 (edit_context::apply_fixit): ...this. Convert to half-open range.
20231 (edited_file::apply_insert): Delete.
20232 (edited_file::apply_replace): Rename to...
20233 (edited_file::apply_fixit): ...this.
20234 (edited_line::~edited_line): Drop deletion of events.
20235 (edited_line::apply_insert): Delete.
20236 (edited_line::apply_replace): Rename to...
20237 (edited_line::apply_fixit): ...this. Convert to half-open range.
20238 Update for change to type of m_line_events.
20239 * edit-context.h (edit_context::apply_insert): Delete.
20240 (edit_context::apply_replace): Rename to...
20241 (edit_context::apply_fixit): ...this.
20243 2017-05-01 Martin Sebor <msebor@redhat.com>
20245 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20248 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20251 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20252 __get_cpuid_max returns 0.
20253 (__get_cpuid_count): Ditto.
20255 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20257 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20258 replacement expression is another instance of one of its arguments.
20260 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20263 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20264 check for stack push/pop autoinc.
20265 * config/i386/i386.c (ix86_agi_dependent): Return false
20266 if the only reason why modified_in_p returned true is that
20267 addr is SP based and set_insn is a push or pop.
20269 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20271 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20274 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20277 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20278 (account_size_time): Use two predicates - exec_pred and
20280 (evaluate_conditions_for_known_args): Compute both clause and
20282 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20283 (inline_summary_t::duplicate): Update.
20284 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20286 (compute_inline_parameters): Likewise.
20287 (estimate_edge_size_and_time): Update caluclation of time.
20288 (estimate_node_size_and_time): Compute both time and nonspecialized
20290 (estimate_ipcp_clone_size_and_time): Update.
20291 (inline_merge_summary): Update.
20292 (do_estimate_edge_time): Update.
20293 (do_estimate_edge_size): Update.
20294 (do_estimate_edge_hints): Update.
20295 (inline_read_section, inline_write_summary): Stream both new predicates.
20296 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20298 (compute_inlined_call_time): Cleanup.
20299 (big_speedup_p): Update.
20300 (edge_badness): Update.
20301 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20302 (size_time_entry): Replace predicate by exec_predicate and
20303 nonconst_predicate.
20304 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20305 (estimate_edge_time): Return also nonspec_time.
20306 (reset_edge_growth_cache): Update.
20308 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20310 PR rtl-optimization/80491
20311 * ifcvt.c (noce_process_if_block): When looking for x setter
20312 with missing else_bb, don't check only the insn right before
20313 cond_earliest, but look for the last insn that x is modified in
20314 within the same bb.
20316 PR rtl-optimization/80491
20317 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20319 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20321 PR tree-optimization/80487
20322 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20324 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20326 PR tree-optimization/79697
20327 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20328 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20329 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20331 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20332 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20334 2017-04-28 Martin Sebor <msebor@redhat.com>
20336 PR tree-optimization/80523
20337 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20338 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20340 (maybe_warn, format_directive, parse_directive): Use new functions.
20341 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20343 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20345 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20347 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20349 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20350 target_header_dir): Set correctly.
20351 * configure: Regenerated.
20352 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20353 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20354 instead of SYSTEM_HEADER_DIR.
20356 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20358 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20359 (estimate_local_effects): Likewise.
20360 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20361 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20362 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20363 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20364 do_estimate_edge_time, estimate_edge_time): Likewise.
20365 * ipa-inline-analysis.c (estimate_node_size_and_time,
20366 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20367 (estimate_time_after_inlining): Remove.
20369 2017-04-28 Martin Liska <mliska@suse.cz>
20371 * doc/gcov.texi: Enhance documentation of gcov.
20373 2017-04-28 Martin Liska <mliska@suse.cz>
20375 * doc/gcov.texi: Sort options in alphabetic order.
20376 * doc/gcov-dump.texi: Likewise.
20377 * doc/gcov-tool.texi: Likewise.
20378 * gcov.c (print_usage): Likewise.
20379 * gcov-dump.c (print_usage): Likewise.
20380 * gcov-tool.c (print_merge_usage_message): Likewise.
20381 (print_rewrite_usage_message): Likewise.
20382 (print_overlap_usage_message): Likewise.
20384 2017-04-28 Martin Liska <mliska@suse.cz>
20386 PR gcov-profile/53915
20387 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20389 2017-04-28 Martin Liska <mliska@suse.cz>
20391 PR gcov-profile/79891
20392 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20393 is marked by compiler as living on a line.
20394 (get_cycles_count): Remove usage of the union.
20395 (output_intermediate_file): Likewise.
20396 (find_source): Fix GNU coding style.
20397 (accumulate_line_counts): Remove old non-all block mode.
20398 (output_lines): Remove usage of the union.
20399 * profile.c (output_location): Include all BBs, even if
20400 belonging to a same line (and file) as a previous BB.
20402 2017-04-28 Martin Liska <mliska@suse.cz>
20404 * gcov.c (process_args): Handle new argument 'w'.
20405 (read_graph_file): Assign ID to BBs.
20406 (output_branch_count): Display BB # if verbose flag is set.
20407 (output_lines): Likewise for arcs.
20408 (print_usage): Add '--verbose' option help.
20409 * doc/gcov.texi: Document --verbose (-w) option.
20411 2017-04-28 Martin Liska <mliska@suse.cz>
20413 * gcov.c (struct block_location_info): New struct.
20414 (process_file): Fill up the new structure.
20415 (read_graph_file): Replace usage of encoding by the newly added
20417 (add_line_counts): Likewise.
20418 (accumulate_line_counts): Remove usage of the union.
20419 (function_info::function_info): New function.
20420 (function_info::~function_info): Likewise.
20421 (process_file): Call delete instead of release_function.
20422 (release_function): Release the function.
20423 (release_structures): Call delete instead of release_function.
20424 (solve_flow_graph): Replace usage of num_blocks.
20425 (find_exception_blocks): Likewise.
20426 (output_lines): Fix GNU coding style.
20428 2017-04-28 Martin Liska <mliska@suse.cz>
20431 * coverage.c (coverage_remove_note_file): New function.
20432 * coverage.h: Declare the function.
20433 * toplev.c (finalize): Clean if an error has been seen.
20435 2017-04-28 Martin Liska <mliska@suse.cz>
20437 PR gcov-profile/80031
20438 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20439 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20440 * gcov.c (read_graph_file): Read just number of blocks.
20441 * profile.c (branch_prob): Do not stream 0 flags per a basic
20444 2017-04-28 Martin Liska <mliska@suse.cz>
20446 * gcov-dump.c (tag_*): Add new argument to declarations.
20447 (dump_gcov_file): Likewise.
20448 (tag_blocks): Add and use new argument depth.
20449 (tag_arcs): Likewise.
20450 (tag_lines): Likewise.
20451 (tag_counters): Likewise.
20452 (tag_summary): Likewise.
20453 (dump_working_sets): Use depth to do a proper indentation.
20455 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20458 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20459 * symtab.c (symtab_node::debug_symtab): Move definition here.
20461 2017-04-28 Richard Biener <rguenther@suse.de>
20463 * lto-streamer.h (LTO_major_version): Bump to 7.
20465 2017-04-28 Richard Biener <rguenther@suse.de>
20467 * tree-vrp.c (assert_info): New struct.
20468 (add_assert_info): New helper.
20469 (register_edge_assert_for_2): Refactor to add asserts to a vector
20471 (register_edge_assert_for_1): Likewise.
20472 (register_edge_assert_for): Likewise.
20473 (finish_register_edge_assert_for): New helper actually registering
20474 asserts where live on edge.
20475 (find_conditional_asserts): Adjust.
20476 (find_switch_asserts): Likewise.
20477 (evrp_dom_walker::try_find_new_range): Generalize.
20478 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20480 2017-04-27 Marek Polacek <polacek@redhat.com>
20483 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20484 arg10 and arg11 to itype.
20486 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20488 * doc/extend.texi (Object Size Checking): Improve grammar.
20490 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20493 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20494 that the logic for permitting reciprocal estimates matches that
20497 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20500 * tree.c (type_cache_hasher::equal): Only compare
20501 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20502 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20503 non-aggregate element types.
20504 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20505 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20507 2017-04-27 Richard Biener <rguenther@suse.de>
20509 PR middle-end/80533
20510 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20511 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20512 keeping a reference to a trailing array.
20514 2017-04-27 Richard Biener <rguenther@suse.de>
20516 PR middle-end/80539
20517 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20518 being in loop-closed SSA form conservatively.
20519 (chrec_fold_multiply_poly_poly): Likewise.
20521 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20523 PR middle-end/79665
20524 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20525 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20527 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20530 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20531 (aarch64_function_arg_alignment): Return unsigned int again, but still
20532 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20533 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20534 Don't emit -Wpsabi note.
20535 (aarch64_function_arg_boundary): Likewise.
20536 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20539 2017-04-26 Nathan Sidwell <nathan@acm.org>
20541 * tree.h (crc32_unsigned_n): Declare.
20542 (crc32_unsigned, crc32_unsigned): Make inline.
20543 * tree.c (crc32_unsigned_bits): Replace with ...
20544 (crc32_unsigned_n): ... this.
20545 (crc32_unsigned, crc32_byte): Remove.
20546 (crc32_string): Remove unnecessary braces.
20548 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20550 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20551 * ipa-inline-analysis.c (MAX_TIME): Remove.
20552 (account_size_time): Use sreal for time.
20553 (dump_inline_summary): Update.
20554 (estimate_function_body_sizes): Update.
20555 (estimate_edge_size_and_time): Update.
20556 (estimate_calls_size_and_time): Update.
20557 (estimate_node_size_and_time): Update.
20558 (inline_merge_summary): Update.
20559 (inline_update_overall_summary): Update.
20560 (estimate_time_after_inlining): Update.
20561 (inline_read_section): Update.
20562 (inline_write_summary): Update.
20563 * ipa-inline.c (compute_uninlined_call_time): Update.
20564 (compute_inlined_call_time): Update.
20565 (recursive_inlining): Update.
20566 (inline_small_functions): Update.
20567 (dump_overall_stats): Update.
20568 * ipa-inline.h: Include sreal.h.
20569 (size_time_entry): Turn time to sreal.
20570 (inline_summary): Turn self_time nad time to sreal.
20572 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20574 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20576 (sreal::stream_out, sreal::stream_in): New.
20577 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20579 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20581 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20584 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20587 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20589 (dimode_scalar_chain::compute_convert_gain): Ditto.
20590 (dimode_scalar_chain::make_vector_copies): Ditto.
20591 (dimode_scalar_chain::convert_reg): Ditto.
20592 (dimode_scalar_chain::convert_insn): Ditto.
20593 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20594 (VI248_AVX512BW_1): New mode iterator.
20595 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20596 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20599 2017-04-25 Martin Sebor <msebor@redhat.com>
20601 PR tree-optimization/80497
20602 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20603 constants are representable in HOST_WIDE_INT.
20604 (parse_directive): Ditto.
20606 2017-04-25 Martin Sebor <msebor@redhat.com>
20609 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20610 (new_zero_array): Adjust signature.
20611 (dom_info::dom_init): Used unsigned rather that size_t.
20612 (dom_info::dom_info): Same.
20614 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20615 Jakub Jelinek <jakub@redhat.com>
20618 * config/arm/arm.c: Include gimple.h.
20619 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20620 returns negative, increment ncrn only if it returned positive.
20621 (arm_needs_doubleword_align): Return int instead of bool,
20622 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20623 members, but if there is any such non-FIELD_DECL
20624 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20625 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20626 returns negative, increment nregs only if it returned positive.
20627 (arm_setup_incoming_varargs): Likewise.
20628 (arm_function_arg_boundary): Emit -Wpsabi note if
20629 arm_needs_doubleword_align returns negative, return
20630 DOUBLEWORD_ALIGNMENT only if it returned positive.
20632 2017-04-25 Marek Polacek <polacek@redhat.com>
20635 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20636 first argument to type.
20638 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20641 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20642 type checks to test for compatibility instead of equality.
20644 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20645 Jakub Jelinek <jakub@redhat.com>
20648 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20650 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20651 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20652 the alignment computation, but return their maximum in warn_alignment.
20653 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20654 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20656 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20657 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20660 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20662 * config/arc/simdext.md (dmpyh): Fix typo.
20664 2017-04-25 Richard Biener <rguenther@suse.de>
20666 PR tree-optimization/80492
20667 * alias.c (compare_base_decls): Handle registers with asm
20668 specification conservatively.
20669 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20670 compare_base_decls returning dont-know properly.
20672 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20674 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20675 (legitimate_offset_address_p): New function.
20676 (arc_legitimate_address_p): Use above function.
20678 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20680 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20682 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20684 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20685 ACCH registers whenever they are available.
20687 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20689 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20690 double regs fix when not used.
20692 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20694 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20696 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20697 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20699 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20701 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20702 when emitting short ADD instructions.
20704 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20706 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20708 (cmpsi_cc_c_insn): Likewise.
20709 (cbranchsi4_scratch): Compute proper instruction length using
20710 compact_hreg_operand.
20711 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20713 2017-04-25 Richard Biener <rguenther@suse.de>
20715 PR middle-end/80509
20716 * passes.c (pass_manager::pass_manager): Initialize
20717 m_name_to_pass_map.
20719 2017-04-25 Richard Biener <rguenther@suse.de>
20721 PR tree-optimization/79201
20722 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20724 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20727 * config/s390/vector.md: Split MEM->GPR vector moves for
20728 non-s_operand addresses.
20730 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20733 * config/s390/predicates.md (reload_const_wide_int_operand): New
20735 * config/s390/s390.md ("movti"): Remove d/P alternative.
20736 ("movti_bigconst"): New pattern definition.
20738 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20741 * s390-protos.h (s390_expand_cs_hqi): Removed.
20742 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20743 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20744 modes as well as CCZ1mode and CCZmode.
20745 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20746 signature of s390_emit_compare_and_swap.
20747 (s390_expand_cs_hqi): Likewise, make static.
20748 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20749 compare-and-swap, in some cases.
20750 (s390_expand_cs): Wrapper function.
20751 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20753 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20754 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20755 patterns for small and large integers. Forbid symref memory operands.
20756 Move expander to s390.c. Require cc register.
20757 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20758 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20759 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20760 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20761 symref memory operands. Remove CC mode and call s390_match_ccmode
20763 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20765 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20767 * config/s390/s390.md (define_peephole2): New peephole to help
20768 combining the load-and-test pattern with volatile memory.
20770 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20772 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20773 with CCZmode for TARGET_Z196.
20775 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20777 PR rtl-optimization/80501
20778 * combine.c (make_compound_operation_int): Set subreg_code to SET
20779 even for AND with mask of the sign bit of mode.
20781 PR rtl-optimization/80500
20782 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20783 sum's initial value.
20785 2017-04-25 Julian Brown <julian@codesourcery.com>
20786 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20788 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20790 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20792 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20794 2017-04-25 Julian Brown <julian@codesourcery.com>
20795 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20797 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20798 (thunderx2t99_sha): New Reservation.
20800 2017-04-25 Julian Brown <julian@codesourcery.com>
20801 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20803 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20804 type for 1-element load.
20806 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20808 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20810 2017-04-24 Martin Jambor <mjambor@suse.cz>
20812 PR tree-optimization/80293
20813 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20814 char arrays not totally scalarizable if it is false.
20815 (analyze_all_variable_accesses): Pass correct value in the new
20816 parameter. Add a statistics counter.
20818 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20820 PR middle-end/79931
20821 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20823 2017-04-24 Richard Biener <rguenther@suse.de>
20825 PR tree-optimization/80494
20826 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20827 out for complex types.
20829 2017-04-24 Richard Biener <rguenther@suse.de>
20831 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20832 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20833 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20834 (DFS): Adjust and never fail.
20835 (sccvn_dom_walker::fail): Remove.
20836 (sccvn_dom_walker::before_dom_children): Adjust.
20837 (run_scc_vn): Likewise and never fail.
20838 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20839 (pass_fre::execute): Likewise.
20841 2017-04-24 Richard Biener <rguenther@suse.de>
20843 PR tree-optimization/79725
20844 * tree-ssa-sink.c (statement_sink_location): Return whether
20845 failure reason was zero uses. Move that check later.
20846 (sink_code_in_bb): Deal with zero uses by removing the stmt
20849 2017-04-24 Richard Biener <rguenther@suse.de>
20852 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20853 pointer-based references.
20855 2017-04-24 Richard Biener <rguenther@suse.de>
20858 * pass_manager.h (pass_manager::operator new): Remove.
20859 (pass_manager::operator delete): Likewise.
20860 * passes.c (pass_manager::operator new): Remove.
20861 (pass_manager::operator delete): Likewise.
20862 (pass_manager::pass_manager): Zero individual pass members.
20864 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20867 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20868 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20869 Check "XEXP (src, 1)" operand here.
20870 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20871 Check "XEXP (src, 1)" operand here.
20872 (dimode_scalar_chain::make_vector_copies): Detect count register
20873 of a shift instruction. Zero extend count register from QImode
20874 to DImode to satisfy vector shift pattern count operand predicate.
20875 Substitute vector shift count operand with a DImode copy.
20876 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20879 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20881 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20882 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20883 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20884 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20885 (UNSPEC_NOREX_MEM): Remove definition.
20887 2017-04-21 Richard Biener <rguenther@suse.de>
20889 PR tree-optimization/79547
20890 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20891 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20892 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20893 without any constraints.
20895 2017-04-21 Richard Biener <rguenther@suse.de>
20897 PR tree-optimization/78847
20898 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20900 2017-04-21 Richard Biener <rguenther@suse.de>
20902 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20903 (build_distinct_type_copy): Likewise.
20904 (build_variant_type_copy): Likewise.
20905 * tree.c (build_qualified_type): Pass down mem-stat info.
20906 (build_distinct_type_copy): Likewise.
20907 (build_variant_type_copy): Likewise.
20909 2017-04-21 Richard Biener <rguenther@suse.de>
20911 PR tree-optimization/80237
20912 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20914 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20915 for a simplified result.
20917 2016-04-21 Richard Biener <rguenther@suse.de>
20919 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20920 sth as strict as a simple_iv but a chrec without symbols and an
20921 operand defined in the loop we are peeling (and not some subloop).
20922 (propagate_constants_for_unrolling): Propagate all constants.
20924 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20927 * config/i386/i386.c (print_reg): Remove assert for disalowed
20928 regno values, call output_operand_lossage instead.
20930 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20933 * config/i386/constraints.md (Yc): New register constraint.
20934 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20935 Use Yc constraint for alternative 2 of operand 0. Remove
20936 preferred_for_speed attribute.
20938 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20940 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20941 lastprivate clauses in SIMT case.
20943 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20945 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20947 2017-04-20 Richard Biener <rguenther@suse.de>
20949 PR tree-optimization/57796
20950 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20951 as N scalar stores.
20952 (vect_model_load_cost): Cost gathers as N scalar loads.
20954 2017-04-20 Richard Biener <rguenther@suse.de>
20956 * ggc-page.c (ggc_allocated_p): Rename to ...
20957 (safe_lookup_page_table_entry): ... this and return the lookup
20959 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20961 2017-04-20 Richard Biener <rguenther@suse.de>
20963 PR tree-optimization/80453
20964 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20965 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20966 from the conditions.
20967 (vn_phi_eq): Pass them down.
20968 (vn_phi_lookup): Record them.
20969 (vn_phi_insert): Likewise.
20971 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20973 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20974 uninitialized variable warning to avoid buffer overrun.
20976 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20979 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20980 is suppressed for '{ 0 }' in C.
20982 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20984 * BASE-VER: Set to 8.0.0.
20986 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20988 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20989 priority .init_array and .fini_array section with SECTION_NOTYPE
20992 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20994 PR middle-end/80423
20995 * tree.h (build_array_type): Add typeless_storage default argument.
20996 * tree.c (type_cache_hasher::equal): Also compare
20997 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20998 (build_array_type): Add typeless_storage argument, set
20999 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21001 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21002 (build_array_type): Likewise. Add typeless_storage argument.
21004 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21005 Jakub Jelinek <jakub@redhat.com>
21007 PR tree-optimization/80426
21008 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21009 operation on symbolic operands, also compute the overflow for the
21010 invariant part when the operation degenerates into a negation.
21012 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21015 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21016 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21019 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21021 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21024 * config/avr/avr.c (tree.h): Include it.
21025 (cgraph.h): Include it.
21026 (avr_encode_section_info): Don't warn for uninitialized progmem
21027 variable if it's just an alias.
21029 2017-04-19 Richard Biener <rguenther@suse.de>
21032 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21033 when needed by AutoPGO.
21035 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21038 * doc/lto.texi: Remove an extra 'that'.
21040 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21042 PR rtl-optimization/80429
21043 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21044 are only used in debug insns.
21046 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21047 Vladimir Makarov <vmakarov@redhat.com>
21049 * config/sparc/predicates.md (input_operand): Add comment. Return
21050 true for any memory operand when LRA is in progress.
21051 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21053 2017-04-18 Jeff Law <law@redhat.com>
21056 * mips.md ({return,simple_return}_internal): Do not overwrite
21059 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21061 PR tree-optimization/80443
21062 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21063 instead of adding 1, subtract -1 and similarly instead of subtracting
21066 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21068 PR rtl-optimization/80357
21069 * haifa-sched.c (tmp_bitmap): New variable.
21070 (model_recompute): Handle duplicate use records.
21071 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21072 (free_global_sched_pressure_data): Free it.
21074 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21077 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21078 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21079 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21080 instead of SYSTEM_HEADER_DIR.
21082 2017-04-18 Jeff Law <law@redhat.com>
21084 PR middle-end/80422
21085 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21086 predecessors after walking up the insn chain.
21088 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21091 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21092 sizetype type into debug info.
21094 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21097 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21098 unneeded test for TARGET_UPPER_REGS_SF.
21099 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21101 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21104 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21105 instead of gsi_after_labels.
21107 2017-04-18 Jeff Law <law@redhat.com>
21109 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21113 2017-04-13 Jeff Law <law@redhat.com>
21114 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21115 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21117 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21120 * config/avr/avr.c (intl.h): Include it.
21121 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21123 2017-04-18 Martin Liska <mliska@suse.cz>
21125 PR gcov-profile/78783
21126 * gcov-tool.c (gcov_output_files): Validate that destination
21127 file is either removed by the tool or by a user.
21129 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21130 Guy Benyei <guybe@mellanox.com>
21132 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21133 block, and do not negate it, the stored id is already negative.
21135 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21137 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21139 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21142 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21143 masks of options that should be turned off if the VSX vector
21144 options are turned off.
21145 (OTHER_P8_VECTOR_MASKS): Likewise.
21146 (OTHER_VSX_VECTOR_MASKS): Likewise.
21147 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21148 rs6000_disable_incompatible_switches to validate no type switches
21150 (rs6000_incompatible_switch): New function to disallow turning on
21151 other vector options if -mno-vsx, -mno-power8-vector, or
21152 -mno-power9-vector are specified.
21154 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21156 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21158 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21160 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21161 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21162 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21163 (ARG_POINTER_CFA_OFFSET): Likewise.
21165 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21167 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21168 conditions to take advantage of various optimizations.
21170 2017-04-13 Jeff Law <law@redhat.com>
21172 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21173 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21174 (zero_extendsidi2_dext): Likewise.
21176 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21179 * fold-const.c (fold_ternary_loc): Revert
21180 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21183 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21185 PR rtl-optimization/80343
21186 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21187 new scratch pseudo.
21189 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21192 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21193 to ubsan_encode_value.
21195 2017-04-13 Jeff Law <law@redhat.com>
21197 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21198 appearing in DEBUG_INSNs.
21200 2017-04-13 Martin Liska <mliska@suse.cz>
21202 PR gcov-profile/80413
21203 * gcov-io.c (gcov_write_string): Copy to buffer just when
21204 allocated size is greater than zero.
21206 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21209 * dwarf2out.c (decls_for_scope): Ignore declarations of
21210 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21212 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21215 * ipa-visibility.c (non_local_p): Fix typos.
21216 (localize_node): When localizing symbol in same comdat group,
21217 dissolve the group only when we know external symbols are going
21219 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21221 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21223 PR tree-optimization/79390
21224 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21225 order does not result in usable sequence, retry with reversed operand
21231 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21232 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21233 op0 instead of fold_convert_loc (loc, type, arg0).
21235 2017-04-12 Jeff Law <law@redhat.com>
21237 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21238 has a delay slot in the generated code.
21240 * config/cris/cris.md (cris_preferred_reload_class): Return
21241 GENNONACR_REGS rather than GENERAL_REGS.
21243 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21246 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21247 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21248 signedness of the result type.
21250 2017-04-12 Richard Biener <rguenther@suse.de>
21251 Jeff Law <law@redhat.com>
21253 PR tree-optimization/80359
21254 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21255 trim stores to TARGET_MEM_REFs.
21257 2017-04-12 Richard Biener <rguenther@suse.de>
21259 PR tree-optimization/79390
21260 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21261 threading case even more.
21263 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21266 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21267 for quad_address_p for TImode, instead of just not indexed_address.
21269 2017-04-12 Richard Biener <rguenther@suse.de>
21270 Bernd Edlinger <bernd.edlinger@hotmail.de>
21272 PR middle-end/79671
21273 * alias.c (component_uses_parent_alias_set_from): Handle
21274 TYPE_TYPELESS_STORAGE.
21275 (get_alias_set): Likewise.
21276 * tree-core.h (tree_type_common): Add typeless_storage flag.
21277 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21278 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21279 for types containing members with TYPE_TYPELESS_STORAGE.
21280 (place_field): Likewise.
21281 (layout_type): Likewise for ARRAY_TYPE.
21282 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21283 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21284 TYPE_TYPELESS_STORAGE.
21285 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21287 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21290 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21291 first argument to type.
21293 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21297 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21298 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21299 (rs6000_expand_binop_builtin): Likewise.
21300 (rs6000_expand_ternop_builtin): Likewise; also add missing
21301 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21302 vshasigma built-ins.
21303 * doc/extend.texi: Document that vec_xxpermdi's third argument
21304 must be a constant.
21306 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21308 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21309 Use shift_const cost parameter when calculating gain of STV shifts.
21311 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21313 PR rtl-optimization/70478
21314 * lra-constraints.c (process_alt_operands): Check memory for
21315 disfavoring memory insn operand.
21317 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21319 PR middle-end/80100
21320 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21321 left shift in unsigned HOST_WIDE_INT type.
21323 PR rtl-optimization/80385
21324 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21325 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21328 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21329 if they have any depend clauses.
21331 2017-04-11 Martin Liska <mliska@suse.cz>
21334 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21335 * ipa-split.c (split_function): Create a local comdat symbol
21336 if caller is in a comdat group.
21338 2017-04-11 Martin Liska <mliska@suse.cz>
21341 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21344 2017-04-11 Martin Sebor <msebor@redhat.com>
21346 PR middle-end/80364
21347 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21348 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21350 (directive::set_width, directive::set_precision, format_character):
21352 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21355 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21358 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21359 conflict, set target->arch_name instead of target->cpu_name.
21361 2017-04-11 Richard Biener <rguenther@suse.de>
21363 PR tree-optimization/80374
21364 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21365 build_zero_cst, remove fold_convertible_p check again.
21367 2017-04-11 Martin Liska <mliska@suse.cz>
21370 * ubsan.c (instrument_object_size): Do not instrument register
21373 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21376 * config/i386/i386-builtin-types.def
21377 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21378 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21379 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21380 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21381 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21382 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21383 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21384 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21385 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21386 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21387 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21388 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21389 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21390 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21391 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21392 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21393 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21394 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21395 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21396 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21397 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21398 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21399 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21400 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21401 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21402 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21403 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21404 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21405 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21406 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21407 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21408 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21409 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21410 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21411 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21412 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21413 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21414 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21415 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21416 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21417 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21418 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21419 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21420 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21421 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21422 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21424 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21425 flag to second_arg_count, handle 4 argument function type _COUNT
21426 aliases, handle second_arg_count on second argument rather than last.
21428 2017-04-10 Jeff Law <law@redhat.com>
21430 PR tree-optimization/80374
21431 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21432 record anything if we can not convert integer_zero_node to the
21435 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21438 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21439 Enhance special handling given to the TARGET_P9_MINMAX option in
21440 relation to certain other options.
21442 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21444 PR tree-optimization/80153
21445 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21446 remove POINTER_PLUS_EXPR's base part directly, rather than through
21449 2017-04-10 Richard Biener <rguenther@suse.de>
21450 Bin Cheng <bin.cheng@arm.com>
21452 PR tree-optimization/80153
21453 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21454 the first element of pointer type aff_tree. Build result expr in
21456 (add_elt_to_tree): Convert to type unconditionally. Remove other
21457 fold_convert calls.
21458 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21459 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21461 2017-04-10 Richard Biener <rguenther@suse.de>
21463 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21466 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21468 PR rtl-optimization/70478
21469 * lra-constraints.c (curr_small_class_check): New.
21470 (update_and_check_small_class_inputs): New.
21471 (process_alt_operands): Update curr_small_class_check. Disfavor
21472 alternative insn memory operands. Check available regs for small
21475 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21478 * config/mips/mips.opt (-mvirt): Update description.
21479 * doc/invoke.texi (-mvirt): Likewise.
21481 2017-04-10 Richard Biener <rguenther@suse.de>
21483 PR middle-end/80362
21484 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21485 looking for NEGATE_EXPR in -A / -B to A / B folding.
21487 2017-04-10 Martin Liska <mliska@suse.cz>
21489 PR gcov-profile/80224
21490 * gcov.c (print_usage): Fix usage string.
21491 (get_gcov_intermediate_filename): Remove.
21492 (output_gcov_file): Use both for normal and intermediate format.
21493 (generate_results): Do not initialize special file for
21494 intermediate format.
21496 2017-04-10 Richard Biener <rguenther@suse.de>
21498 PR tree-optimization/80304
21499 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21502 2017-04-10 Nathan Sidwell <nathan@acm.org>
21505 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21506 (rs6000_init_builtins): Use it.
21508 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21510 * config/arm/arm.md (<mrc>): Add mode to SET source.
21511 (<mrrc>): Likewise.
21513 2017-04-10 Richard Biener <rguenther@suse.de>
21515 PR middle-end/80344
21516 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21518 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21521 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21522 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21523 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21524 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21525 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21526 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21527 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21528 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21529 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21530 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21531 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21532 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21533 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21534 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21535 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21536 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21537 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21538 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21539 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21540 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21541 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21542 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21543 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21545 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21547 PR rtl-optimization/70478
21548 * lra-constraints.c: Reverse the last patch.
21550 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21552 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21553 Add comment for WCHAR_T.
21555 2017-04-08 Martin Liska <mliska@suse.cz>
21558 2017-04-07 Martin Liska <mliska@suse.cz>
21561 * ipa-split.c (split_function): Add function part to a same comdat
21564 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21567 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21569 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21571 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21572 (vec_pairing): ...to this.
21573 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21574 (rs6000_sched_init): Adjust for name change.
21575 (struct rs6000_sched_context): Likewise.
21576 (rs6000_init_sched_context): Likewise.
21577 (rs6000_set_sched_context): Likewise.
21579 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21585 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21587 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21588 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21589 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21591 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21593 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21595 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21597 PR rtl-optimization/70703
21598 * ira-color.c (update_conflict_hard_regno_costs): Use
21599 int64_t instead of HOST_WIDE_INT.
21601 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21603 PR rtl-optimization/70478
21604 * lra-constraints.c (process_alt_operands): Disfavor alternative
21605 insn memory operands.
21607 2017-04-07 Jeff Law <law@redhat.com>
21609 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21610 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21612 2017-04-07 Martin Liska <mliska@suse.cz>
21615 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21616 Show error message instead of an ICE.
21618 2017-04-07 Martin Liska <mliska@suse.cz>
21621 * ipa-split.c (split_function): Add function part to a same comdat
21624 2017-04-07 Richard Biener <rguenther@suse.de>
21626 PR middle-end/80341
21627 * tree.c (get_unwidened): Also handle ! for_type case for
21629 * convert.c (do_narrow): Split out from ...
21630 (convert_to_integer_1): ... here. Do not pass final truncation
21631 type to get_unwidened for TRUNC_DIV_EXPR.
21633 2017-04-07 Richard Biener <rguenther@suse.de>
21635 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21637 (aff_combination_const): Adjust.
21638 (aff_combination_scale): Likewise.
21639 (aff_combination_add_elt): Likewise.
21640 (aff_combination_add_cst): Likewise.
21641 (aff_combination_convert): Likewise.
21642 (add_elt_to_tree): Likewise. Remove unused argument.
21643 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21645 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21647 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21649 * config/arm/arm.c (arm_default_short_enums): Use
21650 ARM_DEFAULT_SHORT_ENUMS.
21651 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21653 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21656 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21657 members with redundant out-of-class redeclaration.
21659 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21662 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21663 * config/i386/i386.md (*zero_extendsidi2):
21664 Add (?*x,*x) and (?*v,*v) alternatives.
21666 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21669 * config/i386/i386.c (ix86_expand_builtin)
21670 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21671 mode from insn data. Convert operands to insn operand mode.
21672 Copy operands that don't satisfy insn predicate to a register.
21674 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21676 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21679 2017-04-06 Richard Biener <rguenther@suse.de>
21681 PR tree-optimization/80334
21682 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21683 preserve alignment of accesses.
21685 2017-04-06 Richard Biener <rguenther@suse.de>
21687 PR tree-optimization/80262
21688 * tree-sra.c (build_ref_for_offset): Preserve address-space
21690 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21691 Drop useless address-space information on MEM_REF offsets.
21693 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21695 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21697 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21699 PR rtl-optimization/70703
21700 * ira-color.c (update_conflict_hard_regno_costs): Use
21701 HOST_WIDE_INT instead of long.
21703 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21706 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21707 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21709 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21710 for x86_64 target. Handle -m3dnowa option.
21712 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21714 PR rtl-optimization/70703
21715 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21716 (update_conflict_hard_regno_costs): Use long instead of unsigned
21717 arithmetic for cost calculation.
21719 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21720 Bernd Edlinger <bernd.edlinger@hotmail.de>
21723 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21726 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21729 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21730 ptr_mode with Pmode throughout.
21731 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21732 into probe_stack_range and use DImode.
21734 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21737 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21738 call_eh_return is true.
21740 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21742 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21743 Initialize last_match_fntype_index.
21745 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21748 * tree-nvr.c: Include internal-fn.h.
21749 (pass_return_slot::execute): Ignore internal calls without
21752 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21753 Richard Biener <rguenther@suse.de>
21756 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21757 captures used multiple times, except for the last use.
21758 * generic-match-head.c: Include gimplify.h.
21760 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21762 PR tree-optimization/79390
21763 * target.h (struct noce_if_info): Declare.
21764 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21765 * target.def (noce_conversion_profitable_p): New target hook.
21766 * ifcvt.h (struct noce_if_info): New type, moved from ...
21767 * ifcvt.c (struct noce_if_info): ... here.
21768 (noce_conversion_profitable_p): Renamed to ...
21769 (default_noce_conversion_profitable_p): ... this. No longer
21771 (noce_try_store_flag_constants, noce_try_addcc,
21772 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21773 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21774 instead of noce_conversion_profitable_p.
21775 * config/i386/i386.c: Include ifcvt.h.
21776 (ix86_option_override_internal): Don't override
21777 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21778 (ix86_noce_conversion_profitable_p): New function.
21779 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21780 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21781 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21782 * doc/tm.texi: Regenerated.
21784 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21786 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21789 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21792 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21793 instructions for small multiply cores.
21795 2017-04-04 Jeff Law <law@redhat.com>
21797 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21799 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21800 that are not set by the loop over the elements.
21802 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21805 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21806 int mode, convert_modes it to mode as unsigned, otherwise use
21807 lowpart_subreg to mode rather than SImode.
21808 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21809 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21810 Use DImode instead of SImode for the shift count operand.
21811 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21814 2017-04-04 Richard Biener <rguenther@suse.de>
21816 PR middle-end/80281
21817 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21818 arithmetic done for the negate or the plus. Simplify.
21819 (A - (-B) -> A + B): Likewise.
21820 * fold-const.c (split_tree): Make sure to not negate pointers.
21822 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21824 PR rtl-optimization/60818
21825 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21826 a compare of comparisons with the thing compared if this results
21827 in a different machine mode.
21829 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21831 * alias.c (base_alias_check): Fix typo in comment.
21832 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21833 * cgraphunit.c (symbol_table::compile): Likewise.
21834 * collect2.c (maybe_run_lto_and_relink): Likewise.
21835 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21836 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21837 * config/avr/avr.c (avr_map_op_t): Likewise.
21838 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21839 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21840 * config/epiphany/epiphany.md (movcc): Likewise.
21841 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21842 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21844 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21845 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21846 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21847 * config/sh/sh.c (sh_rtx_costs): Likewise.
21848 * fold-const.c (fold_truth_andor): Likewise.
21849 * genautomata.c (collapse_flag): Likewise.
21850 * gengtype.h (struct type::u::s): Likewise.
21851 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21852 * input.c (FORMAT_AMOUNT): Likewise.
21853 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21854 (known_aggs_to_agg_replacement_list): Likewise.
21855 * ipa-inline-analysis.c: Likewise.
21856 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21857 * ipa-polymorphic-call.c
21858 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21859 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21860 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21862 * modulo-sched.c (apply_reg_moves): Likewise.
21863 * omp-expand.c (build_omp_regions_1): Likewise.
21864 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21865 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21866 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21867 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21868 * value-prof.c: Likewise.
21869 * var-tracking.c (val_reset): Likewise.
21871 2017-04-03 Richard Biener <rguenther@suse.de>
21873 PR tree-optimization/80275
21874 * fold-const.c (split_address_to_core_and_offset): Handle
21877 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21879 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21880 descriptors is at least equal to that of functions.
21882 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21884 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21886 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21889 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21890 (mov<IMOD4:mode>): New expander.
21891 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21893 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21895 PR rtl-optimization/79405
21896 * fwprop.c (propagations_left): New variable.
21897 (forward_propagate_into): Decrement it.
21898 (fwprop_init): Initialize it.
21899 (fw_prop): If the variable has reached zero, stop propagating.
21900 (fwprop_addr): Ditto.
21902 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21905 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21906 a FUNCTION_DECL, pass it as decl instead of origin to
21909 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21911 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21914 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21917 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21918 TARGET_VSX_SMALL_INTEGER.
21920 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21922 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21923 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21925 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21927 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21928 extraction from odd-numbered MSA register.
21930 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21932 PR middle-end/80173
21933 * expmed.c (store_bit_field_1): Don't attempt to create
21934 a word subreg out of hard registers wider than word if they
21935 have HARD_REGNO_NREGS of 1 for their mode.
21937 PR middle-end/80163
21938 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21939 conversions to integer types wider than word and pointer.
21942 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21943 (rtx_equal_for_cselib_p): Pass 0 to it.
21944 * cselib.c (cselib_hasher::equal): Likewise.
21945 (rtx_equal_for_cselib_1): Add depth argument. If depth
21946 is 128, don't look up VALUE locs and punt. Increment
21947 depth in recursive calls when walking VALUE locs.
21949 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21951 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21952 (make_gcov_file_name): Use the canonical path name for generating
21954 (read_line): Fix handling of files with ascii null bytes.
21956 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21958 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21959 to initialise a vector register instead
21960 of using a const_int.
21962 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21964 PR translation/80189
21965 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21966 diagnostic messages.
21968 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21971 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21972 (dfp_diex_<mode>): Update mode of operand 1.
21973 * doc/extend.texi (dxex, dxexq): Document change to return type.
21974 (diex, diexq): Document change to argument type.
21976 2017-03-30 Martin Jambor <mjambor@suse.cz>
21979 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21980 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21981 it reflects the signature changes performed at the callee side.
21982 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21983 to cgraph_build_function_type_skip_args.
21984 (build_function_decl_skip_args): Adjust call to the above function.
21986 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21989 * config/i386/sse.md
21990 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21991 register as dest whenever it is a MEM not rtx_equal_p to the
21992 corresponding dup operand, and when forcing into reg move the
21993 reg into the memory afterwards.
21994 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21995 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21996 for the force_reg mode.
21997 (avx512vl_vextractf128<mode>): Use register as dest either
21998 always when a MEM, or when it is a MEM not rtx_equal_p to the
21999 corresponding dup operand, or even not when it is a CONST_VECTOR
22000 depending on the mode and lo vs. hi.
22001 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22003 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22004 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22005 Likewise. Require that operands[2] is even.
22006 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22007 Remove extraneous parens. Require that operands[2] is a multiple
22009 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22010 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22011 disallow memory then.
22013 2017-03-30 Richard Biener <rguenther@suse.de>
22015 PR tree-optimization/77498
22016 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22017 to non-constants over backedges.
22019 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22021 PR rtl-optimization/80233
22022 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22023 as last_combined_insn. Do not test for BARRIER_P separately.
22025 2017-03-29 Andreas Schwab <schwab@suse.de>
22028 * calls.c (prepare_call_address): Convert funexp to Pmode before
22029 copying to temp reg.
22031 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22033 PR tree-optimization/80158
22034 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22035 Handle possible future case of more than one alternate
22037 (replace_rhs_if_not_dup): Likewise.
22038 (replace_one_candidate): Likewise.
22040 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22042 PR rtl-optimization/80193
22043 * ira.c (ira): Do not check allocation for LRA.
22045 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22047 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22048 (nvptx_output_simt_exit): Declare.
22049 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22050 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22051 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22052 (nvptx_declare_function_name): ...here. Emit declaration of local
22053 memory space buffer for omp_simt_enter insn.
22054 (nvptx_output_unisimt_switch): New.
22055 (nvptx_output_softstack_switch): New.
22056 (nvptx_output_simt_enter): New.
22057 (nvptx_output_simt_exit): New.
22058 * config/nvptx/nvptx.h (struct machine_function): New fields
22059 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22060 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22061 (UNSPECV_SIMT_EXIT): Ditto.
22062 (omp_simt_enter_insn): New insn.
22063 (omp_simt_enter): New expansion.
22064 (omp_simt_exit): New insn.
22065 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22067 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22068 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22069 (expand_GOMP_SIMT_EXIT): New.
22070 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22071 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22072 (GOMP_SIMT_EXIT): Ditto.
22073 * target-insns.def (omp_simt_enter): New insn.
22074 (omp_simt_exit): Ditto.
22075 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22077 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22078 (lower_rec_input_clauses): Likewise.
22079 (lower_lastprivate_clauses): Handle SIMT privatization.
22081 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22082 (ompdevlow_adjust_simt_enter): New.
22083 (find_simtpriv_var_op): New.
22084 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22085 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22087 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22088 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22089 (copy_decl_for_dup_finish): Ditto.
22091 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22093 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22096 * config/i386/i386.c (ix86_option_override_internal): Always
22097 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22099 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22101 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22103 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22105 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22106 mark new edge's irreducible flag accordign to it.
22107 (vect_do_peeling): Check loop preheader edge's irreducible flag
22108 and pass it to function slpeel_add_loop_guard.
22110 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22112 PR tree-optimization/80218
22113 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22114 Update block frequencies and counts.
22116 2017-03-28 Richard Biener <rguenther@suse.de>
22118 PR tree-optimization/78644
22119 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22120 of a simplification result we may not use it at all.
22122 2017-03-28 Richard Biener <rguenther@suse.de>
22125 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22126 without arguments, generate default definition of a SSA name.
22128 2017-03-28 Richard Biener <rguenther@suse.de>
22130 PR middle-end/80222
22131 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22132 TYPE_REF_CAN_ALIAS_ALL references.
22133 * fold-const.c (fold_indirect_ref_1): Likewise.
22135 2017-03-28 Martin Liska <mliska@suse.cz>
22138 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22139 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22141 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22142 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22144 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22145 (EXTRA_SPECS): Define.
22146 (SUBTARGET_EXTRA_SPECS): Likewise.
22147 (SUBTARGET_CPP_SPEC): Likewise.
22148 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22149 SUBTARGET_EXTRA_SPECS.
22150 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22152 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22154 * config/arc/simdext.md (vst64_insn): Update pattern.
22155 (vld32wh_insn): Likewise.
22156 (vld32wl_insn): Likewise.
22157 (vld64_insn): Likewise.
22158 (vld32_insn): Likewise.
22160 2017-03-28 Marek Polacek <polacek@redhat.com>
22163 * fold-const.c (fold_comparison): Use protected_set_expr_location
22164 instead of SET_EXPR_LOCATION.
22166 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22168 * tree.c (add_expr): Avoid name lookup warning.
22170 2017-03-27 Jeff Law <law@redhat.com>
22172 PR tree-optimization/80216
22173 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22174 function name. Limit recursion depth.
22175 (record_temporary_equivalences): Corresponding changes.
22177 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22179 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22182 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22185 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22187 * cfgcleanup.c (reg_note_cfa_p): New array.
22188 (insns_have_identical_cfa_notes): New function.
22189 (old_insns_match_p): Don't cross-jump in between /f
22190 and non-/f instructions. If both i1 and i2 are frame related,
22191 verify all CFA notes, their order and content.
22193 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22196 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22197 HImode and SImode with zero extend to DImode to one insn.
22198 (bswap<mode>2_extenddi): Likewise.
22199 (bswapsi2_extenddi): Likewise.
22200 (bswaphi2_extendsi): Likewise.
22201 (bswaphi2): Combine bswap HImode and SImode into one insn.
22202 Separate memory insns from swapping register.
22203 (bswapsi2): Likewise.
22204 (bswap<mode>2): Likewise.
22205 (bswaphi2_internal): Delete, no longer used.
22206 (bswapsi2_internal): Likewise.
22207 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22208 store, and gpr<-gpr swap insns.
22209 (bswap<mode>2_store): Likewise.
22210 (bswaphi2_reg): Register only splitter, combine with the splitter.
22211 (bswaphi2 splitter): Likewise.
22212 (bswapsi2_reg): Likewise.
22213 (bswapsi2 splitter): Likewise.
22214 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22215 the insns into load, store, and register/register insns.
22216 (bswapdi2_ldbrx): Likewise.
22217 (bswapdi2_load): Likewise.
22218 (bswapdi2_store): Likewise.
22219 (bswapdi2_reg): Likewise.
22221 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22223 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22224 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22226 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22229 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22231 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22232 special handling for target option conflicts between dform
22233 options (-mpower9-dform, -mpower9-dform-vector,
22234 -mpower9-dform-scalar) and -mno-direct-move.
22236 2017-03-27 Richard Biener <rguenther@suse.de>
22238 PR tree-optimization/80181
22239 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22241 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22243 * config/arc/predicates.md (move_double_src_operand): Replace the
22244 call to move_double_src_operand with a call to address_operand.
22246 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22248 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22249 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22250 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22252 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22254 * config/arc/predicates.md (long_immediate_loadstore_operand):
22255 Consider scaled addresses cases.
22257 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22259 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22260 restored when in interrupt.
22261 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22262 doesn't have delay slot.
22264 2017-03-27 Richard Biener <rguenther@suse.de>
22267 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22268 inlined thunk clones.
22270 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22273 * asan.c (instrument_derefs): Copy over last operand from
22274 original COMPONENT_REF to the new COMPONENT_REF with
22275 DECL_BIT_FIELD_REPRESENTATIVE.
22276 * ubsan.c (instrument_object_size): Likewise.
22278 2017-03-27 Richard Biener <rguenther@suse.de>
22280 PR tree-optimization/80170
22281 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22282 sure DR/SCEV didnt fold in constants we do not see when looking
22283 at the reference base alignment.
22285 2017-03-27 Richard Biener <rguenther@suse.de>
22287 PR middle-end/80171
22288 * gimple-fold.c (fold_ctor_reference): Properly guard against
22289 NULL return value from canonicalize_constructor_val.
22291 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22294 * config/i386/i386.c (ix86_expand_builtin)
22295 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22296 flags reg setting and flags reg using instructions.
22297 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22298 clobbering instructions to zero extend op2.
22300 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22302 * doc/install.texi (Configuration) <--with-aix-soname>:
22303 Update link to AIX ld.
22305 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22307 PR rtl-optimization/80160
22308 PR rtl-optimization/80159
22309 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22310 reg_alternate_class into account.
22312 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22315 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22316 to consider in curr_insn_transform.
22318 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22320 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22321 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22322 and emit_mode_inner.
22324 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22326 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22327 argument to the overloaded builtin variants. Use the new flag to
22328 deprecate certain builtin variants.
22329 * config/s390/s390-builtin-types.def: Add new builtin types.
22330 * config/s390/s390-builtins.h: Support new flags field for
22331 overloaded builtins.
22332 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22333 (s390_macro_to_expand): Enable vector float data type.
22334 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22335 builtins by incrementing the __VEC__ version number.
22336 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22338 (s390_resolve_overloaded_builtin): Emit error messages depending
22339 on the builtin flags.
22340 * config/s390/s390.c (s390_expand_builtin): Support additional
22341 flags argument. Change error message to match the messages
22342 emitted in s390-c.c.
22343 * config/s390/s390.md: New UNSPEC_* constants.
22344 (op_type): Add new instruction types.
22345 * config/s390/vecintrin.h: Add new builtins and test data class
22347 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22348 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22349 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22350 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22351 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22352 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22354 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22355 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22356 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22357 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22359 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22360 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22361 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22362 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22364 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22365 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22366 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22367 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22368 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22369 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22370 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22372 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22373 ("vec_scatter_element<V_HW_4:mode>_DI")
22374 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22375 ("vec_fpint<mode>", "vflls")
22376 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22377 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22378 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22379 ("*vec_cmphe<mode>_cc"): ... these.
22381 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22382 mode constant instead of magic value.
22384 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22386 * config/s390/s390.c (s390_expand_vec_compare): Support other
22387 vector floating point modes than just V2DF.
22388 (s390_expand_vcond): Likewise.
22389 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22390 (s390_cannot_change_mode_class): Prevent mode changes between TF
22391 and V1TF in vector registers.
22392 * config/s390/s390.md (DF, SF): New mode attributes.
22393 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22394 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22395 SFmode support for VRs.
22396 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22398 (VFT, VF_HW): New mode iterators.
22399 (vw, sdx): New mode attributes.
22400 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22401 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22402 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22403 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22404 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22405 also the new vector floating point modes. Renaming to ...
22407 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22408 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22409 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22410 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22411 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22412 ("vec_unordered<mode>"): ... these.
22414 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22415 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22416 ("*vec_extendv2df"): New insn definitions.
22418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22420 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22421 ("mulditi3_2", "*muldi3_sign"): New patterns.
22422 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22423 rename the pattern definition.
22425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22427 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22429 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22431 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22433 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22434 instruction if possible.
22435 * config/s390/vector.md (vec_halfnumelts): New mode
22437 ("*vec_vllezlf<mode>"): New pattern.
22439 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22441 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22442 ("popcountv4si2", "popcountv2di2"): Rename to ...
22443 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22444 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22446 ("popcount<mode>2_vxe"): New pattern.
22448 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22450 * common/config/s390/s390-common.c (processor_flags_table): Add
22452 * config.gcc: Add arch12.
22453 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22454 Default to arch12 for unknown CPU model numbers.
22455 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22456 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22457 PROCESSOR_max sanity check.
22458 * config/s390/s390-opts.h (enum processor_type): Add
22460 * config/s390/s390.c (processor_table): Add arch12.
22461 (s390_expand_builtin): Add check for B_VXE flag.
22462 (s390_issue_rate): Add PROCESSOR_ARCH12.
22463 (s390_get_sched_attrmask): Likewise.
22464 (s390_get_unit_mask): Likewise.
22465 (s390_sched_score): Enable z13 scheduling for arch12.
22466 (s390_sched_reorder): Likewise.
22467 (s390_sched_variable_issue): Likewise.
22468 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22470 (s390_tune_attr): Use z13 scheduling also for arch12.
22471 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22472 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22473 (TARGET_VXE_P): New macros.
22474 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22475 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22476 * config/s390/s390.opt: Add arch12 as processor_type.
22478 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22480 * config/s390/s390.md
22481 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22482 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22483 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22485 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22486 Rename expanders to ...
22488 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22489 ("fixuns_truncdddi2_emu"): ... these.
22491 ("fixuns_trunc<mode>si2_emu"): New expander.
22493 ("*fixuns_truncdfdi2_z13"): Rename to ...
22494 ("*fixuns_truncdfdi2_vx"): ... this.
22496 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22498 * config/s390/2964.md: Remove the single element vector compare
22499 instructions which are no longer used.
22500 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22502 (s390_canonicalize_comparison): Remove handling of DFmode
22504 (s390_expand_vec_compare_scalar): Remove function.
22505 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22506 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22508 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22510 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22512 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22514 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22515 will anyway by matched by mov<mode>_64dfp.
22517 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22519 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22520 vlef/vstef. Add missing operand to vleif.
22522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22524 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22525 pair for all vector types with 64 bit elements.
22526 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22527 * config/s390/vector.md (V_HW_64): ... here.
22528 (V_128_NOSINGLE): New mode iterator.
22529 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22530 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22531 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22532 ("*vec_load_pairv2di"): Change to ...
22533 ("*vec_load_pair<mode>"): ... this one.
22535 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22537 * config/s390/constraints.md: Add comments.
22538 (jKK): Reject element sizes > 8 bytes.
22539 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22541 * config/s390/s390.md: Add the s_operand checks formerly in
22542 s390_split_ok_p to various splitters where they are still
22544 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22545 for 128 bit vectors. Plus two splitters.
22547 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22549 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22552 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22555 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22556 error if the boundary argument is not constant.
22558 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22560 PR rtl-optimization/80112
22561 * loop-doloop.c (doloop_condition_get): Don't check condition
22562 if cmp isn't SET with IF_THEN_ELSE src.
22564 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22566 PR tree-optimization/80158
22567 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22568 replacing a candidate statement, also replace it for the
22569 candidate's alternate interpretation.
22570 (replace_rhs_if_not_dup): Likewise.
22571 (replace_one_candidate): Likewise.
22573 2017-03-24 Richard Biener <rguenther@suse.de>
22575 PR tree-optimization/80167
22576 * graphite-isl-ast-to-gimple.c
22577 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22579 (translate_isl_ast_to_gimple::get_rename): Likewise.
22581 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22583 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22584 handling of certain combinations of target options, including the
22585 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22586 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22588 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22591 * config/arm/arm.md (*load_multiple): Add reload_completed to
22592 matching condition.
22594 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22595 Richard Biener <rguenth@suse.de>
22597 PR tree-optimization/79908
22598 PR tree-optimization/80136
22599 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22600 been cast away, gimplify_and_add suffices.
22602 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22604 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22606 2017-03-23 Richard Biener <rguenther@suse.de>
22608 PR tree-optimization/80032
22609 * gimplify.c (gimple_push_cleanup): Forced unconditional
22610 cleanups still have to go to the conditional_cleanups
22613 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22615 PR tree-optimization/80072
22616 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22618 (next_operand_entry_id): Change type to unsigned int.
22619 (sort_by_operand_rank): Make sure to return the right return value
22620 even if unsigned fields are bigger than INT_MAX.
22621 (struct oecount): Change cnt and id type to unsigned int.
22622 (oecount_hasher::equal): Formatting fix.
22623 (oecount_cmp): Make sure to return the right return value
22624 even if unsigned fields are bigger than INT_MAX.
22625 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22628 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22629 TREE_READONLY on result if writing it more than once.
22632 * doc/invoke.texi (-fsanitize=thread): Document that with
22633 -fnon-call-exceptions atomics are not able to throw
22637 * tsan.c: Include tree-eh.h.
22638 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22639 maybe_clean_or_replace_eh_stmt where needed.
22640 (instrument_memory_accesses): Add cfg_changed argument.
22641 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22642 if it returned true.
22643 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22645 PR rtl-optimization/63191
22646 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22647 wrapper function, moved the whole old content into ...
22648 (ix86_delegitimize_address_1): ... this. New inline function.
22649 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22650 true as last argument instead of ix86_delegitimize_address.
22652 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22654 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22655 cortexa57_branch_cost.
22657 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22659 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22661 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22664 * doc/md.texi (Constraints): Document wA constraint.
22665 * config/rs6000/constraints.md (wA): New.
22666 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22667 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22668 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22669 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22671 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22674 * gimplify.c (is_oacc_declared): New function.
22675 (oacc_default_clause): Use it to set default flags for acc declared
22676 variables inside parallel regions.
22677 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22678 declared variables.
22679 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22680 declare attribute to any decl as necessary.
22682 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22685 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22686 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22687 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22688 (arm_arch_lpae): This.
22689 * config/arm/arm.c (arm_arch7ve): Rename into ...
22690 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22691 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22694 2017-03-22 Martin Liska <mliska@suse.cz>
22697 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22698 error message instead of an ICE.
22700 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22702 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22704 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22706 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22708 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22711 2017-03-21 Martin Sebor <msebor@redhat.com>
22713 * doc/extend.texi: Use "cannot" instead of "can't."
22714 * doc/hostconfig.texi: Same.
22715 * doc/install.texi: Same.
22716 * doc/invoke.texi: Same.
22717 * doc/loop.texi: Same.
22718 * doc/md.texi: Same.
22719 * doc/objc.texi: Same.
22720 * doc/rtl.texi: Same.
22721 * doc/tm.texi: Same.
22722 * doc/tm.texi.in: Same.
22723 * doc/trouble.texi: Same.
22725 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22728 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22729 (collect_checksum_attributes): Set it.
22730 (die_checksum_ordered): Use it.
22732 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22734 PR tree-optimization/79908
22735 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22736 change: For a VA_ARG whose LHS has been cast away, use
22737 force_gimple_operand to construct the side effects.
22739 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22741 PR translation/80001
22742 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22743 more amenable to translation.
22744 (oacc_loop_auto_partitions): Likewise.
22746 2017-03-21 Marek Polacek <polacek@redhat.com>
22747 Martin Sebor <msebor@redhat.com>
22749 PR tree-optimization/80109
22750 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22751 on INTEGRAL_TYPE_P.
22753 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22754 Segher Boessenkool <segher@kernel.crashing.org>
22757 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22758 check reg_used_between_p between insn and one of succ or succ2
22759 depending on if succ is artificial insn not inserted into insn
22762 2017-03-21 Martin Liska <mliska@suse.cz>
22764 PR gcov-profile/80081
22765 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22766 * doc/gcc.texi: Include gcov-dump stuff.
22767 * doc/gcov-dump.texi: New file.
22769 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22771 PR rtl-optimization/79150
22772 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22773 conditional jump, if the jump is the last insn of the loop.
22775 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22776 Richard Biener <rguenth@suse.de>
22778 PR tree-optimization/79908
22779 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22780 been cast away, use force_gimple_operand to construct the side
22783 2017-03-21 Martin Liska <mliska@suse.cz>
22785 PR libfortran/79956
22786 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22789 2017-03-21 Brad Spengler <spender@grsecurity.net>
22792 * plugin.c (htab_hash_plugin): New function.
22793 (add_new_plugin): Use it and adjust.
22794 (parse_plugin_arg_opt): Adjust.
22795 (init_one_plugin): Likewise.
22797 2017-03-21 Richard Biener <rguenther@suse.de>
22799 PR tree-optimization/80032
22800 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22801 if set force the cleanup to happen unconditionally.
22802 (gimplify_target_expr): Push inserted clobbers with force_uncond
22803 to avoid them being removed by control-dependent DCE.
22805 2017-03-21 Richard Biener <rguenther@suse.de>
22807 PR tree-optimization/80122
22808 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22809 va_arg_pack_len when the inlined call stmt requires pack
22811 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22813 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22816 * tsan.c (instrument_builtin_call): If the memory model argument
22817 is not a constant, assume it is valid.
22820 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22823 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22825 PR rtl-optimization/79910
22826 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22827 if its dest is used by an insn before I2 (other than the combined
22828 insns themselves, which are properly handled already).
22830 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22833 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22835 * combine.c (record_used_regs): New static function.
22836 (try_combine): Handle situations where there is an additional
22837 instruction between I2 and I3 which needs to have a LOG_LINK
22841 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22843 * combine.c (try_combine): Delete redundant i1 test. Call
22844 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22846 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22849 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22850 alternatives 13/14.
22852 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22854 PR tree-optimization/80054
22855 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22856 the optimization if a PHI or any of its arguments is not dominated
22857 by the candidate's basis. Use gphi* rather than gimple* as
22859 (replace_profitable_candidates): Clean up a gimple* variable that
22860 should be a gphi* variable.
22862 2017-03-20 Martin Sebor <msebor@redhat.com>
22865 * doc/extend.texi (attribute constructor): Document present limitation.
22867 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22870 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22871 __POWER9_VECTOR__ #ifdef control, change template definition to
22872 use Power9-specific built-in function.
22873 (vec_any_eq): Likewise.
22874 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22875 to control outcomes from this test.
22876 (vector_ae_<mode>p): For VEC_F modes, likewise.
22878 2017-03-20 Ian Lance Taylor <iant@google.com>
22880 * config/i386/i386.c (ix86_function_regparm): Save an extra
22881 register for -fsplit-stack with DECL_STATIC_CHAIN.
22883 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22886 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22887 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22889 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22891 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22893 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22896 2017-03-20 Marek Polacek <polacek@redhat.com>
22899 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22901 2017-03-20 Richard Biener <rguenther@suse.de>
22903 PR tree-optimization/80113
22904 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22905 allocate extra SSA name for PHI def.
22906 (add_close_phis_to_outer_loops): Likewise.
22907 (add_close_phis_to_merge_points): Likewise.
22908 (copy_loop_close_phi_args): Likewise.
22909 (copy_cond_phi_nodes): Likewise.
22911 2017-03-20 Martin Liska <mliska@suse.cz>
22913 PR middle-end/79753
22914 * tree-chkp.c (chkp_build_returned_bound): Do not build
22915 returned bounds for a LHS that's not a BOUNDED_P type.
22917 2017-03-20 Martin Liska <mliska@suse.cz>
22921 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22922 COMPLEX_CST and VECTOR_CST.
22924 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22927 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22928 target operand. A new splitter adds the clobber statement in case
22929 the target operand is dead anyway.
22931 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22933 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22934 to age-old versions of binutils and glibc.
22936 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22938 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22940 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22942 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22944 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22946 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22947 requirement for binutils 2.13.
22949 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22951 * combine.c (try_combine): Delete redundant i1 test. Call
22952 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22954 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22956 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22957 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22959 <riscv64-*-elf>: Re-arrange section
22960 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22961 <riscv32-*-linux>: Likewise.
22962 <riscv64-*-elf>: Likewise
22963 <riscv64-*-linux>: Likewise.
22965 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22968 * aarch64.opt(verbose-cost-dump): Fix typo.
22970 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22973 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22974 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22976 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22978 * reload.c (find_reloads): When reloading a nonoffsettable address,
22979 use RELOAD_OTHER for it and its address reloads.
22981 PR rtl-optimization/79910
22982 * combine.c (record_used_regs): New static function.
22983 (try_combine): Handle situations where there is an additional
22984 instruction between I2 and I3 which needs to have a LOG_LINK
22987 2017-03-17 Jeff Law <law@redhat.com>
22989 PR tree-optimization/71437
22990 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22991 conditional in the hash table first.
22992 (vrp_dom_walker::before_dom_children): Extract condition from
22993 ASSERT_EXPR. Record condition, its inverion and any implied
22994 conditions as well.
22996 2017-03-17 Marek Polacek <polacek@redhat.com>
22997 Markus Trippelsdorf <markus@trippelsdorf.de>
22999 PR tree-optimization/80079
23000 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23003 2017-03-17 Richard Biener <rguenther@suse.de>
23005 PR middle-end/80075
23006 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23007 Properly verify the LHS before the RHS possibly claims to be
23009 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23012 2017-03-17 Martin Jambor <mjambor@suse.cz>
23014 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23015 (List of -O2 options): Likewise.
23016 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23019 2017-03-17 Tom de Vries <tom@codesourcery.com>
23021 * gcov-dump.c (print_usage): Print bug_report_url.
23023 2017-03-17 Richard Biener <rguenther@suse.de>
23025 PR middle-end/80050
23026 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23027 (parser::peek): Likewise.
23029 2017-03-17 Richard Biener <rguenther@suse.de>
23031 PR tree-optimization/80048
23032 * sese.c (free_sese_info): Properly release rename_map and
23033 copied_bb_map elements.
23035 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23037 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23038 Add linked-list forward and backlinks. Insert on
23039 construction, remove on destruction.
23040 (class pass_store_merging): Add m_stores_head field.
23041 (pass_store_merging::terminate_and_process_all_chains):
23042 Iterate over m_stores_head list.
23043 (pass_store_merging::terminate_all_aliasing_chains):
23045 (pass_store_merging::execute): Check for debug stmts first.
23046 Push new chains onto the m_stores_head stack.
23048 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23051 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23052 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23053 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23055 2017-03-16 Jeff Law <law@redhat.com>
23057 PR tree-optimization/71437
23058 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23059 member function. Implementation moved into after_dom_children
23060 member function and into the threader's thread_outgoing_edges
23062 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23063 some code into new thread_outgoing_edges.
23064 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23065 definition. Simplify marker handling (do it here). Assume we always
23066 have the available expression and the const/copies tables.
23067 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23069 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23070 * tree-vrp.c (equiv_stack): No longer file scoped.
23071 (vrp_dom_walker): New class.
23072 (vrp_dom_walker::before_dom_children): New member function.
23073 (vrp_dom_walker::after_dom_children): Likewise.
23074 (identify_jump_threads): Setup domwalker. Use it rather than
23075 walking edges in a random order by hand. Simplify setup/finalization.
23076 (finalize_jump_threads): Remove.
23077 (vrp_finalize): Do not call identify_jump_threads here.
23078 (execute_vrp): Do it here instead and call thread_through_all_blocks
23081 PR tree-optimization/71437
23082 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23084 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23086 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23087 (dom_opt_dom_walker::thread_across_edge): Remove
23088 handle_dominating_asserts argument. All callers changed.
23089 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23090 changes. Remove calls to lhs_of_dominating_assert. Other
23091 uses of handle_dominating_asserts turn into unconditional code
23092 (simplify_control_stmt_condition_1): Likewise.
23093 (simplify_control_stmt_condition): Likewise.
23094 (thread_through_normal_block, thread_across_edge): Likewise.
23095 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23096 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23097 object if it is not an SSA_NAME.
23098 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23099 before calling into the VRP specific simplifiers.
23100 (identify_jump_threads): Remove handle_dominating_asserts
23103 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23106 * tree-diagnostic.c (default_tree_printer): No longer static.
23107 * tree-diagnostic.h (default_tree_printer): New prototype.
23109 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23111 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23112 Change ins into fmov.
23114 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23116 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23117 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23118 Use h_con constraint for operand 1.
23119 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23120 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23122 2017-03-15 Jeff Law <law@redhat.com>
23124 PR tree-optimization/71437
23125 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23126 (record_temporary_equivalences): Use it.
23128 PR tree-optimization/71437
23129 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23130 tree-ssa-scopedtables.
23131 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23132 (record_conditions, record_cond, vuse_eq): Likewise.
23133 (record_edge_info): Adjust to API tweak of record_conditions.
23134 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23135 (record_temporary_equivalences, optimize_stmt): Likewise.
23136 (eliminate_redundant_computations): Likewise.
23137 (record_equivalences_from_stmt): Likewise.
23138 * tree-ssa-scopedtables.c: Include options.h and params.h.
23139 (vuse_eq): New function, moved from tree-ssa-dom.c
23140 (build_and_record_new_cond): Likewise.
23141 (record_conditions): Likewise. Accept vector of conditions rather
23142 than edge_equivalence structure for first argument.
23143 for the first argument.
23144 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23145 from tree-ssa-dom.c.
23146 (avail_exprs_stack::record_cond): Likewise.
23147 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23148 from tree-ssa-dom.c.
23149 (avail_exprs_stack): Add new member functions lookup_avail_expr
23151 (record_conditions): Declare.
23153 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23156 * lra-constraints.c (process_alt_operands): Increase reject for
23157 reloading an input/output operand.
23159 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23162 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23163 insns to convert from signed/unsigned char/short to IEEE 128-bit
23165 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23167 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23170 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23171 subreg of inner mode for values already in registers.
23173 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23175 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23176 iteration reg is used after the loop.
23178 2017-03-14 Martin Sebor <msebor@redhat.com>
23180 PR tree-optimization/79800
23181 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23182 precision in negative-positive range.
23183 (format_floating): Call non-const overload with adjusted precision.
23185 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23188 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23191 2017-03-14 Martin Sebor <msebor@redhat.com>
23193 PR middle-end/80020
23194 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23195 * builtins.def (aligned_alloc): Use it.
23198 * Makefile.in (GTFILES): Add calls.c.
23199 * calls.c: Include "gt-calls.h".
23201 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23203 PR rtl-optimization/79728
23204 * regs.h (struct target_regs): New field
23205 x_contains_allocatable_regs_of_mode.
23206 (contains_allocatable_regs_of_mode): New macro.
23207 * reginfo.c (init_reg_sets_1): Initialize it, and change
23208 contains_reg_of_mode so it includes global regs as well.
23209 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23210 rather than contains_regs_of_mode.
23212 2017-03-14 Martin Liska <mliska@suse.cz>
23214 * doc/invoke.texi: Document options that can't be combined with
23215 -fcheck-pointer-bounds.
23217 2017-03-14 Martin Liska <mliska@suse.cz>
23219 PR middle-end/79831
23220 * doc/invoke.texi (-Wchkp): Document the option.
23222 2017-03-14 Martin Liska <mliska@suse.cz>
23224 * Makefile.in: Install gcov-dump.
23226 2017-03-14 Martin Liska <mliska@suse.cz>
23228 * multiple_target.c (expand_target_clones): Bail out for
23229 an invalid attribute.
23231 2017-03-14 Richard Biener <rguenther@suse.de>
23233 * alias.c (struct alias_set_entry): Pack properly.
23234 * cfgloop.h (struct loop): Likewise.
23235 * cse.c (struct set): Likewise.
23236 * ipa-utils.c (struct searchc_env): Likewise.
23237 * loop-invariant.c (struct invariant): Likewise.
23238 * lra-remat.c (struct cand): Likewise.
23239 * recog.c (struct change_t): Likewise.
23240 * rtl.h (struct address_info): Likewise.
23241 * symbol-summary.h (function_summary): Likewise.
23242 * tree-loop-distribution.c (struct partition): Likewise.
23243 * tree-object-size.c (struct object_size_info): Likewise.
23244 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23245 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23246 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23247 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23248 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23249 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23250 (struct _stmt_vec_info): Likewise.
23252 2017-03-14 Martin Liska <mliska@suse.cz>
23255 * multiple_target.c (create_dispatcher_calls): Check that
23256 a target can create a function dispatcher.
23258 2017-03-14 Martin Liska <mliska@suse.cz>
23261 * multiple_target.c (expand_target_clones): Drop local.local
23262 flag for default implementation.
23264 2017-03-14 Richard Biener <rguenther@suse.de>
23266 PR tree-optimization/80030
23267 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23269 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23271 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23272 gcc_fallthrough() instead of __attribute__((fallthrough));
23274 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23276 * doc/gcc.texi: Remove "up" link to (DIR).
23277 * doc/gccint.texi: Ditto.
23279 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23281 * doc/install.texi (Specific) <avr>: Remove reference to
23284 2017-03-13 Jeff Law <law@redhat.com>
23286 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23287 attribute rather than comments.
23289 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23290 match_scratch operand is highest.
23292 2017-03-13 Martin Liska <mliska@suse.cz>
23294 PR middle-end/78339
23295 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23296 is a CHKP clone, use original declaration.
23298 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23300 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23301 (arc_conditional_register_usage): Use a different allocation order
23302 when optimizing for size.
23303 * common/config/arc/arc-common.c (arc_option_optimization_table):
23304 Section anchors default on when optimizing for size.
23306 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23308 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23310 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23312 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23313 * config/arc/arc.md (cpu_facility): Add cd variant.
23314 (*movqi_insn): Add code density variant.
23315 (*movhi_insn): Likewise.
23316 (*movqi_insn): Likewise.
23317 (*addsi3_mixed): Likewise.
23318 (subsi3_insn): Likewise.
23320 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23322 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23324 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23326 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23327 expressions with MINUS and UNARY ops.
23329 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23332 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23334 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23335 between vec_select and vector argument.
23336 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23337 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23338 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23339 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23340 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23341 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23343 2017-03-13 Richard Biener <rguenther@suse.de>
23346 * params.def (vect-max-peeling-for-alignment): Fix typo.
23348 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23350 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23351 issue that only occurred with binutils below 2.18.
23353 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23355 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23356 refer to binutils 2.11/2.12 minimum.
23358 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23360 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23361 ftp.kernel.org and simplify binutils requirement.
23363 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23365 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23367 (Optimize Options): Ditto. Also remove redundancy.
23369 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23371 PR translation/79848
23372 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23374 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23375 to G_ to avoid double translation.
23377 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23379 PR translation/79923
23380 * auto-profile.c (get_combined_location): Convert leading
23381 character of diagnostics to lower case and remove trailing period.
23382 (read_profile): Likewise for various diagnostics.
23383 * config/arm/arm.c (arm_option_override): Remove trailing period
23384 from various diagnostics.
23385 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23386 (msp430_expand_delay_cycles): Likewise.
23388 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23391 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23392 full command-line argument, rather than just "str".
23393 (aarch64_validate_march): Likewise.
23394 (aarch64_validate_mtune): Likewise.
23396 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23398 PR rtl-optimization/78911
23399 * lra-assigns.c (must_not_spill_p): New function.
23400 (spill_for): Use it.
23402 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23404 PR tree-optimization/79981
23405 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23406 ATOMIC_COMPARE_EXCHANGE ifn result.
23407 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23408 IFN_ATOMIC_COMPARE_EXCHANGE.
23410 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23413 * opts.c (parse_sanitizer_options): Add missing question mark to
23414 "did you mean" message.
23416 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23418 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23420 (VMULEUH_UNS): Likewise.
23421 (VMULOUB_UNS): Likewise.
23422 (VMULOUH_UNS): Likewise.
23423 * config/rs6000/rs6000.c (builtin_function_type): Remove
23424 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23426 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23429 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23430 x with result of extra_parsing_for_operand_code_0.
23431 (function_reader::extra_parsing_for_operand_code_0): Convert
23432 return type from void to rtx, returning x. When reading
23433 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23434 larger size containing struct block_symbol.
23436 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23438 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23439 -mfloat128-hardware without -m64.
23441 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23444 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23445 entries to the case statement that marks unsigned arguments to
23446 overloaded functions.
23448 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23451 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23453 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23456 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23457 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23459 2017-03-10 Martin Liska <mliska@suse.cz>
23463 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23464 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23467 2017-03-10 Olivier Hainque <hainque@adacore.com>
23469 * tree-switch-conversion (array_value_type): Start by resetting
23470 candidate type to it's main variant.
23472 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23474 PR rtl-optimization/79909
23475 * combine.c (try_combine): Use simplify_replace_rtx on individual
23476 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23477 of the whole CALL_INSN_FUNCTION_USAGE.
23479 PR tree-optimization/79972
23480 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23481 get_range_info on SSA_NAMEs. Formatting fixes.
23483 2017-03-10 Richard Biener <rguenther@suse.de>
23484 Jakub Jelinek <jakub@redhat.com>
23486 PR tree-optimization/77975
23487 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23488 edge to be constant.
23489 (get_val_for): For constant x return it. Formatting fix.
23490 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23491 would use the same bases as the current one.
23493 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23495 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23496 instead of vec_select for V1TImode.
23497 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23499 (VSX_LE_128): Add V1TI to this mode iterator.
23500 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23501 (*vsx_le_perm_store_<mode>): Likewise.
23502 (pre-reload splitter for VSX stores): Likewise.
23503 (post-reload splitter for VSX stores): Likewise.
23504 (*vsx_xxpermdi2_le_<mode>): Likewise.
23505 (*vsx_lxvd2x2_le_<mode>): Likewise.
23506 (*vsx_stxvd2x2_le_<mode>): Likewise.
23508 2017-03-09 Michael Eager <eager@eagercon.com>
23510 Correct failures with --enable-checking=yes,rtl.
23512 * config/microblaze/microblaze.c (microblaze_expand_shift):
23513 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23514 test for const0_rtx.
23515 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23516 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23518 2017-03-09 Richard Biener <rguenther@suse.de>
23520 PR tree-optimization/79977
23521 * graphite-scop-detection.c (scop_detection::merge_sese):
23522 Handle the case of extra exits to blocks dominating the entry.
23524 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23526 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23529 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23531 PR rtl-optimization/79949
23532 * lra-constraints.c (process_alt_operands): Check memory when
23533 trying to predict a cycle. Print about the overall increase.
23535 2017-03-09 Richard Biener <rguenther@suse.de>
23537 PR middle-end/79971
23538 * gimple-expr.c (useless_type_conversion_p): Preserve
23539 TYPE_SATURATING for fixed-point types.
23541 2017-03-09 Richard Biener <rguenther@suse.de>
23544 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23545 alignment of BLKmode params.
23547 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23550 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23551 (VALL_NO_V2Q): Likewise.
23553 * config/aarch64/aarch64-simd.md
23554 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23556 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23557 VALL_NO_V2Q mode iterator.
23558 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23560 2017-03-09 Martin Liska <mliska@suse.cz>
23562 PR tree-optimization/79631
23563 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23564 tree_int_cst_sign_bit just for INTEGER constants.
23566 2017-03-09 Martin Liska <mliska@suse.cz>
23570 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23573 2017-03-09 Marek Polacek <polacek@redhat.com>
23576 * tree.c (inchash::add_expr): Handle TREE_VEC.
23578 2017-03-09 Martin Liska <mliska@suse.cz>
23581 (chkp_narrow_size_and_offset): New function.
23582 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23583 (void chkp_parse_bit_field_ref): New function.
23584 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23585 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23587 2017-03-09 Martin Liska <mliska@suse.cz>
23590 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23591 (chkp_find_bounds_1): Remove gcc_unreachable.
23593 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23596 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23597 BUILT_IN_SYNC*, determine the access type from the size suffix and
23598 always build a MEM_REF with that type. Handle forgotten
23599 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23602 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23603 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23604 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23605 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23606 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23607 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23608 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23609 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23610 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23611 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23612 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23613 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23614 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23615 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23616 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23617 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23618 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23619 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23620 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23621 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23622 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23623 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23624 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23625 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23626 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23627 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23628 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23629 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23630 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23631 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23632 definitions outside of __OPTIMIZE__ guarded section.
23635 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23636 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23637 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23638 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23641 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23643 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23644 ("vfenez<mode>"): Add missing constraints.
23646 2017-03-08 Martin Sebor <msebor@redhat.com>
23649 * config/nds32/nds32.c (nds32_option_override):
23650 Fix misspelled diagnostic.
23652 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23655 * gimplify.c (gimplify_omp_for): Replace index var in outer
23656 taskloop statement with an artificial variable and add
23657 OMP_CLAUSE_PRIVATE clause for it.
23659 2017-03-08 Richard Biener <rguenther@suse.de>
23661 PR tree-optimization/79955
23662 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23663 for accesses that are completely outside of the variable.
23665 2017-03-08 Andrew Haley <aph@redhat.com>
23667 PR tree-optimization/79943
23668 * tree-ssa-loop-split.c (compute_new_first_bound): When
23669 calculating the new upper bound, (END-BEG) should be added, not
23672 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23674 * config/avr/avr.md (setmemhi): Make sure match_dup
23675 operand number comes before match_scratch.
23677 2017-03-08 Richard Biener <rguenther@suse.de>
23679 PR tree-optimization/79920
23680 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23681 with ncopies == 1 to ...
23682 (vect_transform_slp_perm_load): ... here. Properly compute
23683 all element loads by iterating VF times over the group. Do
23684 not handle ncopies (computed in a broken way) in
23685 vect_create_mask_and_perm.
23687 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23690 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23691 is a uniform vector, use uniform_vector_p return value instead of
23692 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23694 2017-03-07 Marek Polacek <polacek@redhat.com>
23696 PR middle-end/79809
23697 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23698 (alloca_call_type): Likewise.
23700 2017-03-07 Martin Liska <mliska@suse.cz>
23702 * gcov.c (process_args): Put comment to correct location.
23704 2017-03-07 Martin Liska <mliska@suse.cz>
23706 PR middle-end/68270
23707 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23708 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23709 (chkp_narrow_bounds_for_field): Likewise.
23710 (chkp_parse_array_and_component_ref): Pass one more argument to
23713 2017-03-07 Richard Biener <rguenther@suse.de>
23715 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23718 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23720 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23721 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23723 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23726 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23727 to end of description.
23728 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23730 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23732 PR rtl-optimization/79901
23733 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23735 (*avx512f_<code><mode>3<mask_name>): ... this.
23736 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23737 iterator instead of VI8_AVX2_AVX512BW.
23739 PR rtl-optimization/79901
23740 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23741 min/max expander, expand it using expand_vec_cond_expr.
23744 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23747 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23750 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23751 to void * for PCH reasons.
23752 * dwarf2out.c (output_loc_operands, output_die): Cast
23753 v.val_vec.array to unsigned char *.
23755 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23758 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23761 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23763 PR rtl-optimization/79571
23764 * lra-constraints.c (process_alt_operands): Calculate static
23765 reject and subtract it from overall when only addresses will be
23768 2017-03-06 Julia Koval <julia.koval@intel.com>
23771 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23772 incoming stack boundary to 128 for 64-bit targets.
23774 2017-03-06 Richard Biener <rguenther@suse.de>
23776 PR tree-optimization/79894
23777 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23778 to NULL after folding it.
23780 2017-03-06 Richard Biener <rguenther@suse.de>
23782 PR tree-optimization/79824
23783 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23784 check disabling peeling for gaps.
23786 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23788 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23789 attributes): Document gettimeofday.
23791 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23793 * config/s390/s390.c (s390_option_override_internal): Set
23794 PARAM_MIN_VECT_LOOP_BOUND
23796 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23798 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23799 * config/s390/s390.md: Likewise.
23801 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23804 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23805 (<avx2_avx512>_perm<mode>): Rename to ...
23806 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23808 (<avx512>_perm<mode>_mask): Rename to ...
23809 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23811 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23812 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23813 instead of VI8F_256_512.
23814 (avx512f_perm<mode>): New define_expand.
23815 (avx512f_perm<mode>_mask): Likewise.
23816 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23817 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23819 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23821 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23822 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23824 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23826 2017-03-06 Martin Liska <mliska@suse.cz>
23829 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23830 when having a SSA NAME w/o VAR_DECL assigned to it.
23832 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23834 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23835 msa_dpsub_<su>_d): Fix MODE for vec_select.
23837 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23839 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23841 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23843 2017-03-06 Richard Biener <rguenther@suse.de>
23845 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23846 * plugin.c (register_plugin_info): Likewise.
23847 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23849 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23851 * config/i386/sse.md (sse_storehps, sse_storelps,
23852 avx_<castmode><avxsizesuffix>_<castmode>,
23853 avx512f_<castmode><avxsizesuffix>_<castmode>,
23854 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23855 in condition that at least one operand is not a MEM.
23857 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23859 PR middle-end/79805
23860 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23861 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23863 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23864 gimple_call_nothrow_p flag based on whether original builtin can throw.
23865 If it can, emit following stmts on the fallthrough edge.
23866 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23867 don't create new bb if inserting just debug stmts on the edge, try to
23868 insert them on the fallthru bb or just reset debug stmts.
23870 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23873 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23874 restore recog_data (including the operand rtxes inside it) around
23875 the call to get_insn_template.
23877 2017-03-03 Martin Sebor <msebor@redhat.com>
23879 PR tree-optimization/79699
23880 * context.c (context::~context): Free MPFR caches to avoid
23881 a memory leak on program exit.
23883 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23885 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23886 Use wide_int::ulow () instead of .elt (0).
23888 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23890 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23891 (*pushxf): Limit oF constraint to 32bit targets and add oC
23892 constraint for 64bit targets.
23893 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23894 (*pushdf): Change rmF constraint to rmC.
23896 2017-03-03 Martin Liska <mliska@suse.cz>
23898 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23899 Remove unused variable.
23901 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23904 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23905 is a memory operand, increase num_memory.
23906 (ix86_expand_args_builtin): Likewise.
23908 2017-03-03 Jan Hubicka <jh@suse.cz>
23911 * ipa-devirt.c (maybe_record_node): Properly handle
23912 __cxa_pure_virtual visibility.
23914 2017-03-03 Martin Liska <mliska@suse.cz>
23916 PR tree-optimization/79803
23917 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23919 (pass_loop_prefetch::execute): Disabled optimization if an
23920 assumption about L1 cache size is not met.
23922 2017-03-03 Martin Liska <mliska@suse.cz>
23924 PR rtl-optimization/79574
23925 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23926 (hash_scan_set): Likewise.
23927 (dump_hash_table): Likewise.
23928 (hoist_code): Likewise.
23930 2017-03-03 Richard Biener <rguenther@suse.de>
23932 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23933 in place of uhigh ().
23934 (fixed_convert_from_real): Likewise.
23936 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23939 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23941 2017-03-03 Richard Biener <rguenther@suse.de>
23943 PR middle-end/79818
23944 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23945 TYPE_OVERFLOW_UNDEFINED check.
23947 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23949 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23951 (vector_ae_<mode>_p): Likewise.
23952 (vector_nez_<mode>_p): Likewise.
23953 (vector_ne_v2di_p): Likewise.
23954 (vector_ae_v2di_p): Likewise.
23955 (vector_ne_<mode>_p): Likewise.
23956 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23958 (vsx_tsqrt<mode>2_fe): Likewise.
23960 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23963 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23965 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23967 PR rtl-optimization/79780
23968 * cprop.c (one_cprop_pass): When second and further conditional trap
23969 in a single basic block is turned into an unconditional trap, turn it
23970 into a deleted note to avoid RTL verification failures.
23972 2017-03-02 Richard Biener <rguenther@suse.de>
23974 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23976 2017-03-02 Richard Biener <rguenther@suse.de>
23978 PR tree-optimization/79345
23980 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23981 param and abort the walk, returning -1 if it is hit.
23982 (walk_aliased_vdefs): Take a limit param and pass it on.
23983 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23984 defaulting to 0 and return a signed int.
23985 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23986 (check_defs): New helper.
23987 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23988 about uninitialized memory.
23989 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23990 bogus uninitialized warning.
23991 (fixed_convert_from_real): Likewise.
23993 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23995 PR tree-optimization/66768
23996 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23997 iv_use if base object can't be determined.
23999 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24001 PR tree-optimization/79345
24002 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24003 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24004 (get_pattern_stats): Initialize it.
24005 * genemit.c (gen_expand): Verify match_scratch numbers come after
24006 match_operand/match_dup numbers.
24007 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24008 match_scratch numbers.
24009 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24011 * config/s390/s390.md (trunctdsd2): Likewise.
24013 2017-03-02 Richard Biener <rguenther@suse.de>
24015 * wide-int.h (wide_int_storage::operator=): Implement in terms
24018 2017-03-02 Richard Biener <rguenther@suse.de>
24020 PR tree-optimization/79777
24021 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24022 the to insert expression to sth existing.
24024 2017-03-01 Martin Sebor <msebor@redhat.com>
24026 PR middle-end/79692
24027 * gimple-ssa-sprintf.c
24028 (directive::known_width_and_precision): New function.
24029 (format_integer): Use it.
24030 (get_mpfr_format_length): Consider the full range of precision
24031 when computing %g output with the # flag. Set the likely byte
24032 count to 3 rather than 1 when precision is indeterminate.
24033 (format_floating): Correct the lower bound of precision.
24035 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24037 * doc/invoke.texi: Document default code model for 64-bit Linux.
24039 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24042 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24043 udiv rather than div since input pattern is unsigned.
24045 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24047 * config/i386/i386.c (print_reg): Warn for values of
24048 unsupported size in integer register.
24050 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24053 * config/rs6000/predicates.md (current_file_function_operand): Do
24054 not allow self calls to be local if the function is replaceable.
24056 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24059 * config/rs6000/altivec.h (vec_ctz and others): Change the
24060 preprocessor macro that controls conditional compilation from
24061 _ARCH_PWR9 to __POWER9_VECTOR__.
24062 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24063 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24064 control (instead of _ARCH_PWR9 control) so that template
24065 definition uses power9-specific function.
24066 (vec_any_eq): Likewise.
24067 (vec_all_ne): Change macro definition to use a power9-specific
24068 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24069 _ARCH_PWR9 control).
24070 (vec_any_eq) Likewise.
24071 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24072 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24073 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24074 support for xvcmpnedp instruction.
24075 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24076 macro expansion so that Power9 implementation of vec_all_ne does
24077 not use the AltiVec predicate framework.
24078 (VCMPNEH_P): Likewise.
24079 (VCMPNEW_P): Likewise.
24080 (VCMPNED_P): Likewise.
24081 (VCMPNEFP_P): Likewise.
24082 (VCMPNEDP_P): Likewise.
24083 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24084 implementation of vec_any_eq to not use AltiVec predicate
24086 (VCMPAEH_P): Likewise.
24087 (VCMPAEW_P): Likewise.
24088 (VCMPAED_P): Likewise.
24089 (VCMPAEFP_P): Likewise.
24090 (VCMPAEDP_P): Likewise.
24091 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24092 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24093 not use the AltiVec predicate framework.
24094 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24095 of vec_any_eq to not use AltiVec predicate framework.
24096 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24097 support for predefined __POWER9_VECTOR__ macro to indicate that
24098 Power9 instruction selection is enabled.
24099 (altivec_overloaded_builtins): Remove extraneous
24100 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24101 function argument types RS6000_BTI_bool_V16QI and
24102 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24103 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24104 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24105 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24106 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24107 Power9 for implementations of vec_cmpne. Change the signature for
24108 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24109 (representing vec_all_ne) to remove the previously described first
24110 argument of type RS6000_BTI_INTSI, as this was an artifact of
24111 reliance on the AltiVec predicate framework, which is no longer
24112 used in the implementation of these functions. Add
24113 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24114 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24115 since, unlike the AltiVec predicate framework implementation, we
24116 do not share function descriptors between vec_alle and vec_anyeq.
24117 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24118 set of modes that receive special treatment even when
24119 TARGET_P9_VECTOR is true. The special treatment emits code that
24120 does not depend on Power9 instructions.
24121 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24122 define_expand to not rely on AltiVec predicate framework.
24123 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24125 (vector_ne_v2di_p): Change this define_expand to not rely on
24126 AltiVec predicate framework.
24127 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24129 (vector_ne_<mode>_p): Change this define_expand to not rely on
24130 AltiVec predicate framework.
24131 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24133 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24134 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24135 define_insn pattern.
24136 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24137 define_insn pattern because the xvcmpne<VSs>. instruction is not
24139 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24140 instruction is not supported.
24142 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24144 * config/nvptx/nvptx.c: Include intl.h.
24146 2017-03-01 Martin Jambor <mjambor@suse.cz>
24149 * ipa-prop.h (ipa_bits): Removed field known.
24150 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24151 to pointers. Adjusted their comments to warn about their sharing.
24152 (ipcp_transformation_summary): Change bits to a vector of pointers.
24153 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24154 (ipa_get_ipa_bits_for_value): Declare.
24155 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24156 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24157 (ipa_bits_hash_table): Likewise.
24158 (ipa_vr_ggc_hash_traits): Likewise.
24159 (ipa_vr_hash_table): Likewise.
24160 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24161 being pointers and vr_known being removed.
24162 (ipa_set_jf_unknown): Likewise.
24163 (ipa_get_ipa_bits_for_value): New function.
24164 (ipa_set_jfunc_bits): Likewise.
24165 (ipa_get_value_range): New overloaded functions.
24166 (ipa_set_jfunc_vr): Likewise.
24167 (ipa_compute_jump_functions_for_edge): Use the above functions to
24168 construct bits and vr parts of jump functions.
24169 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24170 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24172 (ipcp_grow_transformations_if_necessary): Also allocate
24173 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24175 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24176 them. Fix too long lines.
24177 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24178 vr_known being removed.
24179 (ipa_read_jump_function): Use new setter functions to construct bits
24180 and vr parts of jump functions or set them to NULL.
24181 (write_ipcp_transformation_info): Adjust for bits being pointers.
24182 (read_ipcp_transformation_info): Likewise.
24183 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24185 Include gt-ipa-prop.h.
24186 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24188 (ipcp_store_bits_results): Likewise.
24189 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24190 Do not write to existing jump functions but use a temporary instead.
24192 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24195 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24196 attempt to use its first operand as BIT_FIELD_REF base.
24198 2017-03-01 Richard Biener <rguenther@suse.de>
24200 PR middle-end/79721
24201 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24202 interpolating formula in wrapping arithmetic.
24203 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24205 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24207 PR tree-optimization/79734
24208 * tree-vect-generic.c (expand_vector_condition): Optimize
24209 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24210 Handle VEC_COND_EXPR where comparison has different inner width from
24211 type's inner width.
24213 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24215 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24216 markup, and similar issues. Remove @opindex entries for things
24217 that aren't options. Add missing -mmpy-option entries.
24219 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24221 PR tree-optimization/79737
24222 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24223 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24224 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24225 instead of byte_size. Formatting fix.
24226 (shift_bytes_in_array_right): Formatting fix.
24228 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24231 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24232 condition on optimize for the leaf function test.
24234 2017-02-28 Martin Liska <mliska@suse.cz>
24237 * read-rtl-function.c (function_reader::handle_unknown_directive):
24238 Bail out when one uses -flto.
24240 2017-02-28 Martin Liska <mliska@suse.cz>
24242 * common.opt: Replace space with tabular for options of <number>
24244 * config/i386/i386.opt: Show <number> value for
24245 -mlarge-data-threshold.
24246 * opts.c (print_filtered_help): Do not display number in hexadecimal
24249 2017-02-28 Martin Liska <mliska@suse.cz>
24251 * common.opt: Fix --help=option -Q for options which are of
24254 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24256 * config/i386/i386.c (print_reg): Error out for values
24257 of 8-bit size in invalid integer register.
24259 2017-02-28 Martin Sebor <msebor@redhat.com>
24261 PR tree-optimization/79691
24262 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24264 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24267 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24268 gcc_unreachable with output_operand_lossage.
24270 2017-02-28 Richard Biener <rguenther@suse.de>
24272 PR tree-optimization/79740
24273 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24275 (visit_nary_op): Insert the nary into the hashtable if we
24276 pattern-matched sth.
24277 * tree-ssa-pre.c (eliminate_insert): Robustify.
24279 2017-02-28 Richard Biener <rguenther@suse.de>
24281 PR middle-end/79731
24282 * fold-const.c (decode_field_reference): Reject out-of-bound
24285 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24287 * config/i386/i386.c: Include intl.h.
24288 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24289 instead of just cond ? "..." : "...".
24290 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24291 * coverage.c (read_counts_file): Likewise.
24292 * omp-offload.c: Include intl.h.
24293 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24294 of just cond ? "..." : "...".
24295 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24296 of just cond ? "..." : "...".
24298 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24301 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24303 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24304 'tune for' CPU name.
24305 * config/arm/arm-cpu-data.h: Regenerated.
24307 2017-02-28 Richard Biener <rguenther@suse.de>
24309 PR tree-optimization/79732
24310 * tree-inline.c (expand_call_inline): Do not shadow var.
24312 2017-02-28 Richard Biener <rguenther@suse.de>
24314 PR tree-optimization/79723
24315 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24316 address-space properly.
24318 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24320 * doc/optinfo.texi (Optimization groups): Fix option used for
24322 * doc/invoke.texi (-fopt-info): Document "omp".
24323 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24324 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24325 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24326 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24327 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24329 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24331 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24332 OPTGROUP_OMP with "omp".
24334 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24337 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24338 for arithmetic shift of unsigned V2DI.
24340 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24342 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24343 arc/linux.h headers.
24344 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24345 (LINK_SPEC): Likewise.
24346 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24347 (EXTRA_SPECS): Likewise.
24348 (STARTFILE_SPEC): Likewise.
24349 (ENDFILE_SPEC): Likewise.
24350 (LIB_SPEC): Likewise.
24351 (TARGET_SDATA_DEFAULT): Likewise.
24352 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24353 (MULTILIB_DEFAULTS): Likewise.
24354 (DWARF2_UNWIND_INFO): Likewise.
24355 * config/arc/big.h: New file.
24356 * config/arc/elf.h: Likewise.
24357 * config/arc/linux.h: Likewise.
24358 * config/arc/t-uClibc: Remove.
24360 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24362 PR tree-optimization/77536
24363 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24364 (tree_transform_and_unroll_loop): Use above function to compute the
24365 estimated niter of unrolled loop and use it when scaling profile.
24366 Also use count info rather than frequency if it's non-zero.
24367 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24368 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24369 (vect_transform_loop): Call above function.
24371 2017-02-27 Richard Biener <rguenther@suse.de>
24373 PR tree-optimization/45397
24374 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24375 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24376 (visit_nary_op): Add pattern matching for CSEing sign-changed
24377 or truncated operations with wider ones.
24379 2017-02-27 Richard Biener <rguenther@suse.de>
24381 PR tree-optimization/79690
24382 * tree-vect-stmts.c (vectorizable_store): Use vector type
24383 built from the DR with address-space.
24385 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24387 * doc/invoke.texi (Optimize Options): Refine the description
24388 of asan-use-after-return.
24390 2017-02-25 Alan Modra <amodra@gmail.com>
24392 PR rtl-optimization/79584
24393 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24394 base, not ad->base_term, the reg within base. Remove assertion
24395 that ad->base == ad->base_term. Replace gen_int_mode using
24396 bogus mode with const0_rtx.
24398 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24400 PR middle-end/79396
24401 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24402 FMA_EXPR like tcc_binary or tcc_unary.
24404 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24407 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24409 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24410 (output_loc_operands): Handle DW_OP_call_ref and
24411 DW_OP_GNU_variable_value.
24412 (struct variable_value_struct): New type.
24413 (struct variable_value_hasher): Likewise.
24414 (variable_value_hash): New variable.
24415 (string_types): Remove.
24416 (copy_loc_descr): New function.
24417 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24418 (prepend_loc_descr_to_each): New function.
24419 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24420 instead of add_loc_descr_to_each if the first argument is single
24421 location list and the second has multiple.
24422 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24423 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24424 when looking for variable value which doesn't have other location info.
24425 (loc_list_from_tree): Formatting fix.
24426 (gen_array_type_die): Simplify DW_AT_string_length handling.
24427 (adjust_string_types): Remove.
24428 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24429 string_types. Call resolve_variable_values.
24430 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24431 (resolve_addr_in_expr): Likewise. Add A argument.
24432 (copy_deref_exprloc): Remove deref argument. Adjust for the
24433 original expression being DW_OP_GNU_variable_value with optionally
24434 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24435 optionally after it.
24436 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24437 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24438 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24439 (variable_value_hasher::hash, variable_value_hasher::equal): New
24441 (resolve_variable_value_in_expr, resolve_variable_value,
24442 resolve_variable_values, note_variable_value_in_expr,
24443 note_variable_value): New functions.
24444 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24447 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24450 * opts.h (handle_generated_option): Add GENERATED_P argument.
24451 * opts-common.c (handle_option): Adjust function comment.
24452 (handle_generated_option): Add GENERATED_P argument, pass it to
24454 (control_warning_option): Pass false to handle_generated_option
24456 * opts.c (maybe_default_option): Pass true to handle_generated_option
24458 * optc-gen.awk: Likewise.
24460 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24462 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24463 a REG, look at the REG it is a SUBREG of.
24464 (splitter for cmpeqsi_t): Ditto.
24466 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24468 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24469 the special USEs with the pattern of the insn, not the insn itself.
24471 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24474 * doc/invoke.texi: Document -mload-store-pairs.
24476 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24477 Sandra Loosemore <sandra@codesourcery.com>
24479 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24480 argument isn't a CONST_INT.
24481 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24482 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24483 (nios2_validate_compare): Bypass alternate compare logic if *op2
24484 is not a CONST_INT.
24485 (ldstwm_operation_p): Return false if first_base is not a REG or
24486 if first_offset is not a CONST_INT.
24488 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24490 * config/cris/cris.md: Use correct operand in a define_peephole2.
24492 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24494 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24496 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24498 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24499 this_insn if it is an INSN or JUMP_INSN.
24500 (force_offsettable): Look at base, not at addr.
24501 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24502 on things that aren't necessarily CONST_INTs.
24504 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24506 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24507 -mfpmath=sse is the default also for x86-32 targets with SSE2
24508 instruction set when @option{-ffast-math} is enabled
24510 2017-02-24 Jeff Law <law@redhat.com>
24512 PR rtl-optimizatoin/79286
24513 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24516 2017-02-24 Richard Biener <rguenther@suse.de>
24518 PR tree-optimization/79389
24519 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24522 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24524 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24525 function comment to reflect reality.
24526 (loop_exits_before_overflow): Fix typo in function description.
24528 2017-02-24 Richard Biener <rguenther@suse.de>
24530 PR tree-optimization/79389
24531 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24532 properly that a threading opportunity exists. Detect conditional
24533 copy/constant propagation opportunities.
24535 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24537 * config/visium/visium.md (type): Add trap.
24538 (b): New mode attribute.
24539 (*btst): Rename into...
24540 (*btst<mode>): ...this and adjust.
24541 (*cbranchsi4_btst_insn): Rename into...
24542 (*cbranch<mode>4_btst_insn): ...this and adjust.
24543 (trap): New define_insn.
24545 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24547 PR tree-optimization/79389
24548 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24549 (noce_reversed_cond_code): New function.
24550 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24551 reversed_comparison_code. Formatting fix.
24552 (noce_try_store_flag): Test rev_cond != NULL in addition to
24553 reversed_comparison_code.
24554 (noce_try_store_flag_constants): Likewise.
24555 (noce_try_store_flag_mask): Likewise.
24556 (noce_try_addcc): Use rev_cond if non-NULL instead of
24557 reversed_comparison_code.
24558 (noce_try_cmove_arith): Likewise. Formatting fixes.
24559 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24560 (noce_find_if_block): Initialize rev_cond.
24561 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24562 instead of false as last argument never attempt to reverse it
24565 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24567 PR tree-optimization/79663
24568 * tree-predcom.c (combine_chains): Process refs in reverse order
24569 only for ZERO length chains, and add explaining comment.
24571 2017-02-23 Jeff Law <law@redhat.com>
24573 PR tree-optimization/79578
24574 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24575 in call to operand_equal_p.
24577 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24580 * config/i386/cpuid.h: Fix another undefined behavior.
24582 2017-02-23 Richard Biener <rguenther@suse.de>
24584 PR tree-optimization/79683
24585 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24586 vector types for data-refs.
24588 2017-02-23 Martin Liska <mliska@suse.cz>
24590 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24592 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24594 PR middle-end/79665
24595 * internal-fn.c (get_range_pos_neg): Moved to ...
24596 * tree.c (get_range_pos_neg): ... here. No longer static.
24597 * tree.h (get_range_pos_neg): New prototype.
24598 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24599 are known to be in between 0 and signed maximum inclusive, try to
24600 expand both unsigned and signed divmod and use the cheaper one from
24603 2017-02-22 Jeff Law <law@redhat.com>
24605 PR tree-optimization/79578
24606 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24607 to compare base operands.
24609 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24612 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24613 gpc_reg_operand instead of fpr_reg_operand.
24615 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24617 * config/mips/mips.c (mips_return_in_memory): Force FP
24618 vector types to be returned in memory for o32 ABI.
24620 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24622 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24623 instead of DW_TAG_member for static data member declarations and don't
24624 set no_linkage_name for static inline data members.
24625 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24628 2017-02-22 Martin Liska <mliska@suse.cz>
24630 * doc/invoke.texi: Replace inequality signs with square brackets
24633 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24635 PR tree-optimization/68644
24636 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24638 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24641 * lra-constraints.c (simplify_operand_subreg): Handle
24642 WORD_REGISTER_OPERATIONS targets.
24644 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24647 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24648 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24649 elimination by swapping fld*.
24651 2017-02-22 Richard Biener <rguenther@suse.de>
24653 PR tree-optimization/79673
24654 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24655 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24656 irrelevant address-space qualifiers and avoiding a
24657 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24659 2017-02-22 Richard Biener <rguenther@suse.de>
24661 PR tree-optimization/79666
24662 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24663 to not symbolically negate if that may introduce undefined
24666 2017-02-22 Martin Liska <mliska@suse.cz>
24669 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24670 * data-streamer-out.c (streamer_write_gcov_count_stream):
24672 * value-prof.c (stream_out_histogram_value): Make assert more
24673 precise based on type of counter.
24675 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24678 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24679 Use nonimmediate_operand instead of memory_operand for operand 1.
24680 (float-extend standard_x87sse_constant_load splitter): Ditto.
24682 2017-02-21 Jeff Law <law@redhat.com>
24684 PR tree-optimization/79621
24685 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24686 blocks with edges to themselves.
24688 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24691 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24692 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24693 Use gimple_call_builtin_p.
24696 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24697 on temporarily removed DEBUG_INSNs.
24699 PR tree-optimization/79649
24700 * tree-loop-distribution.c (classify_partition): Give up on
24701 non-generic address space loads/stores.
24703 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24705 * doc/loop.texi (Loop manipulation): Remove nonexistent
24706 tree_ssa_loop_version from the documentation.
24707 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24709 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24712 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24713 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24714 * config/rs6000/rs6000.c: Include except.h.
24715 (rs6000_expand_split_stack_prologue): Call
24716 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24718 2017-02-21 Martin Jambor <mjambor@suse.cz>
24721 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24722 have been analyzed.
24724 2017-02-21 Martin Jambor <mjambor@suse.cz>
24726 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24727 for backward compatibility only.
24728 * doc/invoke.texi (Option Summary): Remove all references to
24729 -fipa-cp-alignment.
24731 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24735 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24737 * lra-constraints.c (curr_insn_transform): Handle
24738 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24740 2017-02-21 Martin Liska <mliska@suse.cz>
24742 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24744 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24746 PR translation/79638
24747 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24749 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24752 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24753 (arm_function_ok_for_sibcall): Return false for an indirect call by
24754 descriptor if all the argument registers are used.
24755 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24756 alignment of the function.
24758 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24760 PR tree-optimization/61441
24761 * simplify-rtx.c (simplify_const_unary_operation): For
24762 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24763 the sNaN unmodified.
24765 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24767 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24768 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24769 instead of SYSTEM_HEADER_DIR.
24771 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24772 Martin Liška <mliska@suse.cz>
24774 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24775 Fix typos and grammar, use active voice, and clarify.
24777 2017-02-20 Marek Polacek <polacek@redhat.com>
24779 PR middle-end/79537
24780 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24783 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24785 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24788 * config/i386/i386.c (ix86_expand_builtin): Handle
24789 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24790 ix86_builtins_isa[fcode].isa as a requirement of those
24791 flags and any other flag in the bitmask.
24792 (ix86_init_mmx_sse_builtins): Use 0 instead of
24793 ~OPTION_MASK_ISA_64BIT as mask.
24794 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24795 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24796 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24797 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24799 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24802 * lra-constraints.c (split_reg): Check requested split mode
24803 is supported by the register.
24805 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24807 * lra-constraints.c (simplify_operand_subreg): Remove early
24810 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24813 * lra-constraints.c (curr_insn_transform): Tighten condition
24814 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24816 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 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24825 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24827 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24829 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24832 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24835 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24837 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24840 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24842 * final.c (last_columnnum, override_columnnum): New variables.
24843 (final_start_function): Set last_columnnum, pass it to begin_prologue
24844 hook and pass 0 to dwarf2out_begin_prologue.
24845 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24846 to source_line debug hook.
24847 (notice_source_line): Compute last_columnnum and for debug_column_info
24848 return true on column changes.
24849 * debug.h (struct gcc_debug_hooks): Add column argument to
24850 source_line and begin_prologue hooks.
24851 (debug_nothing_int_charstar_int_bool): Remove prototype.
24852 (debug_nothing_int_int_charstar,
24853 debug_nothing_int_int_charstar_int_bool): New prototypes.
24854 (dwarf2out_begin_prologue): Add column argument.
24855 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24856 begin_prologue hooks.
24857 (debug_nothing_int_charstar_int_bool): Remove.
24858 (debug_nothing_int_int_charstar,
24859 debug_nothing_int_int_charstar_int_bool): New functions.
24860 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24861 through to dwarf2out_source_line.
24862 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24863 (dwarf2out_source_line): Add column argument, emit it if requested.
24864 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24866 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24867 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24868 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24869 through to dwarf2out_begin_prologue.
24870 (vmsdbgout_source_line): Add column argument, pass it through to
24871 dwarf2out_source_line.
24872 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24873 dbxout_source_line caller.
24874 (dbxout_source_line): Add column argument.
24876 * common.opt (gno-column-info, gcolumn-info): New options.
24877 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24878 (check_die): Also test for multiple DW_AT_decl_column attributes.
24879 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24880 DW_AT_decl_column if requested.
24881 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24883 (gen_variable_die): Likewise.
24884 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24885 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24888 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24889 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24890 (ix86_handle_option): Handle OPT_m3dnowa.
24891 * doc/invoke.texi (-m3dnowa): Document.
24892 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24893 -m3dnowa instead of -m3dnow -march=athlon.
24896 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24897 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24899 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24902 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24903 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24904 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24905 generator for vsx_xxpermdi_<mode>_be.
24906 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24907 force big-endian semantics.
24908 (vsx_xxpermdi_<mode>_be): New define_expand with same
24909 implementation as previous version of vsx_xxpermdi_<mode>.
24911 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24913 PR tree-optimization/79327
24914 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24915 variable, its initialization and use.
24917 2017-02-17 Julia Koval <julia.koval@intel.com>
24919 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24920 (OPTION_MASK_ISA_PKU_UNSET): New.
24921 (ix86_handle_option): Handle -mrdpid.
24922 * config/i386/cpuid.h (bit_RDPID): New.
24923 * config/i386/driver-i386.c (host_detect_local_cpu):
24924 Detect RDPID feature.
24925 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24926 * config/i386/i386-c.c (ix86_target_macros_internal):
24928 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24929 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24930 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24931 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24932 * config/i386/i386.md (define_insn "rdpid"): New.
24933 * config/i386/i386.opt Add -mrdpid.
24934 * config/i386/immintrin.h (_rdpid_u32): New.
24936 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24938 PR rtl-optimization/79541
24939 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24940 instead of transforming it into USE.
24942 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24944 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24945 If HONOR_SNANS (SFmode) force the input to a register.
24946 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24947 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24948 an frsp or similar insn.
24950 2017-02-17 Martin Liska <mliska@suse.cz>
24952 PR rtl-optimization/79577
24953 * params.def (selsched-max-sched-times): Increase minimum to 1.
24955 2017-02-17 Martin Liska <mliska@suse.cz>
24957 PR rtl-optimization/79574
24958 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24960 2017-02-17 Martin Liska <mliska@suse.cz>
24962 PR tree-optimization/79529
24963 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24964 ssa_defined_default_def_p to handle cases which are implicitly
24966 * tree-ssa.c (ssa_defined_default_def_p): New function.
24967 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24968 which are implicitly defined.
24969 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24971 2017-02-17 Richard Biener <rguenther@suse.de>
24973 PR middle-end/79576
24974 * params.def (max-ssa-name-query-depth): Limit to 10.
24976 2017-02-17 Richard Biener <rguenther@suse.de>
24978 PR tree-optimization/79552
24979 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24982 2017-02-17 Richard Biener <rguenther@suse.de>
24985 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24987 2017-02-17 Marek Polacek <polacek@redhat.com>
24989 PR middle-end/79536
24990 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24991 (fold_negate_expr): New wrapper.
24993 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24995 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24996 Correct terminology and de-emphasize pre-standard behavior.
24998 2017-02-16 Alan Modra <amodra@gmail.com>
25000 PR rtl-optimization/79286
25001 * ira.c (def_dominates_uses): New function.
25002 (update_equiv_regs): Don't create an equivalence for insns that
25003 may trap where the register def does not dominate the use.
25005 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25007 PR rtl-optimization/78127
25008 * lra.c (lra): Call lra_eliminate before finish the loop after
25011 2017-02-16 Richard Biener <rguenther@suse.de>
25013 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25015 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25016 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25017 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25018 (isl_val_int_from_wi): New function.
25019 (extract_affine_gmp): Rename to ...
25020 (extract_affine_wi): ... this, take a widest_int.
25021 (extract_affine_int): Just wrap extract_affine_wi.
25022 (add_param_constraints): Use isl_val_int_from_wi.
25023 (add_loop_constraints): Likewise, and extract_affine_wi.
25025 2017-02-15 Jeff Law <law@redhat.com>
25027 PR middle-end/79521
25028 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25029 ira_init_register_move_cost_if_necessary.
25031 2017-02-15 Martin Sebor <msebor@redhat.com>
25033 PR middle-end/32003
25034 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25035 removed in a prior commit.
25037 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25039 PR tree-optimization/79347
25040 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25041 counters during peeling.
25043 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25045 * Makefile.in (site.exp): Remove "set ISLVER".
25047 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25050 * real.c (real_from_integer): Call real_convert even for decimal.
25052 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25055 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25057 2017-02-14 Andrew Pinski <apinski@cavium.com>
25059 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25060 cores and change the partno/implementer to be correct.
25061 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25062 the 'B" as the implementer.
25063 * config/aarch64/aarch64-tune.md: Regenerate.
25065 2017-02-14 Carl Love <cel@us.ibm.com>
25067 * config/rs6000/rs6000.c: Add case statement entry to make the
25068 xvcvuxdsp built-in argument unsigned.
25069 * config/rs6000/vsx.md: Fix the source and return operand types so they
25070 match the instruction definitions from the ISA document. Fix typo
25071 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25074 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25077 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25078 member early_clobber_alts.
25079 * lra-lives.c (reg_early_clobber_p): New.
25080 (process_bb_lives): Use it.
25081 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25082 (debug_operand_data): Initialize early_clobber_alts.
25083 (setup_operand_alternative): Set up early_clobber_alts.
25084 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25085 alternatives to new_insn_reg.
25086 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25088 (lra_update_insn_regno_info): Pass the new arg.
25090 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25092 PR middle-end/79505
25093 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25094 (new_oacc_loop_raw): Don't clear already cleared fields.
25097 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25098 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25099 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25101 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25104 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25106 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25109 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25110 the extra instruction to the right place to store 128-bit constant
25113 2017-02-14 Martin Sebor <msebor@redhat.com>
25115 PR middle-end/79448
25116 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25117 warning for strings of unknown length.
25119 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25121 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25123 2017-02-14 Jeff Law <law@redhat.com>
25126 * ira-costs.c (scan_one_insn): Initialize register move costs
25127 for pseudos seen in USE/CLOBBER insns.
25129 PR tree-optimization/79095
25130 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25131 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25132 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25133 if the operands are known to be not equal, then the resulting range
25135 (intersect_ranges): If the new range is ~[0,0] and the old range is
25136 wide, then prefer ~[0,0].
25137 * tree-vrp.c (overflow_comparison_p_1): New function.
25138 (overflow_comparison_p): New function.
25139 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25140 if NAME is used in an overflow test.
25141 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25142 overflow check that can be expressed as an equality test, then adjust
25143 ops to be that equality test.
25145 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25147 * config/s390/s390-builtin-types.def: Remove flags argument.
25148 * config/s390/s390.c (s390_init_builtins): Likewise.
25150 2017-02-14 Martin Liska <mliska@suse.cz>
25152 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25153 vector. Fix trailing white spaces.
25155 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25157 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25160 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25162 PR rtl-optimization/68664
25163 * config/arm/arm.c (arm_sched_can_speculate_insn):
25164 New function. Declare prototype.
25165 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25167 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25169 PR rtl-optimization/68664
25170 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25172 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25174 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25176 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25177 max skip bytes for function, loop and jump.
25179 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25181 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25182 ABS_EXPR for gimple dump.
25184 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25187 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25189 PR tree-optimization/79408
25190 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25191 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25192 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25193 also if rhs1 is INTEGER_CST.
25195 2017-02-14 Richard Biener <rguenther@suse.de>
25197 PR middle-end/79432
25198 * tree-into-ssa.c (insert_phi_nodes): When the function can
25199 have abnormal edges rewrite SSA names with broken use-def
25200 dominance out of SSA and register them for PHI insertion.
25202 2017-02-13 Martin Sebor <msebor@redhat.com>
25204 PR middle-end/79496
25205 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25206 clearing info.nowrite flag when snprintf size argument is a range.
25208 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25210 * cprop.c (cprop_jump): Add missing space in string literal.
25211 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25212 (get_constraint_for_component_ref): Likewise.
25213 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25214 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25215 * lra-constraints.c (process_alt_operands): Likewise.
25216 * ipa-inline.c (inline_small_functions): Likewise.
25217 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25218 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25219 * trans-mem.c (diagnose_tm_1_op): Likewise.
25220 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25221 (grid_parallel_clauses_gridifiable): Likewise.
25223 * config/nvptx/mkoffload.c (process): Add space in between
25226 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25227 "MOD4_SSE_REGS" and "ALL_REGS".
25229 * spellcheck.c (test_data): Add , in between "foo" and "food".
25231 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25234 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25235 boundary crossing check and subsequent code generation agree.
25237 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25239 * config/aarch64/aarch64.c (has_memory_op): Delete.
25240 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25243 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25245 PR rtl-optimization/79388
25246 PR rtl-optimization/79450
25247 * combine.c (distribute_notes): When removing TEM_INSN for which
25248 corresponding dest has last value recorded, invalidate that last
25251 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25253 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25254 of explicit '@'. Add missing assembly comment marker on branch costs
25257 2017-02-13 Nathan Sidwell <nathan@acm.org>
25259 * gengtype-lex.l (<in_struct>): Add '/'.
25261 2017-02-13 Martin Liska <mliska@suse.cz>
25264 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25266 2017-02-13 Richard Biener <rguenther@suse.de>
25268 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25270 * configure: Re-generate.
25271 * config.in: Likewise.
25272 * graphite-dependences.c: Simplify as if
25273 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25274 * graphite-isl-ast-to-gimple.c: Likewise.
25275 * graphite-optimize-isl.c: Likewise.
25276 * graphite-poly.c: Likewise.
25277 * graphite-sese-to-poly.c: Likewise.
25278 * graphite.h: Likewise.
25279 * toplev.c: Include isl/version.h and use isl_version () for
25280 printing the ISL version.
25281 * doc/install.texi: Update ISL requirement.
25283 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25285 * doc/standards.texi (Standards): Update reference to
25288 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25290 * doc/extend.texi (Named Address Spaces): sourceware.org now
25292 * doc/install.texi (Binaries): Ditto.
25295 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25297 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25298 standard terminology "stringize"/"stringizing" throughout.
25299 * doc/cppinternals.texi: Likewise.
25301 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25303 * doc/extend.texi: Fix some spelling mistakes and typos.
25304 * doc/invoke.texi: Likewise.
25306 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25309 * params.def (inline-min-speedup) Change from 10 to 8.
25311 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25313 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25316 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25319 * ipa-inline-analysis.c (get_minimal_bb): New function.
25320 (record_modified): Use it.
25321 (remap_edge_change_prob): Handle also ancestor functions.
25323 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25325 * doc/contrib.texi (Contributors): Remove broken link into
25326 the Mauve CVS repository.
25328 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25330 PR middle-end/79454
25331 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25332 result computation whenever lhs doesn't have vector mode, not
25333 just when it has BLKmode.
25335 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25337 * doc/makefile.texi (profiledbootstrap): Refer to the
25338 installation instructions only in textual form.
25340 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25343 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25345 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25347 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25348 (Specific): Update mingw-w64 reference.
25350 (Specific): Remove broken link to Renesas RX processor.
25352 2017-02-10 Richard Biener <rguenther@suse.de>
25354 * toplev.c (process_options): Do not mention obsolete graphite
25355 options when printing sorry message about missing graphite support.
25356 Mention -floop-nest-optimize.
25358 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25360 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25361 (vtst_p16): Likewise.
25362 (vtstq_p8): Likewise.
25363 (vtstq_p16): Likewise.
25365 (vtstq_p64): Likewise.
25366 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25367 (vset_lane_p64): New.
25368 (vsetq_lane_p64): New.
25370 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25372 PR tree-optimization/79411
25373 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25374 stmt operands are SSA_NAMEs used in abnormal phis.
25375 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25378 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25381 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25384 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25386 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25388 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25390 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25394 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25395 not arbitrary TREE_CONSTANT.
25398 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25399 "omp declare target link" attribute unless is_global_var.
25400 * omp-offload.c (find_link_var_op): Likewise.
25402 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25403 Chung-Lin Tang <cltang@codesourcery.com>
25405 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25407 (gimplify_adjust_omp_clauses): Don't delete TILE.
25408 (gimplify_omp_for): Deal with TILE.
25409 * internal-fn.c (expand_GOACC_TILE): New function.
25410 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25412 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25413 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25415 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25416 avoid DIV for outermost collapse var.
25417 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25418 Remove out of date comments, fix whitespace.
25419 * omp-general.c (omp_extract_for_data): Deal with tiling.
25420 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25421 adjust OLF_DIM_BASE value.
25422 (struct omp_for_data): Add tiling field.
25423 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25424 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25425 for auto loops. Remove default auto determining, moved to
25426 oacc_loop_fixed_partitions.
25427 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25428 stmts, add e_mask field.
25429 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25430 (oacc_thread_numbers): Use oacc_dim_call.
25431 (oacc_xform_tile): New.
25432 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25433 (finish_oacc_loop): Adjust for ifns vector.
25434 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25435 add case for GOACC_TILE fns.
25436 (oacc_loop_xform_loop): Delete.
25437 (oacc_loop_process): Iterate over call list directly, and add
25438 handling for GOACC_TILE fns.
25439 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25441 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25442 vector partitioning to outer loops. Assign 2 partitions to loops
25443 when available. Add TILE handling.
25444 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25445 (execite_oacc_device_lower): Process GOACC_TILE fns,
25446 ignore unknown specs.
25447 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25448 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25449 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25451 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25453 * configure.ac (ACX_BUGURL): Update.
25454 * configure: Regenerate.
25456 2017-02-09 Richard Biener <rguenther@suse.de>
25458 PR tree-optimization/69823
25459 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25460 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25462 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25464 * config/arc/arc-c.def: Add __NPS400__ definition.
25465 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25466 (TARGET_NPS400): Define.
25468 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25470 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25472 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25473 pointer, arch_info.
25474 (arc_cpu_types): Fill the arch_info field with a pointer into the
25475 arc_arch_types table.
25476 (arc_selected_cpu): Declare.
25477 * config/arc/arc.c (arc_selected_cpu): Make global.
25478 (arc_selected_arch): Delete.
25479 (arc_base_cpu): Delete.
25480 (arc_override_options): Remove references to deleted variables,
25481 update access to arch information.
25482 (ARC_OPT): Update access to arch information.
25483 (ARC_OPTX): Likewise.
25484 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25485 (TARGET_ARC600): Update access to arch information.
25486 (TARGET_ARC601): Likewise.
25487 (TARGET_ARC700): Likewise.
25488 (TARGET_EM): Likewise.
25489 (TARGET_HS): Likewise.
25490 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25493 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25496 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25497 condition/operands for integer GE/LE/GEU/LEU operations.
25499 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25501 PR translation/79397
25502 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25505 2017-02-08 Martin Jambor <mjambor@suse.cz>
25508 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25509 whether allocation happened.
25510 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25511 nothing was allocated.
25513 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25515 PR tree-optimization/79408
25516 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25517 constant, but SSA_NAME with a known integer range, use the minimum
25518 of that range instead of op1 to determine if modulo can be replaced
25519 with its first operand.
25521 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25523 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25525 2017-02-08 Richard Biener <rguenther@suse.de>
25527 PR tree-optimization/71824
25528 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25529 Check all loops contained in the merged region.
25531 2017-02-07 Andrew Pinski <apinski@cavium.com>
25533 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25535 2017-02-07 Andrew Pinski <apinski@cavium.com>
25537 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25538 (thunderxt88): Likewise.
25539 (thunderxt81): Disable LSE and change v8.1 to v8.
25540 (thunderxt83): Likewise.
25542 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25543 Richard Biener <rguenther@suse.de>
25545 PR middle-end/79399
25546 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25547 type from int to size_t.
25548 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25550 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25552 PR rtl-optimization/79386
25553 * cprop.c (bypass_conditional_jumps): Initialize
25554 bypass_last_basic_block already before splitting bbs after
25555 unconditional traps...
25556 (bypass_conditional_jumps): ... rather than here.
25559 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25560 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25561 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25562 fix -masm=intel patterns.
25564 2017-02-07 Richard Biener <rguenther@suse.de>
25566 PR tree-optimization/79256
25567 PR middle-end/79278
25568 * builtins.c (get_object_alignment_2): Use min_align_of_type
25569 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25570 and ADJUST_FIELD_ALIGN.
25572 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25574 * doc/tm.texi: Regenerate.
25575 * stor-layout.c (layout_decl): Adjust.
25576 (update_alignment_for_field): Likewise.
25577 (place_field): Likewise.
25578 (min_align_of_type): Likewise.
25579 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25580 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25581 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25582 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25583 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25584 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25585 * config/i386/i386.c (x86_field_alignment): Likewise.
25586 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25587 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25588 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25589 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25590 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25591 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25595 2017-01-30 Richard Biener <rguenther@suse.de>
25597 PR tree-optimization/79256
25598 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25599 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25602 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25604 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25605 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25606 builtins to SImode and emit a zero-extend, if necessary.
25608 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25610 * docs/invoke.texi (RISC-V Options): Alphabetize.
25612 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25614 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25617 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25619 * config/riscv/riscv.c: New file.
25620 * common/config/riscv/riscv-common.c: Likewise.
25621 * config.gcc: Likewise.
25622 * config/riscv/constraints.md: Likewise.
25623 * config/riscv/elf.h: Likewise.
25624 * config/riscv/generic.md: Likewise.
25625 * config/riscv/linux.h: Likewise.
25626 * config/riscv/multilib-generator: Likewise.
25627 * config/riscv/peephole.md: Likewise.
25628 * config/riscv/pic.md: Likewise.
25629 * config/riscv/predicates.md: Likewise.
25630 * config/riscv/riscv-builtins.c: Likewise.
25631 * config/riscv/riscv-c.c: Likewise.
25632 * config/riscv/riscv-ftypes.def: Likewise.
25633 * config/riscv/riscv-modes.def: Likewise.
25634 * config/riscv/riscv-opts.h: Likewise.
25635 * config/riscv/riscv-protos.h: Likewise.
25636 * config/riscv/riscv.h: Likewise.
25637 * config/riscv/riscv.md: Likewise.
25638 * config/riscv/riscv.opt: Likewise.
25639 * config/riscv/sync.md: Likewise.
25640 * config/riscv/t-elf-multilib: Likewise.
25641 * config/riscv/t-linux: Likewise.
25642 * config/riscv/t-linux-multilib: Likewise.
25643 * config/riscv/t-riscv: Likewise.
25644 * configure.ac: Likewise.
25645 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25646 Waterman as RISC-V maintainers.
25647 * doc/install.texi: Add RISC-V entries.
25648 * doc/invoke.texi: Add RISC-V options section.
25649 * doc/md.texi: Add RISC-V constraints section.
25650 * configure: Regenerated.
25652 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25655 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25656 false values to be constant vectors with all 0 or all 1 bits set.
25657 (vcondu<mode><mode>): Likewise.
25658 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25660 (fpmask_comparison_operator): Update comment.
25661 (vecint_comparison_operator): New predicate.
25662 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25663 vector conditionals when the true and false values are constant
25664 vectors with all 0 bits or all 1 bits set.
25666 2017-02-06 Martin Sebor <msebor@redhat.com>
25668 PR tree-optimization/79376
25669 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25671 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25673 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25674 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25675 to simplify split condition.
25677 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25679 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25680 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25683 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25685 PR rtl-optimization/68664
25686 * target.def (can_speculate_insn): New hook.
25687 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25688 * doc/tm.texi: Regenerate.
25689 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25690 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25691 (rs6000_sched_can_speculate_insn): New function.
25693 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25695 PR tree-optimization/79284
25696 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25697 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25698 vectorizable_mask_load_store, vectorizable_operation,
25699 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25700 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25701 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25702 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25703 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25704 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25705 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25706 is_gimple_assign (stmt). Replace another such test with
25707 is_gimple_assign (stmt).
25709 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25712 * config/avr/avr.c (rtl-iter.h): Include it.
25713 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25714 (avr_legitimate_combined_insn): ...and implementation.
25716 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25718 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25719 * config/s390/s390.c (s390_const_operand_ok)
25720 (s390_canonicalize_comparison, s390_extract_part)
25721 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25722 (s390_contiguous_bitmask_p, s390_rtx_costs)
25723 (legitimize_pic_address): Likewise.
25724 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25725 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25726 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25727 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25728 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25730 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25732 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25733 REGNO($0) == REGNO($1).
25735 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25737 * config/s390/linux.h(SIZE_TYPE): Add comment.
25739 2017-02-06 Julian Brown <julian@codesourcery.com>
25740 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25741 Virendra Pathak <virendra.pathak@broadcom.com>
25743 * config/aarch64/aarch64-cores.def: Change the scheduler
25745 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25746 * config/aarch64/thunderx2t99.md: New file.
25748 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25750 * doc/standards.texi (Go Language): Update link to language
25753 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25755 * tree-eh.c (lower_resx): Sanitize profile.
25756 (cleanup_empty_eh_move_lp): Likewise.
25758 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25761 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25763 * cfgloopmanip.h (loop_version): Update prototype.
25764 * modulo-sched.c (sms_schedule): Update call of loop_version.
25765 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25766 * tree-parloops.c (gen_parallel_loop): Likewise.
25767 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25768 * tree-ssa-loop-split.c (split_loop): Likewise.
25769 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25770 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25772 2017-02-05 Martin Liska <mliska@suse.cz>
25775 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25777 (print_operand_address): Initialize a struct to zero.
25779 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25781 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25782 garbage collector only in textual form.
25784 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25786 * doc/extend.texi (x86 specific memory model extensions for
25787 transactional memory): Simplify a phrase.
25789 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25792 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25793 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25794 (atomic_storedi_1): Likewise.
25796 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25798 PR tree-optimization/79338
25799 * tree-parloops.c (gather_scalar_reductions): Don't call
25800 vect_analyze_loop_form for loop->inner before destroying loop's
25803 2017-02-03 Martin Sebor <msebor@redhat.com>
25805 PR tree-optimization/79327
25806 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25807 when precision has resulted in leading zeros.
25808 (format_integer): Adjust the likely counter to assume an unknown
25809 argument that may be zero is non-zero.
25811 2017-02-03 Jason Merrill <jason@redhat.com>
25814 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25815 avoid copying non-taken branch.
25817 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25819 PR tree-optimization/79340
25820 * tree-vect-loop.c (vectorizable_reduction): Release
25821 vec_defs elements after safe_splicing them into other vectors.
25824 PR tree-optimization/79327
25825 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25826 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25828 (format_integer): Use wide_int_to_tree instead of build_int_cst
25829 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25830 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25831 of shortest and longest sequence.
25833 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25835 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25836 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25838 2017-02-03 Walter Lee <walt@tilera.com>
25841 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25842 after initial stackframe link reg save.
25843 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25845 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25848 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25849 wu for stxssp alternative.
25851 2017-02-03 Martin Sebor <msebor@redhat.com>
25853 PR tree-optimization/79352
25854 * gimple-fold.c (get_range_strlen): Add argument.
25855 (get_range_strlen): Change return type to bool.
25856 (get_maxval_strlen): Pass in a dummy argument.
25857 * gimple-fold.h (get_range_strlen): Change return type to bool.
25858 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25859 * tree.h (array_at_struct_end_p): Add argument.
25860 * tree.c (array_at_struct_end_p): Handle it.
25862 2017-02-03 Martin Liska <mliska@suse.cz>
25865 * multiple_target.c (create_dispatcher_calls): Redirect edge
25866 from a caller of a dispatcher.
25867 (expand_target_clones): Make the clones local.
25868 (ipa_target_clone): Do both target clones and resolvers.
25869 (ipa_dispatcher_calls): Remove the pass.
25870 (pass_dispatcher_calls::gate): Likewise.
25871 (make_pass_dispatcher_calls): Likewise.
25872 * passes.def (pass_target_clone): Put as very first IPA early
25875 2017-02-03 Martin Liska <mliska@suse.cz>
25877 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25878 in case of a function with ifunc attribute.
25880 2017-02-03 Martin Liska <mliska@suse.cz>
25882 * cgraph.c (cgraph_node::dump): Dump function version info.
25883 * symtab.c (symtab_node::dump_base): Add missing new line.
25885 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25887 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25888 (ifcombine_ifandif): Use it.
25890 2017-02-03 Martin Liska <mliska@suse.cz>
25892 * doc/invoke.texi: Document default value for
25893 use-after-scope-direct-emission-threshold.
25895 2017-02-03 Martin Liska <mliska@suse.cz>
25897 PR tree-optimization/79339
25898 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25899 (format_floating): Likewise.
25901 2017-02-03 Martin Liska <mliska@suse.cz>
25904 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25906 (ipa_node_params_t::remove): Likewise.
25907 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25908 initialization from removed ipa_node_params_t::insert.
25909 (ipa_node_params::~ipa_node_params): Move from removed
25910 ipa_node_params_t::release.
25911 * symbol-summary.h (symbol_summary::m_released): New member.
25912 Do not release a summary twice. Do not allow to call finalizer
25913 for types of a summary that live in GGC memory.
25915 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25917 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25920 2017-02-02 Martin Sebor <msebor@redhat.com>
25922 PR middle-end/79275
25923 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25924 (format_string): Tighten up the range of output for non-constant
25925 strings and correct the expected range for wide non-constant strings.
25927 2017-02-02 Martin Sebor <msebor@redhat.com>
25929 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25931 PR middle-end/32003
25932 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25934 (-fdump-tree-@var): Add to index and document how to come up
25935 with pass-specific option and dump file names.
25936 (-fdump-passes): Clarify where to look for output.
25938 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25940 PR middle-end/77445
25941 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25942 statistics of the analyzed path; allow threading for speed when
25943 any of BBs along the path are optimized for speed.
25945 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25947 PR middle-end/78468
25948 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25949 settings of the virtual registers.
25952 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25954 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25955 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25958 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25960 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25961 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25963 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25965 * config/s390/s390.md: Add missing comments with the expanded
25967 * config/s390/vector.md: Likewise.
25968 * config/s390/vx-builtins.md: Likewise.
25970 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25973 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25974 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25975 conditions on a single line.
25977 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25979 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25980 __S390_VX__ to __VX__.
25982 2017-02-01 Andrew Pinski <apinski@cavium.com>
25984 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25985 stmt_info to record_stmt_cost.
25986 (vect_get_known_peeling_cost): Pass stmt_info if known to
25988 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25989 cpu_vector_cost field into
25990 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25991 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25992 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25993 splitting of scalar_stmt_cost and vec_stmt_cost.
25994 (thunderx_vector_cost): Likewise.
25995 (cortexa57_vector_cost): LIkewise.
25996 (exynosm1_vector_cost): Likewise.
25997 (xgene1_vector_cost): Likewise.
25998 (thunderx2t99_vector_cost): Improve after the splitting of the two
26000 (aarch64_builtin_vectorization_cost): Update for the splitting of
26001 scalar_stmt_cost and vec_stmt_cost.
26003 2017-02-01 Torvald Riegel <triegel@redhat.com>
26004 Richard Henderson <rth@redhat.com>
26006 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26007 conditional on existance of a fast atomic load.
26008 * optabs-query.c (can_atomic_load_p): New function.
26009 * optabs-query.h (can_atomic_load_p): Declare it.
26010 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26011 no fast atomic load is available for the particular size of access.
26012 (expand_atomic_compare_and_swap): Likewise.
26013 (expand_atomic_load): Likewise.
26014 (expand_atomic_store): Likewise.
26015 (expand_atomic_fetch_op): Likewise.
26016 * testsuite/lib/target-supports.exp
26017 (check_effective_target_sync_int_128): Remove x86 because it provides
26018 no fast atomic load.
26019 (check_effective_target_sync_int_128_runtime): Likewise.
26021 2017-02-01 Richard Biener <rguenther@suse.de>
26023 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26024 (graphite_transform_loops): Provide opt-info for optimized nests.
26025 * tree-parloop.c (parallelize_loops): Provide opt-info for
26026 parallelized loops.
26028 2017-02-01 Richard Biener <rguenther@suse.de>
26030 PR middle-end/79315
26031 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26032 was not set before.
26034 2017-02-01 Richard Biener <rguenther@suse.de>
26036 PR tree-optimization/71824
26037 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26038 Verify the loops are valid in the merged SESE region.
26039 (scop_detection::can_represent_loop_1): Check analyzing the
26040 evolution of the number of iterations in the region succeeds.
26042 2017-01-31 Ian Lance Taylor <iant@golang.org>
26044 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26045 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26047 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26049 PR preprocessor/79210
26050 * input.c (get_substring_ranges_for_loc): Replace line_width
26051 assertion with error-handling.
26053 2017-01-31 Richard Biener <rguenther@suse.de>
26055 PR tree-optimization/77318
26056 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26057 (create_pw_aff_from_tree): Take loop parameter.
26058 (add_condition_to_pbb): Pass loop of the condition to
26059 create_pw_aff_from_tree.
26061 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26063 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26064 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26066 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26070 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26072 (convert_int_to_float128): Likewise.
26073 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26074 (convert_int_to_float128): Likewise.
26075 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26076 (UNSPEC_IEEE128_CONVERT): Likewise.
26077 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26078 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26079 Use local variables for IBM extended format.
26080 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26081 (fix_trunc<mode>si2_fprs): Likewise.
26082 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26083 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26084 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26085 to know that we can now have integers of all sizes in vector
26087 (fix<uns>_<mode>di2_hw): Likewise.
26088 (float<uns>_<mode>si2_hw): Likewise.
26089 (fix_<mode>si2_hw): Likewise.
26090 (fixuns_<mode>si2_hw): Likewise.
26091 (float<uns>_<mode>di2_hw): Likewise.
26092 (float_<mode>di2_hw): Likewise.
26093 (float_<mode>si2_hw): Likewise.
26094 (floatuns_<mode>di2_hw): Likewise.
26095 (floatuns_<mode>si2_hw): Likewise.
26096 (xscvqp<su>wz_<mode>): Delete, no longer used.
26097 (xscvqp<su>dz_<mode>): Likewise.
26098 (xscv<su>dqp_<mode>): Likewise.
26099 (ieee128_mfvsrd_64bit): Likewise.
26100 (ieee128_mfvsrd_32bit): Likewise.
26101 (ieee128_mfvsrwz): Likewise.
26102 (ieee128_mtvsrw): Likewise.
26103 (ieee128_mtvsrd_64bit): Likewise.
26104 (ieee128_mtvsrd_32bit): Likewise.
26106 2017-01-31 Martin Liska <mliska@suse.cz>
26109 * ipa-prop.c (ipa_free_all_node_params): Call release method
26110 instead of ~sumbol_summary to not to trigger double times
26113 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26115 PR tree-optimization/71691
26116 * bitmap.h (class auto_bitmap): New.
26117 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26118 is_maybe_undefined instead of ssa_undefined_value_p.
26120 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26122 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26123 __S390_ARCH_LEVEL__ to __ARCH__.
26125 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26127 PR tree-optimization/79267
26128 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26129 if should_remove_lhs_p is true.
26131 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26134 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26135 (add_alignment_attribute): New.
26136 (base_type_die): Add alignment attribute.
26137 (subrange_type_die): Likewise.
26138 (modified_type_die): Likewise.
26139 (gen_array_type_die): Likewise.
26140 (gen_descr_array_type_die: Likewise.
26141 (gen_enumeration_type_die): Likewise.
26142 (gen_subprogram_die): Likewise.
26143 (gen_variable_die): Likewise.
26144 (gen_field_die): Likewise.
26145 (gen_ptr_to_mbr_type_die): Likewise.
26146 (gen_struct_or_union_type_die): Likewise.
26147 (gen_subroutine_type_die): Likewise.
26148 (gen_typedef_die): Likewise.
26149 (base_type_cmp): Compare alignment attribute.
26151 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26154 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26155 (setb_unsigned) New pattern for setb with CCUNS.
26156 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26157 subfc./subfe sequence to avoid overflow problems. Generate a
26158 shorter sequence with cmpld/setb for power9.
26159 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26160 for generating subfc. instruction.
26161 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26162 now uses this instruction.
26164 2017-01-30 Ian Lance Taylor <iant@google.com>
26167 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26168 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26170 2017-01-30 Martin Sebor <msebor@redhat.com>
26172 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26173 Move constant to the right of a relational operator.
26174 (get_mpfr_format_length, format_character, format_string): Ditto.
26175 (should_warn_p, maybe_warn): Same.
26177 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26179 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26182 * asan.c (get_translation_unit_decl): Remove function.
26183 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26185 2017-01-30 Martin Liska <mliska@suse.cz>
26187 PR gcov-profile/79259
26188 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26189 -fprofile-generate.
26191 2017-01-30 Martin Liska <mliska@suse.cz>
26194 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26195 Initialize variables with NULL value.
26197 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26200 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26202 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26204 2017-01-30 Richard Biener <rguenther@suse.de>
26206 PR tree-optimization/79276
26207 * tree-vrp.c (process_assert_insertions): Properly adjust common
26208 when removing a duplicate.
26210 * gcc.dg/torture/pr79276.c: New testcase.
26212 2017-01-30 Richard Biener <rguenther@suse.de>
26214 PR tree-optimization/79256
26215 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26216 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26218 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26220 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26223 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26224 ("*r<noxa>sbg_<mode>_sll_bitmask")
26225 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26226 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26227 Use contiguous_bitmask_nowrap_operand.
26229 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26232 * config/rs6000/altivec.h (vec_xl): Revise #define.
26233 (vec_xst): Likewise.
26235 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26237 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26239 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26241 PR rtl-optimization/79194
26242 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26243 traps before call to bypass_conditional_jumps.
26245 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26247 PR tree-optimization/71374
26248 * lra-constraints.c (check_conflict_input_operands): New.
26249 (match_reload): Use it.
26251 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26254 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26255 account to calculate conflict_set.
26257 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26259 PR rtl-optimization/78559
26260 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26261 other_insn in combine.
26263 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26265 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26266 uint16_type_node for BT_UINT16.
26268 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26270 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26271 "RTL Tests" to menu.
26272 (GIMPLE Tests): New node.
26273 (RTL Tests): New node.
26275 2017-01-27 Richard Biener <rguenther@suse.de>
26277 PR tree-optimization/79245
26278 * tree-loop-distribution.c (distribute_loop): Apply cost
26279 modeling also to detected patterns.
26281 2017-01-27 Richard Biener <rguenther@suse.de>
26283 PR tree-optimization/71433
26284 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26285 (compare_assert_loc): New function.
26286 (process_assert_insertions): Sort and optimize assert locations
26287 to remove duplicates and push down identical assertions on
26288 edges to their destination block.
26290 2017-01-27 Richard Biener <rguenther@suse.de>
26292 PR tree-optimization/79244
26293 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26294 out SSA names even if abnormal.
26296 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26298 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26299 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26300 instead of MPFR_RNDN.
26302 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26305 * arm.c (arm_option_override): Don't call build_target_option_node
26306 until after doing all option overrides.
26307 (arm_valid_target_attribute_tree): Likewise.
26309 2017-01-27 Martin Liska <mliska@suse.cz>
26311 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26312 for {cd}tors and C++ {cd}tors.
26314 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26316 * config/s390/s390.md ("*setmem_long_and")
26317 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26319 2017-01-26 Martin Sebor <msebor@redhat.com>
26321 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26324 2017-01-26 Martin Sebor <msebor@redhat.com>
26326 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26327 HAVE_DFmode before using XFmode or DFmode.
26328 (parse_directive): Avoid using the z length modifier to avoid
26329 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26331 PR middle-end/78703
26332 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26333 to accept adjustment as an array.
26334 (get_int_range): New function.
26335 (struct directive): Make width and prec arrays.
26336 (directive::set_width, directive::set_precision): Call get_int_range.
26337 (format_integer, format_floating): Handle width and precision ranges.
26338 (format_string, parse_directive): Same.
26340 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26343 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26344 comdat_type_p set, just clone them, but keep the children in the
26348 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26349 which have direct callers with -fvar-tracking-assignments enabled
26351 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26352 inside of type units.
26354 2017-01-26 Martin Sebor <msebor@redhat.com>
26356 PR middle-end/78703
26357 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26359 (struct format_result): Replace number_chars, number_chars_min,
26360 and number_chars_max with a single member of struct result_range.
26362 (format_result::operator+=): Adjust.
26363 (struct fmtresult): Remove bounded. Handle likely and unlikely
26365 (fmtresult::adjust_for_width_or_precision): New function.
26366 (fmtresult:type_max_digits): New function.
26367 (bytes_remaining): Handle likely and unlikely counters.
26368 (min_bytes_remaining): Remove.
26369 (format_percent): Simplify.
26370 (format_integer, format_floating): Set likely and unlikely counters.
26371 (get_string_length, format_character, format_string): Same.
26372 (format_plain, should_warn_p): New function.
26373 (maybe_warn): Call should_warn_p. Update diagnostic messages
26374 and handle those for all directives, including plain strings.
26375 (format_directive): Handle likely and unlikely counters.
26376 Remove unnecessary quoting from diagnostics. Add an informational
26378 (add_bytes): Remove.
26379 (pass_sprintf_length::compute_format_length): Simplify.
26380 (try_substitute_return_value): Handle likely and unlikely counters.
26382 2017-01-26 Carl Love <cel@us.ibm.com>
26384 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26385 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26387 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26390 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26391 endianess for subregs into account.
26392 * lra-constraints.c (lra_constraints): Do risky transformations
26393 always on the first iteration.
26394 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26395 last_call_used_reg_set.
26396 (process_bb_lives): Define and use last_call_used_reg_set.
26397 * lra.c (lra): Always continue after lra_constraints on the first
26400 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26402 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26404 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26406 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26408 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26409 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26410 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26411 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26412 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26413 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26414 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26415 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26416 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26418 2017-01-26 Marek Polacek <polacek@redhat.com>
26421 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26422 for the third operand.
26424 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26426 PR middle-end/79236
26427 * omp-low.c (struct omp_context): Add simt_stmt field.
26428 (scan_omp_for): Return omp_context *.
26429 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26430 context to the _simt_ SIMD stmt.
26431 (lower_omp_for): For combined SIMD with sibling _simt_
26432 SIMD, make sure to use the same decls in _looptemp_
26433 clauses as in the sibling.
26435 2017-01-26 David Sherwood <david.sherwood@arm.com>
26437 PR middle-end/79212
26438 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26441 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26444 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26445 emit fld b; fld a; if possible.
26447 * brig-builtins.def: Update copyright years.
26448 * config/arm/arm_acle_builtins.def: Update copyright years.
26450 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26453 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26454 constraint instead of o for the stxsd instruction.
26456 2017-01-25 Carl Love <cel@us.ibm.com>
26458 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26459 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26461 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26463 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26465 2017-01-25 Richard Biener <rguenther@suse.de>
26467 PR tree-optimization/69264
26468 * target.def (vector_alignment_reachable): Improve documentation.
26469 * doc/tm.texi: Regenerate.
26470 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26472 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26473 earlier changes with respect to TYPE_USER_ALIGN.
26474 (vector_alignment_reachable_p): Likewise. Improve dumping.
26476 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26479 * config/arm/arm.md (xordi3): Force constant operand into a register
26482 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26484 * doc/invoke.texi (-fstore-merging): Correct default optimization
26485 levels at which it is enabled.
26486 (-O): Move -fstore-merging from list to...
26489 2017-01-25 Richard Biener <rguenther@suse.de>
26492 * omp-expand.c: Include debug.h.
26493 (expand_omp_taskreg): Make sure to generate early debug before
26494 outlining anything from a function.
26495 (expand_omp_target): Likewise.
26496 (grid_expand_target_grid_body): Likewise.
26498 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26501 * asan.c (get_translation_unit_decl): New function.
26502 (asan_add_global): Extract modules file name from globals
26503 TRANSLATION_UNIT_DECL name.
26505 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26508 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26509 for long calls with APCS frame and VFP.
26511 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26513 * cfg.c (original_copy_tables_initialized_p): New function.
26514 * cfg.h (original_copy_tables_initialized_p): New decl.
26515 * cfgrtl.c (relink_block_chain): Guard the call to
26516 free_original_copy_tables with a call to
26517 original_copy_tables_initialized_p.
26518 * cgraph.h (symtab_node::native_rtl_p): New decl.
26519 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26520 (symtab_node::needed_p): Don't assert for early assembly output
26521 for __RTL functions.
26522 (cgraph_node::finalize_function): Set "force_output" for __RTL
26524 (cgraph_node::analyze): Bail out early for __RTL functions.
26525 (analyze_functions): Update assertion to support __RTL functions.
26526 (cgraph_node::expand): Bail out early for __RTL functions.
26527 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26529 * function.h (struct function): Update comment for field
26531 * gimple-expr.c: Include "tree-pass.h".
26532 (gimple_has_body_p): Return false for __RTL functions.
26533 * Makefile.in (OBJS): Add run-rtl-passes.o.
26534 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26536 (gcc::pass_manager::get_clean_slate): New accessor.
26537 * passes.c: Include "insn-addr.h".
26538 (should_skip_pass_p): Add logging. Update logic for running
26539 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26540 property-provider override so it is only done for gimple passes.
26542 (skip_pass): New function.
26543 (execute_one_pass): Call skip_pass when skipping passes.
26544 * read-md.c (md_reader::read_char): Support filtering
26545 the input to a subset of line numbers.
26546 (md_reader::md_reader): Initialize fields
26547 m_first_line and m_last_line.
26548 (md_reader::read_file_fragment): New function.
26549 * read-md.h (md_reader::read_file_fragment): New decl.
26550 (md_reader::m_first_line): New field.
26551 (md_reader::m_last_line): New field.
26552 * read-rtl-function.c (function_reader::create_function): Only
26553 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26554 curr_properties. Set DECL_INITIAL to a dummy block.
26555 (read_rtl_function_body_from_file_range): New function.
26556 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26558 * run-rtl-passes.c: New file.
26559 * run-rtl-passes.h: New file.
26561 2017-01-24 Jeff Law <law@redhat.com>
26563 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26566 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26568 PR tree-optimization/79159
26569 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26570 (record_nonwrapping_iv): Improve boundary using above function if no
26571 value range information.
26573 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26574 Martin Jambor <mjambor@suse.cz>
26576 * brig-builtins.def: New file.
26577 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26578 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26579 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26580 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26581 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26582 * builtin-types.def (BT_INT8): New.
26583 (BT_INT16): Likewise.
26584 (BT_UINT8): Likewise.
26585 (BT_UINT16): Likewise.
26586 (BT_FN_ULONG): Likewise.
26587 (BT_FN_UINT_INT): Likewise.
26588 (BT_FN_UINT_ULONG): Likewise.
26589 (BT_FN_UINT_LONG): Likewise.
26590 (BT_FN_UINT_PTR): Likewise.
26591 (BT_FN_ULONG_PTR): Likewise.
26592 (BT_FN_INT8_FLOAT): Likewise.
26593 (BT_FN_INT16_FLOAT): Likewise.
26594 (BT_FN_UINT32_FLOAT): Likewise.
26595 (BT_FN_UINT16_FLOAT): Likewise.
26596 (BT_FN_UINT8_FLOAT): Likewise.
26597 (BT_FN_UINT64_FLOAT): Likewise.
26598 (BT_FN_UINT16_UINT32): Likewise.
26599 (BT_FN_UINT32_UINT16): Likewise.
26600 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26601 (BT_FN_INT_PTR_INT): Likewise.
26602 (BT_FN_UINT_PTR_UINT): Likewise.
26603 (BT_FN_LONG_PTR_LONG): Likewise.
26604 (BT_FN_ULONG_PTR_ULONG): Likewise.
26605 (BT_FN_VOID_UINT64_UINT64): Likewise.
26606 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26607 (BT_FN_INT8_INT8_INT8): Likewise.
26608 (BT_FN_INT16_INT16_INT16): Likewise.
26609 (BT_FN_INT_INT_INT): Likewise.
26610 (BT_FN_UINT_FLOAT_UINT): Likewise.
26611 (BT_FN_FLOAT_UINT_UINT): Likewise.
26612 (BT_FN_ULONG_UINT_UINT): Likewise.
26613 (BT_FN_ULONG_UINT_PTR): Likewise.
26614 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26615 (BT_FN_UINT_UINT_UINT): Likewise.
26616 (BT_FN_VOID_UINT_PTR): Likewise.
26617 (BT_FN_UINT_UINT_PTR: Likewise.
26618 (BT_FN_UINT32_UINT64_PTR): Likewise.
26619 (BT_FN_INT_INT_UINT_UINT): Likewise.
26620 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26621 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26622 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26623 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26624 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26625 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26626 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26627 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26628 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26629 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26630 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26631 * doc/frontends.texi: List BRIG FE.
26632 * doc/install.texi (Testing): Add BRIG tesring requirements.
26633 * doc/invoke.texi (Overall Options): Mention BRIG.
26634 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26636 2017-01-24 Richard Biener <rguenther@suse.de>
26638 PR translation/79208
26639 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26641 2017-01-24 Martin Jambor <mjambor@suse.cz>
26644 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26645 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26646 and known_contexts.
26648 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26650 PR middle-end/79123
26651 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26652 casts from signed to unsigned really don't have a range.
26654 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26656 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26657 GMP_RNDx for compatiblity.
26659 2017-01-24 Martin Liska <mliska@suse.cz>
26662 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26663 that would prevent us to call alloca with -1 as argument.
26665 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26667 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26668 Avoid -Wformat-security warning.
26670 2017-01-23 Andrew Pinski <apinski@cavium.com>
26672 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26675 2017-01-23 Martin Sebor <msebor@redhat.com>
26677 PR middle-end/78703
26678 * gimple-ssa-sprintf.c (warn_level): New global.
26679 (format_integer): Use it here and throughout the rest of the file.
26680 Use the same switch to compute sign as base.
26681 (maybe_warn): New function.
26682 (format_directive): Factor out warnings into maybe_warn.
26683 Add debugging output. Use warn_level.
26684 (add_bytes): Use warn_level.
26685 (pass_sprintf_length::compute_format_length): Add debugging output.
26686 (try_substitute_return_value): Same.
26687 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26689 PR middle-end/78703
26690 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26691 (struct fmtresult, format_integer, format_floating): Adjust.
26692 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26693 (get_string_length, format_string,format_directive): Same.
26694 (pass_sprintf_length::compute_format_length): Same.
26695 (try_substitute_return_value): Simplify slightly.
26697 PR middle-end/78703
26698 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26699 (fmtresult::operator+=): Outlined.
26700 (struct fmtresult): Add ctors.
26701 (struct conversion_spec): Rename...
26702 (struct directive): ...to this. Add and remove data members.
26703 (directive::set_width, directive::set_precision): New functions.
26704 (format_percent): Use fmtresult ctor.
26705 (get_width_and_precision): Remove.
26706 (format_integer): Make naming changes. Avoid computing width and
26708 (format_floating): Same. Adjust indentation.
26709 (format_character, format_none): New functions.
26710 (format_string): Moved character handling to format_character.
26711 (format_directive): Remove arguments, change return type.
26712 (parse_directive): New function.
26713 (pass_sprintf_length::compute_format_length): Move directive
26714 parsing to parse_directive.
26716 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26718 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26719 (assign_assembler_name_if_needed): ... this.
26720 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26721 (assign_assembler_name_if_needed): ... this.
26722 (free_lang_data_in_cgraph): Adjust callers.
26723 * cgraphunit.c (cgraph_node::analyze): Likewise.
26724 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26727 2017-01-23 Richard Biener <rguenther@suse.de>
26729 PR tree-optimization/79088
26730 PR tree-optimization/79188
26731 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26732 resetting loop bounds after last path deletion. Reset loop
26733 bounds of the target loop, make code match the comments.
26734 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26735 Make sure loops need no fixups.
26737 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26739 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26740 exponent support with double type for first argument.
26741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26742 type returned by __builtin_vec_extract_sig,
26743 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26744 functions from "vector int" to "vector unsigned int" or from
26745 "vector long long int" to "vector unsigned long long int".
26746 Changed type returned by __builtin_vec_extract_exp,
26747 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26748 functions from "vector int" to "vector unsigned int" or from
26749 "vector long long int" to "vector unsigned long long int".
26750 Changed return type of __builtin_vec_test_data_class,
26751 __builtin_vec_test_data_class_sp, and
26752 __builtin_vec_test_data_class_dp from "vector int" to
26753 "vector bool int" or from "vector long long int" to "vector bool
26754 long long int" and changed second argument type from "unsigned
26755 int" to "int". Added new overloaded function forms "vector float
26756 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26757 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26758 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26759 double, vector unsigned long long int)" and "vector double
26760 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26761 long int)". Changed return type of
26762 __builtin_scalar_test_data_class and
26763 __builtin_scalar_test_data_class_sp and
26764 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26765 int" and changed second argument from "unsigned int" to "int".
26766 Changed type returned by __builtin_scalar_test_neg,
26767 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26768 from "int" to "bool int". Added new overloaded function form
26769 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26770 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26771 exponent double-precision with floating point first argument.
26772 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26773 documentation of scalar_test_data_class, scalar_test_neg,
26774 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26775 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26776 vec_test_data_class built-in functions to reflect refinements in
26777 their type signatures.
26779 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26781 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26783 (aarch64_elf_asm_destructor): Likewise.
26785 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26787 PR rtl-optimization/78634
26788 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26789 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26790 * ifcvt.c (noce_try_cmove): Add missing cost check.
26792 PR rtl-optimization/71724
26793 * combine.c (if_then_else_cond): Look for situations where it is
26794 beneficial to undo the work of one of the recursive calls.
26796 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26798 PR tree-optimization/70754
26799 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26800 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26801 combined stmt before it if not NULL.
26802 (combine_chains): Process refs reversely and compute dominance point
26805 2017-01-23 Martin Liska <mliska@suse.cz>
26807 PR tree-optimization/79196
26808 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26809 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26811 (strlen_optimize_stmt): Call the renamed function.
26813 2017-01-23 Michael Matz <matz@suse.de>
26815 PR tree-optimization/78384
26816 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26818 2017-01-23 Richard Biener <rguenther@suse.de>
26820 PR tree-optimization/79186
26821 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26822 both incoming edges before moving an assert.
26824 2017-01-23 Martin Jambor <mjambor@suse.cz>
26826 * ipa-prop.c (load_from_param_1): Removed.
26827 (load_from_unmodified_param): Bits from load_from_param_1 put back
26829 (load_from_param): Removed.
26830 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26831 with stmt. Reverted back to use of load_from_unmodified_param.
26833 2017-01-23 Martin Jambor <mjambor@suse.cz>
26836 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26837 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26838 field a pointer to garbage collected vector, mark lattices and
26839 ipcp_orig_node with GTY((skip)).
26840 (ipa_get_param_count): Adjust to descriptors being a pointer.
26841 (ipa_get_param): Likewise.
26842 (ipa_get_type): Likewise.
26843 (ipa_get_param_move_cost): Likewise.
26844 (ipa_set_param_used): Likewise.
26845 (ipa_get_controlled_uses): Likewise.
26846 (ipa_set_controlled_uses): Likewise.
26847 (ipa_is_param_used): Likewise.
26848 (ipa_node_params_t): Move into garbage collector. New methods insert
26850 (ipa_node_params_sum): Annotate wth GTY(()).
26851 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26853 (ipa_load_from_parm_agg): Adjust declaration.
26854 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26855 * ipa-profile.c (ipa_profile): Likewise.
26856 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26857 (ipa_populate_param_decls): Make descriptors parameter garbage
26859 (ipa_dump_param): Adjust to descriptors being a pointer.
26860 (ipa_alloc_node_params): Likewise.
26861 (ipa_initialize_node_params): Likewise.
26862 (load_from_param_1): Make descriptors parameter garbage collected.
26863 (load_from_unmodified_param): Likewise.
26864 (load_from_param): Likewise.
26865 (ipa_load_from_parm_agg): Likewise.
26866 (ipa_node_params::~ipa_node_params): Removed.
26867 (ipa_free_all_node_params): Remove call to delete operator.
26868 (ipa_node_params_t::insert): New.
26869 (ipa_node_params_t::remove): Likewise.
26870 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26871 copy known_csts and known_contexts vectors.
26872 (ipa_read_node_info): Adjust to descriptors being a pointer.
26873 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26875 (ipcp_transform_function): Make descriptors variable garbage
26878 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26880 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26881 * config/i386/avx512dqintrin.h: Ditto.
26882 * config/i386/avx512fintrin.h: Ditto.
26883 * config/i386/i386.c: Handle new builtins.
26884 * config/i386/i386-builtin.def: Add new builtins.
26885 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26886 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26888 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26889 Martin Liska <mliska@suse.cz>
26891 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26892 * asan.c (asan_expand_poison_ifn): Support stores and use
26893 appropriate ASAN report function.
26894 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26895 * internal-fn.def (ASAN_POISON_USE): Declare.
26896 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26897 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26898 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26899 ASAN_POISON calls w/o LHS.
26900 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26901 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26902 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26903 * gimplify.c (asan_poison_variables): Add attribute
26904 use_after_scope_memory to variables that really needs to live
26906 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26907 having the attribute.
26909 2017-01-23 Martin Liska <mliska@suse.cz>
26911 * asan.c (create_asan_shadow_var): New function.
26912 (asan_expand_poison_ifn): Likewise.
26913 * asan.h (asan_expand_poison_ifn): New declaration.
26914 * internal-fn.c (expand_ASAN_POISON): Likewise.
26915 * internal-fn.def (ASAN_POISON): New builtin.
26916 * sanopt.c (pass_sanopt::execute): Expand
26917 asan_expand_poison_ifn.
26918 * tree-inline.c (copy_decl_for_dup_finish): Make function
26920 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26921 * tree-ssa.c (is_asan_mark_p): New function.
26922 (execute_update_addresses_taken): Rewrite local variables
26923 (identified just by use-after-scope as addressable) into SSA.
26925 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26927 * doc/install.texi (Specific): opensource.apple.com uses https
26928 now. Remove trailing slash.
26930 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26932 * README.Portability: Remove note on an Irix compatibility issue.
26934 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26936 * gcov.c (INCLUDE_ALGORITHM): Define.
26937 (INCLUDE_VECTOR): Define.
26938 No longer include <vector> and <algorithm> directly.
26940 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26942 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26944 * doc/invoke.texi (Code Gen Options): Ditto.
26946 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26949 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26951 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26953 rtl-optimization/79125
26954 * cprop.c (local_cprop_pass): Handle cases where we make an
26955 unconditional trap.
26957 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26961 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26962 read from, for big endian.
26964 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26966 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26967 register pauth builtins for LP64 only.
26969 2017-01-20 Marek Polacek <polacek@redhat.com>
26972 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26975 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26977 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26979 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26980 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26981 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26983 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26986 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26987 in big-endian mode when they are not a single duplicated value.
26989 2017-01-20 Richard Biener <rguenther@suse.de>
26991 * BASE-VER: Bump to 7.0.1.
26993 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26995 * omp-low.c (omplow_simd_context): New struct. Use it...
26996 (lower_rec_simd_input_clauses): ...here and...
26997 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26998 references to idx, lane, max_vf, is_simt.
27000 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27002 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27005 2017-01-20 Martin Jambor <mjambor@suse.cz>
27007 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27008 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27010 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27011 (GTFILES): Rename hsa.c to hsa-common.c.
27012 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27013 * hsa-dump.c: Likewise.
27014 * hsa-gen.c: Likewise.
27015 * hsa-regalloc.c: Likewise.
27016 * ipa-hsa.c: Likewise.
27017 * omp-expand.c: Likewise.
27018 * omp-low.c: Likewise.
27019 * toplev.c: Likewise.
27021 2017-01-20 Marek Polacek <polacek@redhat.com>
27024 * doc/invoke.texi: Document -Wduplicated-branches.
27025 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27026 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27027 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27028 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27029 return 0 only when not OEP_LEXICOGRAPHIC.
27030 (fold_build_cleanup_point_expr): Use the expression
27031 location when building CLEANUP_POINT_EXPR.
27032 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27033 * tree.c (add_expr): Handle error_mark_node.
27035 2017-01-20 Martin Liska <mliska@suse.cz>
27038 * tree-profile.c (init_ic_make_global_vars): Do not call
27040 (gimple_init_gcov_profiler): Likewise.
27042 2017-01-20 Martin Liska <mliska@suse.cz>
27045 * cgraph.h (maybe_create_reference): Remove argument and
27047 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27049 * ipa-cp.c (create_specialized_node): Likewise.
27050 * symtab.c (symtab_node::maybe_create_reference): Handle
27051 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27053 2017-01-20 Martin Liska <mliska@suse.cz>
27055 * read-rtl-function.c (function_reader::create_function): Use
27056 build_decl instread of build_decl_stat.
27058 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27060 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27061 * config/i386/avx512dqintrin.h: Ditto.
27062 * config/i386/avx512fintrin.h: Ditto.
27063 * config/i386/i386-builtin-types.def: Add new types.
27064 * config/i386/i386.c: Handle new types.
27065 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27066 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27067 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27068 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27069 (__builtin_ia32_kshiftridi): New.
27070 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27072 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27076 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27077 define to rs6000_init_stack_protect_guard.
27078 (rs6000_init_stack_protect_guard): New function.
27080 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27081 Yunqiang Su <yunqiang.su@imgtec.com>
27083 * config.gcc (supported_defaults): Add madd4.
27084 (with_madd4): Add validation.
27085 (all_defaults): Add madd4.
27086 * config/mips/mips.opt (mmadd4): New option.
27087 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27089 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27091 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27092 (ISA_HAS_FUSED_MADD4): Likewise.
27093 * doc/invoke.texi (-mmadd4): Document the new option.
27094 * doc/install.texi (--with-madd4): Document the new option.
27096 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27098 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27099 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27100 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27101 (aarch64_init_pauth_hint_builtins): New.
27102 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27103 (aarch64_expand_builtin): Expand new builtins.
27105 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27107 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27108 * combine-stack-adj.c (no_unhandled_cfa): Handle
27109 REG_CFA_TOGGLE_RA_MANGLE.
27110 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27111 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27112 info for return address signing.
27113 (aarch64_expand_epilogue): Likewise.
27115 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27117 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27118 * config/aarch64/aarch64-protos.h
27119 (aarch64_return_address_signing_enabled): New declaration.
27120 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27122 (aarch64_expand_prologue): Sign return address before it's pushed onto
27124 (aarch64_expand_epilogue): Authenticate return address fetched from
27126 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27127 (aarch64_attributes): New function attributes for "sign-return-address".
27128 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27129 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27130 ("*do_return"): Generate combined instructions according to key index.
27131 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27132 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27134 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27135 * config/aarch64/aarch64.opt (msign-return-address=): New.
27136 * doc/extend.texi (AArch64 Function Attributes): Documents
27137 "sign-return-address=".
27138 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27140 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27142 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27143 overall option summary.
27145 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27147 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27148 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27149 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27150 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27152 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27154 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27155 -mpower9-minmax by default for -mcpu=power9.
27156 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27157 128-bit floating point.
27159 2017-01-20 Alan Modra <amodra@gmail.com>
27161 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27162 optimizing for size.
27164 2017-01-20 Alan Modra <amodra@gmail.com>
27167 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27168 for strcmp and strncmp from corresponding builtin decl.
27170 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27172 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27173 instead of i386/rtems-64.h.
27174 * config/i386/rtems-64.h: Remove.
27176 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27180 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27182 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27184 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27186 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27187 Change int to HOST_WIDE_INT.
27188 * config/aarch64/aarch64-protos.h
27189 (aarch64_simd_gen_const_vector_dup): Likewise.
27190 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27192 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27194 * langhooks-def.h (lhd_type_for_size): New decl.
27195 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27196 * langhooks.c (lhd_type_for_size): New function, taken from
27199 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27201 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27202 define_bypass for CR latency.
27203 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27204 (power9-alu2): Add define_bypass for CR latency.
27206 (power9-mul): Update insn latency.
27207 (power9-mul-compare): Update insn latency, bypass latency and remove
27210 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27212 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27214 * config/aarch64/aarch64.md
27215 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27216 aarch64_nopcrelative_literal_loads.
27217 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27219 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27221 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27222 TARGET_LOONGSON_3A.
27223 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27225 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27228 * config.gcc (supported_defaults): Add lxc1-sxc1.
27229 (with_lxc1_sxc1): Add validation.
27230 (all_defaults): Add lxc1-sxc1.
27231 * config/mips/mips.opt (mlxc1-sxc1): New option.
27232 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27234 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27235 __mips_no_lxc1_sxc1.
27236 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27237 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27238 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27240 2017-01-19 Richard Biener <rguenther@suse.de>
27242 PR tree-optimization/72488
27243 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27244 sure to restore SSA info.
27245 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27247 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27249 PR rtl-optimization/79121
27250 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27251 of the inner type when shifting an extended value.
27253 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27256 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27257 interposable aliases.
27259 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27262 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27263 Use the evmergelohi instruction.
27264 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27265 (mov_sitf_e500_subreg8_2_be): Likewise.
27266 (mov_sitf_e500_subreg12_2_le): Likewise.
27267 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27268 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27269 (mov_sitf_e500_subreg8_2_le): Likewise.
27270 (mov_sitf_e500_subreg12_2_be): Likewise.
27272 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27274 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27275 attribute from vecsimple to vecperm.
27276 (altivec_vbpermq2): Likewise.
27278 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27281 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27283 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27284 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27285 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27286 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27287 case where N arg is SIZE_MAX.
27288 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27289 (cmpstrsi): Add pattern.
27291 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27293 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27294 __builtin_vec_revb builtins.
27295 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27296 built-in functions to support generation of the ISA 3.0 XXBR<x>
27297 vector byte reverse instructions.
27298 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27299 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27300 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27301 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27302 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27303 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27304 (P9V_BUILTIN_VEC_REVB): Likewise.
27305 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27306 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27307 (p9_xxbrq_v16qi): Likewise.
27308 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27309 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27310 (p9_xxbrh_v8hi): Likewise.
27311 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27312 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27313 vec_revb built-in functions.
27315 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27317 PR rtl-optimization/78952
27318 * config/i386/i386.md (any_extract): New code iterator.
27319 (*insvqi_2): Use any_extract for source operand.
27320 (*insvqi_3): Use any_shiftrt for source operand.
27322 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27324 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27326 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27328 2017-01-18 Matthias Klose <doko@ubuntu.com>
27330 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27332 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27334 * config/rs6000/altivec.h (vec_bperm): Change #define.
27335 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27336 (altivec_vbpermq2): New define_insn.
27337 (altivec_vbpermd): Likewise.
27338 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27339 function interface.
27340 (VBPERMD): Likewise.
27341 (VBPERM): New polymorphic function interface.
27342 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27343 Add entries for P9V_BUILTIN_VEC_VBPERM.
27344 * doc/extend.texi: Add interfaces for vec_bperm.
27346 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27348 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27349 first letter of error messages.
27350 (s390_resolve_overloaded_builtin): Likewise.
27351 * config/s390/s390.c (s390_expand_builtin): Likewise.
27352 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27353 (s390_valid_target_attribute_inner_p): Likewise.
27354 * config/s390/s390.md ("tabort"): Likewise.
27356 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27358 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27359 (ISA_AVOID_DIV_HILO): New macro.
27360 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27361 (ISA_HAS_DDIV): Likewise.
27363 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27365 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27367 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27369 * doc/invoke.texi (fabi-version): Spelling fix.
27371 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27374 * doc/invoke.texi (fabi-version): Mention mangling fix for
27377 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27380 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27382 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27385 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27386 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27388 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27390 (stack_protect_test): Ditto.
27391 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27392 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27394 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27395 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27396 -mstack-protector-guard-offset=.
27397 (RS/6000 and PowerPC Options): Ditto.
27399 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27401 * config/i386/i386.h (MASK_CLASS_P): New define.
27402 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27403 there are no registers from different register sets also when
27404 mask registers are used. Update function comment.
27405 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27406 to (*k/*r) and (*k/*km) alternatives.
27408 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27410 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27411 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27412 (EH_RETURN_HANDLER_RTX): New define.
27413 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27414 Force frame pointer in EH return functions.
27415 (aarch64_expand_epilogue): Add barrier for eh_return.
27416 (aarch64_final_eh_return_addr): Remove.
27417 (aarch64_eh_return_handler_rtx): New function.
27418 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27420 (aarch64_eh_return_handler_rtx): New prototype.
27422 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27424 * config/rs6000/altivec.h (vec_rlmi): New #define.
27425 (vec_vrlnm): Likewise.
27426 (vec_rlnm): Likewise.
27427 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27428 (UNSPEC_VRLNM): Likewise.
27429 (VIlong): New mode iterator.
27430 (altivec_vrl<VI_char>mi): New define_insn.
27431 (altivec_vrl<VI_char>nm): Likewise.
27432 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27434 (VRLDNM): Likewise.
27435 (RLNM): New polymorphic function entry.
27436 (VRLWMI): New monomorphic function entry.
27437 (VRLDMI): Likewise.
27438 (RLMI): New polymorphic function entry.
27439 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27440 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27441 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27444 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27447 * dwarf2out.c (field_byte_offset): Restore the
27448 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27449 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27450 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27453 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27456 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27458 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27461 * dwarf2out.c (add_data_member_location_attribute): For constant
27462 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27463 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27464 DW_AT_byte_size attributes.
27466 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27468 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27469 after forcing to constant memory when the code model is medium.
27471 2017-01-17 Julia Koval <julia.koval@intel.com>
27474 * config/i386/avx512fintrin.h
27475 (_mm512_i32gather_ps): Change __addr type to void const*.
27476 (_mm512_mask_i32gather_ps): Ditto.
27477 (_mm512_i32gather_pd): Ditto.
27478 (_mm512_mask_i32gather_pd): Ditto.
27479 (_mm512_i64gather_ps): Ditto.
27480 (_mm512_mask_i64gather_ps): Ditto.
27481 (_mm512_i64gather_pd): Ditto.
27482 (_mm512_mask_i64gather_pd): Ditto.
27483 (_mm512_i32gather_epi32): Ditto.
27484 (_mm512_mask_i32gather_epi32): Ditto.
27485 (_mm512_i32gather_epi64): Ditto.
27486 (_mm512_mask_i32gather_epi64): Ditto.
27487 (_mm512_i64gather_epi32): Ditto.
27488 (_mm512_mask_i64gather_epi32): Ditto.
27489 (_mm512_i64gather_epi64): Ditto.
27490 (_mm512_mask_i64gather_epi64): Ditto.
27491 (_mm512_i32scatter_ps): Change __addr type to void*.
27492 (_mm512_mask_i32scatter_ps): Ditto.
27493 (_mm512_i32scatter_pd): Ditto.
27494 (_mm512_mask_i32scatter_pd): Ditto.
27495 (_mm512_i64scatter_ps): Ditto.
27496 (_mm512_mask_i64scatter_ps): Ditto.
27497 (_mm512_i64scatter_pd): Ditto.
27498 (_mm512_mask_i64scatter_pd): Ditto.
27499 (_mm512_i32scatter_epi32): Ditto.
27500 (_mm512_mask_i32scatter_epi32): Ditto.
27501 (_mm512_i32scatter_epi64): Ditto.
27502 (_mm512_mask_i32scatter_epi64): Ditto.
27503 (_mm512_i64scatter_epi32): Ditto.
27504 (_mm512_mask_i64scatter_epi32): Ditto.
27505 (_mm512_i64scatter_epi64): Ditto.
27506 (_mm512_mask_i64scatter_epi64): Ditto.
27507 * config/i386/avx512pfintrin.h
27508 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27509 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27510 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27511 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27512 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27513 (_mm512_prefetch_i32scatter_ps): Ditto.
27514 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27515 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27516 (_mm512_prefetch_i64scatter_pd): Ditto.
27517 (_mm512_prefetch_i64scatter_ps): Ditto.
27518 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27519 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27520 * config/i386/avx512vlintrin.h
27521 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27522 (_mm_mmask_i32gather_ps): Ditto.
27523 (_mm256_mmask_i32gather_pd): Ditto.
27524 (_mm_mmask_i32gather_pd): Ditto.
27525 (_mm256_mmask_i64gather_ps): Ditto.
27526 (_mm_mmask_i64gather_ps): Ditto.
27527 (_mm256_mmask_i64gather_pd): Ditto.
27528 (_mm_mmask_i64gather_pd): Ditto.
27529 (_mm256_mmask_i32gather_epi32): Ditto.
27530 (_mm_mmask_i32gather_epi32): Ditto.
27531 (_mm256_mmask_i32gather_epi64): Ditto.
27532 (_mm_mmask_i32gather_epi64): Ditto.
27533 (_mm256_mmask_i64gather_epi32): Ditto.
27534 (_mm_mmask_i64gather_epi32): Ditto.
27535 (_mm256_mmask_i64gather_epi64): Ditto.
27536 (_mm_mmask_i64gather_epi64): Ditto.
27537 (_mm256_i32scatter_ps): Change __addr type to void*.
27538 (_mm256_mask_i32scatter_ps): Ditto.
27539 (_mm_i32scatter_ps): Ditto.
27540 (_mm_mask_i32scatter_ps): Ditto.
27541 (_mm256_i32scatter_pd): Ditto.
27542 (_mm256_mask_i32scatter_pd): Ditto.
27543 (_mm_i32scatter_pd): Ditto.
27544 (_mm_mask_i32scatter_pd): Ditto.
27545 (_mm256_i64scatter_ps): Ditto.
27546 (_mm256_mask_i64scatter_ps): Ditto.
27547 (_mm_i64scatter_ps): Ditto.
27548 (_mm_mask_i64scatter_ps): Ditto.
27549 (_mm256_i64scatter_pd): Ditto.
27550 (_mm256_mask_i64scatter_pd): Ditto.
27551 (_mm_i64scatter_pd): Ditto.
27552 (_mm_mask_i64scatter_pd): Ditto.
27553 (_mm256_i32scatter_epi32): Ditto.
27554 (_mm256_mask_i32scatter_epi32): Ditto.
27555 (_mm_i32scatter_epi32): Ditto.
27556 (_mm_mask_i32scatter_epi32): Ditto.
27557 (_mm256_i32scatter_epi64): Ditto.
27558 (_mm256_mask_i32scatter_epi64): Ditto.
27559 (_mm_i32scatter_epi64): Ditto.
27560 (_mm_mask_i32scatter_epi64): Ditto.
27561 (_mm256_i64scatter_epi32): Ditto.
27562 (_mm256_mask_i64scatter_epi32): Ditto.
27563 (_mm_i64scatter_epi32): Ditto.
27564 (_mm_mask_i64scatter_epi32): Ditto.
27565 (_mm256_i64scatter_epi64): Ditto.
27566 (_mm256_mask_i64scatter_epi64): Ditto.
27567 (_mm_i64scatter_epi64): Ditto.
27568 (_mm_mask_i64scatter_epi64): Ditto.
27569 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27570 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27571 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27572 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27573 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27574 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27575 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27576 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27577 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27578 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27579 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27580 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27581 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27582 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27583 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27584 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27585 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27586 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27587 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27588 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27589 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27590 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27591 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27592 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27593 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27594 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27595 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27596 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27597 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27598 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27599 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27600 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27601 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27602 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27603 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27604 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27605 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27606 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27607 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27608 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27609 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27610 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27611 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27612 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27613 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27614 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27615 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27616 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27617 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27618 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27619 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27620 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27621 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27622 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27623 definitions accordingly.
27625 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27626 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27629 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27632 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27635 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27636 hard regs for inner regno.
27638 2017-01-17 Martin Liska <mliska@suse.cz>
27641 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27642 assumption and add comment.
27644 2017-01-17 Nathan Sidwell <nathan@acm.org>
27646 * ipa-visibility.c (localize_node): New function, broken out of ...
27647 (function_and_variable_visibility): ... here. Call it.
27649 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27651 PR middle-end/77445
27652 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27653 correctly set frequency of oudgoing edge.
27654 (duplicate_thread_path): Fix profile updating.
27656 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27659 * configure.ac: Add GCC_BASE_VER.
27660 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27661 version from BASE-VER file.
27662 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27663 (gcc.o): Depend on $(BASEVER).
27664 * common.opt (dumpfullversion): New option.
27665 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27666 * doc/invoke.texi: Document -dumpfullversion.
27667 * doc/install.texi: Document --with-gcc-major-version-only.
27668 * configure: Regenerated.
27670 2017-01-17 Richard Biener <rguenther@suse.de>
27672 PR tree-optimization/71433
27673 * tree-vrp.c (register_new_assert_for): Merge same asserts
27674 on all incoming edges.
27675 (process_assert_insertions_for): Handle insertions at the
27678 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27680 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27681 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27683 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27686 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27689 2017-01-17 Alan Modra <amodra@gmail.com>
27692 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27693 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27694 symbolic stack limit when pic.
27696 2017-01-16 Martin Sebor <msebor@redhat.com>
27698 PR tree-optimization/78608
27699 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27701 2017-01-16 Jeff Law <law@redhat.com>
27704 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27705 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27706 for several include directories that may be relative to sysroot.
27707 * config/i386/x-mingw32 (gplus_includedir): Define.
27708 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27709 (native_system_includedir): Likewise.
27710 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27711 override if TARGET_SYSTEM_ROOT is defined.
27712 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27714 PR tree-optimization/79090
27715 PR tree-optimization/33562
27716 PR tree-optimization/61912
27717 PR tree-optimization/77485
27718 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27719 and computed trims into the dump file.
27721 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27723 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27725 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27728 * gimplify.c (gimplify_init_constructor): If want_value and
27729 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27733 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27734 sequence. Formatting fixes.
27735 (doloop_optimize): Formatting fixes.
27738 * gcc.c (debug_level_greater_than_spec_func): New function.
27739 (static_spec_functions): Add debug-level-gt spec function.
27740 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27742 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27743 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27744 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27745 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27746 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27747 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27749 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27751 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27752 QImode fixups to general and mask registers only.
27754 2017-01-16 Carl Love <cel@us.ibm.com>
27756 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27757 for built-in functions
27758 vector signed char vec_nabs (vector signed char)
27759 vector signed short vec_nabs (vector signed short)
27760 vector signed int vec_nabs (vector signed int)
27761 vector signed long long vec_nabs (vector signed long long)
27762 vector float vec_nabs (vector float)
27763 vector double vec_nabs (vector double)
27764 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27766 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27767 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27768 * doc/extend.texi: Update the documentation file for the new built-in
27771 2017-01-16 Martin Sebor <msebor@redhat.com>
27773 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27776 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27778 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27779 UNSPEC_VSX__XXSPLTD to require special splat handling.
27781 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27784 * system.h: Poison strndup.
27786 2017-01-16 Alan Modra <amodra@gmail.com>
27789 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27792 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27794 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27796 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27798 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27799 call recog here. Assert that INSN_CODE (insn) is non-negative.
27801 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27804 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27806 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27807 in the currently scheduled RTL fragment.
27809 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27811 PR rtl-optimization/78751
27812 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27815 2017-01-14 Jeff Law <law@redhat.com>
27817 PR tree-optimization/79090
27818 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27819 variable length stores.
27820 (compute_trims): Delete dead assignment to *trim_tail.
27821 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27824 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27826 PR rtl-optimization/78626
27827 PR rtl-optimization/78727
27828 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27829 of a block, and split such blocks after everything else is finished.
27831 2017-01-14 Alan Modra <amodra@gmail.com>
27834 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27835 target legitimate_combined_insn.
27836 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27837 (rs6000_legitimate_combined_insn): New function.
27838 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27840 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27841 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27842 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27844 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27846 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27848 2017-01-13 Jeff Law <law@redhat.com>
27850 PR tree-optimization/33562
27851 PR tree-optimization/61912
27852 PR tree-optimization/77485
27853 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27855 (delete_dead_assignment): Likewise.
27856 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27857 statement to delete_dead_call and delete_dead_assignment.
27859 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27862 * substring-locations.c (format_warning_va): Strengthen case 1 so
27863 that both endpoints of the substring must be within the format
27864 range for just the substring to be printed.
27866 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27868 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27869 * config/i386/i386.c (ix86_target_string): Add missing options
27870 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27871 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27872 flags_other and ix86_target_other to flags2_other. Display unknown
27874 (ix86_valid_target_attribute_inner_p): Add missing options and
27875 reorder options by implied ISAs, as in ix86_target_string.
27877 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27879 * hash-table.h (hash_table::too_empty_p): New function.
27880 (hash_table::expand): Use it.
27881 (hash_table::traverse): Likewise.
27882 (hash_table::empty_slot): Use sizeof (value_type) instead of
27883 sizeof (PTR) to convert bytes to elements. Shrink the table
27884 if the current size is excessive for the current number of
27887 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27889 * ira-costs.c (record_reg_classes): Break from the inner loop
27890 early once alt_fail is known to be true. Update outer loop
27891 handling accordingly.
27893 2017-01-13 Jeff Law <law@redhat.com>
27895 * tree-ssa-dse.c (decrement_count): New function.
27896 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27897 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27898 when we know the partially dead statement is a mem* function.
27900 PR tree-optimization/61912
27901 PR tree-optimization/77485
27902 * tree-ssa-dse.c: Include expr.h.
27903 (maybe_trim_constructor_store): New function.
27904 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27906 PR tree-optimization/33562
27907 PR tree-optimization/61912
27908 PR tree-optimization/77485
27909 * doc/invoke.texi: Document new dse-max-object-size param.
27910 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27911 * tree-ssa-dse.c: Include params.h.
27912 (dse_store_status): New enum.
27913 (initialize_ao_ref_for_dse): New, partially extracted from
27915 (valid_ao_ref_for_dse, normalize_ref): New.
27916 (setup_live_bytes_from_ref, compute_trims): Likewise.
27917 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27918 (maybe_trim_partially_dead_store): Likewise.
27919 (maybe_trim_complex_store): Likewise.
27920 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27921 Track what bytes live from the original store. Return tri-state
27922 for dead, partially dead or live.
27923 (dse_dom_walker): Add constructor, destructor and new private members.
27924 (delete_dead_call, delete_dead_assignment): New extracted from
27926 (dse_optimize_stmt): Make a member of dse_dom_walker.
27927 Use initialize_ao_ref_for_dse.
27929 PR tree-optimization/33562
27930 PR tree-optimization/61912
27931 PR tree-optimization/77485
27932 * sbitmap.h (bitmap_count_bits): Prototype.
27933 (bitmap_clear_range, bitmap_set_range): Likewise.
27934 * sbitmap.c (bitmap_clear_range): New function.
27935 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27937 2017-01-13 Martin Liska <mliska@suse.cz>
27940 * function.c (set_cfun): Add new argument force.
27941 * function.h (set_cfun): Likewise.
27942 * ipa-inline-transform.c (inline_call): Use the function when
27943 strict alising from is dropped for function we inline to.
27945 2017-01-13 Richard Biener <rguenther@suse.de>
27947 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27948 for dumping GIMPLE INTEGER_CSTs.
27950 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27952 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27953 to 201112L since C++17.
27955 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27958 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27959 if -fsanitize=kernel-address is present.
27961 2017-01-13 Richard Biener <rguenther@suse.de>
27963 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27964 as _Literal ( type ) number in case usual suffixes do not
27965 preserve all information.
27967 2017-01-13 Richard Biener <rguenther@suse.de>
27969 PR tree-optimization/77283
27970 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27971 and ssa-iterators.h.
27972 (is_feasible_trace): Implement a cost model based on joiner
27975 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27978 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27979 char or short to __float128/_Float128 directly.
27981 2017-01-12 Martin Sebor <msebor@redhat.com>
27983 to -Wformat-overflow.
27984 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27985 (min_bytes_remaining): Same.
27986 (get_string_length): Same.
27987 (format_string): Same.
27988 (format_directive): Same.
27990 (pass_sprintf_length::handle_gimple_call): Same.
27992 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27994 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27995 info.nowrite calls with no lhs that can't throw. Return bool
27996 whether gsi_remove has been called or not.
27997 (pass_sprintf_length::handle_gimple_call): Return bool whether
27998 try_substitute_return_value called gsi_remove. Formatting fix.
27999 (pass_sprintf_length::execute): Don't use gsi_remove if
28000 handle_gimple_call returned true.
28003 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28004 be removed due to side-effects, don't remove following barrier nor
28005 turn the successor edge into fallthru edge.
28007 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28010 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28011 element-reversing loads and stores as not swappable.
28013 2017-01-12 Nathan Sidwell <nathan@acm.org>
28014 Nicolai Stange <nicstange@gmail.com>
28016 * combine.c (try_combine): Don't ignore result of overlap checking
28017 loop. Combine overlap & asm check into single loop.
28019 2017-01-12 Richard Biener <rguenther@suse.de>
28021 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28022 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28024 2017-01-12 Richard Biener <rguenther@suse.de>
28026 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28027 and TS_TARGET_OPTION directly derive from TS_BASE.
28028 * tree-core.h (tree_optimization_option): Derive from tree_base.
28029 (tree_target_option): Likewise.
28031 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28033 * config/i386/i386.c (memory_address_length): Increase len
28034 only when rip_relative_addr_p returns false.
28036 2017-01-11 Julia Koval <julia.koval@intel.com>
28038 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28039 (OPTION_MASK_ISA_SGX_SET): New.
28040 (ix86_handle_option): Handle OPT_msgx.
28041 * config.gcc: Added sgxintrin.h.
28042 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28043 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28044 * config/i386/i386.c (ix86_target_string): Add -msgx.
28046 (ix86_option_override_internal): Handle new options.
28047 (ix86_valid_target_attribute_inner_p): Add sgx.
28048 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28049 * config/i386/i386.opt: Add msgx.
28050 * config/i386/sgxintrin.h: New file.
28051 * config/i386/x86intrin.h: Add sgxintrin.h.
28053 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28056 * fold-const.c (maybe_nonzero_address): Return 1 for function
28058 (tree_single_nonzero_warnv_p): Don't handle function local objects
28062 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28065 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28068 * opts.c: Include "spellcheck.h"
28069 (struct string_fragment): New struct.
28070 (struct edit_distance_traits<const string_fragment &>): New
28072 (get_closest_sanitizer_option): New function.
28073 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28075 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28077 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28079 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28080 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28081 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28082 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28083 for initial die_offset if dwarf_split_debug_info.
28084 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28085 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28087 (output_skeleton_debug_sections): Formatting fix. Use
28088 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28089 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28091 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28093 * config/arm/cortex-a53.md: Add bypasses for
28094 cortex_a53_r2f_cvt.
28095 (cortex_a53_r2f): Only use for transfers.
28096 (cortex_a53_f2r): Likewise.
28097 (cortex_a53_r2f_cvt): Add reservation for conversions.
28098 (cortex_a53_f2r_cvt): Likewise.
28100 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28102 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28103 to all inlined functions, change static to extern.
28105 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28108 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28110 (arm_assemble_integer): Likewise.
28112 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28114 * config.gcc: Use new awk script to check CPU, FPU and architecture
28115 parameters for --with-... options.
28116 * config/arm/parsecpu.awk: New file
28117 * config/arm/arm-cpus.in: New file.
28118 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28120 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28122 * config/arm/t-arm: Update dependency rules.
28123 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28124 of processing .def files.
28125 * config/arm/genopt.sh: Deleted.
28126 * config/arm/gentune.sh: Deleted.
28127 * config/arm/arm-cores.def: Deleted.
28128 * config/arm/arm-arches.def: Deleted.
28129 * config/arm/arm-fpus.def: Deleted.
28130 * config/arm/arm-tune.md: Regenerated.
28131 * config/arm/arm-tables.opt: Regenerated.
28132 * config/arm/arm-cpu.h: New generated file.
28133 * config/arm/arm-cpu-data.h: New generated file.
28134 * config/arm/arm-cpu-cdata.h: New generated file.
28136 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28139 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28141 (input_varpool_node): Unpack dynamically_initialized bit.
28143 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28145 PR rtl-optimization/79032
28146 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28147 the alignment of the adjusted memory reference against that of MODE,
28148 instead of the alignment of the original memory reference.
28150 2017-01-11 Martin Jambor <mjambor@suse.cz>
28152 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28154 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28155 decorated functions.
28157 2017-01-11 Richard Biener <rguenther@suse.de>
28159 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28160 set range/nonnull info for PHI results. Do not set it on
28161 stmts marked for removal.
28163 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28165 * expr.c (store_field): In the bitfield case, fetch the return value
28166 from the registers before applying a single big-endian adjustment.
28167 Always do a final load for a BLKmode value not larger than a word.
28169 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28172 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28173 that we correctly handle column numbers greater than
28174 LINE_MAP_MAX_COLUMN_NUMBER.
28176 2017-01-10 Martin Sebor <msebor@redhat.com>
28178 PR middle-end/78245
28179 * gimple-ssa-sprintf.c (get_destination_size): Call
28180 {init,fini}object_sizes.
28181 * tree-object-size.c (addr_object_size): Adjust.
28182 (pass_through_call): Adjust.
28183 (pass_object_sizes::execute): Adjust.
28184 * tree-object-size.h (fini_object_sizes): Declare.
28186 2017-01-10 Martin Sebor <msebor@redhat.com>
28188 PR tree-optimization/78775
28189 * builtins.c (get_size_range): Move...
28190 * calls.c: ...to here.
28191 (alloc_max_size): Accept zero argument.
28192 (operand_signed_p): Remove.
28193 (maybe_warn_alloc_args_overflow): Call get_size_range.
28194 * calls.h (get_size_range): Declare.
28196 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28198 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28199 from TI's devices.csv file as of September 2016.
28200 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28202 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28204 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28205 * doc/invoke.texi: Likewise.
28206 * doc/md.texi: Likewise.
28207 * doc/objc.texi: Likewise.
28209 2017-01-10 Joshua Conner <joshconner@google.com>
28211 * config/arm/fuchsia-elf.h: New file.
28212 * config/fuchsia.h: New file.
28213 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28214 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28216 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28218 2016-01-10 Richard Biener <rguenther@suse.de>
28220 PR tree-optimization/79034
28221 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28222 Propagate out degenerate PHIs in the joiner.
28224 2017-01-10 Martin Liska <mliska@suse.cz>
28226 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28227 (sort_congruence_classes_by_decl_uid): Likewise.
28228 (sort_congruence_class_groups_by_decl_uid): Likewise.
28229 (sem_item_optimizer::merge_classes): Sort class, groups in these
28230 classes and members in the groups by DECL_UID of declarations.
28231 This would make merge operations stable.
28233 2017-01-10 Martin Liska <mliska@suse.cz>
28235 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28236 usage of m_classes_vec.
28237 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28238 (sem_item_optimizer::get_group_by_hash): Likewise.
28239 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28240 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28241 (sem_item_optimizer::verify_classes): Likewise.
28242 (sem_item_optimizer::process_cong_reduction): Likewise.
28243 (sem_item_optimizer::dump_cong_classes): Likewise.
28244 (sem_item_optimizer::merge_classes): Likewise.
28245 * ipa-icf.h (congruence_class_hash): Rename from
28246 congruence_class_group_hash. Remove declaration of m_classes_vec.
28248 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28250 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28251 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28252 * config.gcc: Add avx512vpopcntdqintrin.h.
28253 * config/i386/avx512vpopcntdqintrin.h: New.
28254 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28255 * config/i386/i386-builtin-types.def: Add new types.
28256 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28257 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28258 __builtin_ia32_vpopcountq_v8di_mask): New.
28259 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28260 __AVX512VPOPCNTDQ__.
28261 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28262 (PTA_AVX512VPOPCNTDQ): Define.
28263 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28264 TARGET_AVX512VPOPCNTDQ_P): Define.
28265 * config/i386/i386.opt: Add mavx512vpopcntdq.
28266 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28267 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28269 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28271 PR middle-end/77484
28272 * predict.def (PRED_CALL): Set to 67.
28274 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28276 * expr.c (store_field): In the bitfield case, if the value comes from
28277 a function call and is of an aggregate type returned in registers, do
28278 not modify the field mode; extract the value in all cases if the mode
28279 is BLKmode and the size is not larger than a word.
28281 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28284 * config/i386/cpuid.h: Fix undefined behavior.
28286 2017-01-04 Jeff Law <law@redhat.com>
28288 PR tree-optimization/79007
28289 PR tree-optimization/67955
28290 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28291 conservative for pt.null when flag_non_call_exceptions is on.
28293 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28295 PR translation/79019
28296 PR translation/79020
28297 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28298 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28299 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28301 * config/avr/avr.opt (maccumulate-args): Likewise.
28302 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28303 * common.opt (freport-bug): Likewise.
28304 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28305 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28306 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28307 translatable string.
28308 * config/i386/i386.c (function_value_32): Likewise.
28309 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28310 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28312 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28313 * common/config/msp430/msp430-common.c (msp430_handle_option):
28315 * symtab.c (symtab_node::verify_base): Likewise.
28316 * opts.c (set_debug_level): Likewise.
28317 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28318 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28319 missing whitespace to translatable strings.
28320 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28321 * config/sh/superh.h: Likewise.
28322 * config/i386/xopintrin.h: Likewise.
28323 * config/i386/znver1.md: Likewise.
28324 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28325 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28326 * double-int.h (struct double_int): Likewise.
28327 * double-int.c (div_and_round_double): Likewise.
28328 * wide-int.cc: Likewise.
28329 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28330 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28331 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28332 (crossjumps_occurred): ... this.
28333 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28336 PR tree-optimization/78899
28337 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28338 returning bool return struct loop *, NULL for failure and the new
28340 (versionable_outer_loop_p): Don't version outer loop if it has
28341 dont_vectorized bit set.
28342 (tree_if_conversion): When versioning outer loop, ensure
28343 tree_if_conversion is performed also on the inner loop of the
28344 non-vectorizable outer loop copy.
28345 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28346 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28347 prevent vectorization of it.
28348 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28349 the outer loop vectorization of the non-scalar version is attempted
28350 before vectorization of the inner loop in scalar version. If
28351 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28352 vectorization of its inner loop.
28353 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28354 has 2 inner loops, rename also on edges from bb whose single pred
28355 is outer_loop->header. Fix typo in function comment.
28357 2017-01-09 Martin Sebor <msebor@redhat.com>
28360 * asan.c (asan_emit_stack_protection): Increase local buffer size
28361 to avoid snprintf truncation warning.
28363 2017-01-09 Andrew Pinski <apinski@cavium.com>
28365 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28366 to reference thunderx2t99 for the tuning structure
28367 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28369 (thunderx2t99_extra_costs): This.
28370 * config/aarch64/aarch64-tune.md: Regenerate.
28371 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28372 (vulcan_addrcost_table): This.
28373 (vulcan_regmove_cost): Rename to ...
28374 (thunderx2t99_regmove_cost): This.
28375 (vulcan_vector_cost): Rename to ...
28376 (thunderx2t99_vector_cost): this.
28377 (vulcan_branch_cost): Rename to ...
28378 (thunderx2t99_branch_cost): This.
28379 (vulcan_tunings): Rename to ...
28380 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28381 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28383 2017-01-09 Martin Jambor <mjambor@suse.cz>
28387 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28388 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28389 (propagate_vr_accross_jump_function): Use the above function for all
28390 value range computations for pass-through jump functions and type
28391 converasion from explicit value range values.
28392 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28393 parameters from TYPE_ARG_TYPES.
28394 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28395 (ipa_write_node_info): Stream type of the actual argument.
28396 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28398 2017-01-09 Martin Liska <mliska@suse.cz>
28401 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28402 (lookup_compiler): Do not show error message with have_E.
28404 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28406 PR tree-optimization/78938
28407 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28408 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28409 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28410 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28413 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28415 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28418 2017-01-09 Richard Biener <rguenther@suse.de>
28420 PR tree-optimization/78997
28421 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28422 name condition properly.
28424 2017-01-09 Richard Biener <rguenther@suse.de>
28427 * dwarf2out.c (is_cxx): New overload with context.
28428 (is_naming_typedef_decl): Use it.
28430 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28432 * invoke.texi (Option Summary): Correct spacing in option lists
28433 and add line breaks to fix over-long lines.
28435 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28437 PR middle-end/17660
28439 * extend.texi (Common Variable Attributes): Add xref to GCC
28440 Internals manual to explain mode attribute keywords.
28442 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28445 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28446 and Preprocessor Options.
28447 (Options for Linking): Document -pthread here....
28448 (RS/6000 and PowerPC Options): ...not here.
28449 (Solaris 2 Options): ...or here.
28450 * doc/cppopts.texi: Document -pthread.
28452 2017-01-08 Martin Sebor <msebor@redhat.com>
28454 PR middle-end/77708
28455 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28456 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28457 New member functions.
28458 (format_directive): Used them.
28460 (pass_sprintf_length::handle_gimple_call): Same.
28461 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28462 to avoid truncation for any argument.
28463 (extract_affine_mul): Same.
28464 * tree.c (get_file_function_name): Same.
28466 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28468 PR middle-end/77484
28469 * predict.def (PRED_INDIR_CALL): Set to 86.
28471 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28473 PR preprocessor/54124
28474 * doc/cppopts.texi: Reformat -d subtable to list the full name
28475 of the options. Add cross-reference to the docs for the general
28476 compiler -d options.
28477 * doc/invoke.texi (Developer Options): Add cross-reference to the
28478 preprocessor-specific -d option documentation.
28480 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28482 PR preprocessor/13498
28483 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28484 redudant material, and reflect new command-line options.
28485 (System Headers): Likewise.
28487 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28489 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28490 -isystem, and -idirafter. Copy-edit.
28491 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28492 default for -ftrack-macro-expansion. Delete obsolete and
28493 badly-formatted implementation details about -fdebug-cpp output.
28494 * doc/cppwarnopts.texi: Copy-edit.
28496 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28499 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28500 that the transition from a max line width >= 1<<10 to narrower
28501 lines works correctly.
28503 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28505 * doc/options.texi (PerFunction): New.
28506 * opt-functions.awk (switch_flags): Map both Optimization and
28507 PerFunction to CL_OPTIMIZATION.
28508 * opth-gen.awk: Test for PerFunction flag along with
28510 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28511 it only when the latter is present. Skip those that don't in
28512 the hash function generator.
28513 * common.opt (fvar-tracking): Mark as PerFunction instead of
28515 (fvar-tracking-assignments): Likewise.
28516 (fvar-tracking-assignments-toggle): Likewise.
28517 (fvar-tracking-uninit): Likewise.
28519 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28521 PR translation/79018
28522 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28525 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28528 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28529 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28530 TARGET_LONG_JUMP_TABLE_OFFSETS.
28531 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28532 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28533 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28534 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28535 * config/m68k/m68k.md (tablejump expander): Likewise.
28536 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28537 TARGET_LONG_JUMP_TABLE_OFFSETS.
28538 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28539 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28541 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28542 David Holsgrove <david.holsgrove@xilinx.com>
28544 * common/config/microblaze/microblaze-common.c
28545 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28546 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28548 * config/microblaze/microblaze.c (microblaze_must_save_register)
28549 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28551 (microblaze_eh_return): New function.
28552 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28553 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28554 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28555 * config/microblaze/microblaze.md (eh_return): New pattern.
28557 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28559 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28560 GCC_DIAGNOSTIC_STRINGIFY): Define.
28562 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28564 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28566 * config/arm/arm.md (<mcrr>): New.
28568 * config/arm/arm.c (arm_arch5te): New.
28569 (arm_option_override): Set arm_arch5te.
28570 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28572 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28573 (arm_mcrr_qualifiers): ... this. New.
28574 (MRRC_QUALIFIERS): Define to...
28575 (arm_mrrc_qualifiers): ... this. New.
28576 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28578 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28579 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28580 (MRRCI, mrrc, MRRC): New.
28581 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28582 VUNSPEC_MRRC2): New.
28584 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28586 * config/arm/arm.md (<mcr>): New.
28588 * config/arm/arm.c (arm_coproc_builtin_available): Add
28589 support for mcr, mrc, mcr2 and mrc2.
28590 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28591 (arm_mcr_qualifiers): ... this. New.
28592 (MRC_QUALIFIERS): Define to ...
28593 (arm_mrc_qualifiers): ... this. New.
28594 (MCR_QUALIFIERS): Define to ...
28595 (arm_mcr_qualifiers): ... this. New.
28596 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28598 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28599 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28600 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28601 VUNSPEC_MRC2): New.
28603 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28605 * config/arm/arm.md (*ldc): New.
28609 * config/arm/arm.c (arm_coproc_builtin_available): Add
28610 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28611 (arm_coproc_ldc_stc_legitimate_address): New.
28612 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28613 'qualifier_const_pointer'.
28614 (LDC_QUALIFIERS): Define to...
28615 (arm_ldc_qualifiers): ... this. New.
28616 (STC_QUALIFIERS): Define to...
28617 (arm_stc_qualifiers): ... this. New.
28618 * config/arm/arm-protos.h
28619 (arm_coproc_ldc_stc_legitimate_address): New.
28620 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28621 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28622 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28623 stc2, stcl, stc2l): New.
28624 * config/arm/constraints.md (Uz): New.
28625 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28626 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28627 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28628 VUNSPEC_STC2L): New.
28630 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28632 * config/arm/arm.md (<cdp>): New.
28633 * config/arm/arm.c (neon_const_bounds): Rename this ...
28634 (arm_const_bounds): ... this.
28635 (arm_coproc_builtin_available): New.
28636 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28637 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28638 (CDP_QUALIFIERS): Define to...
28639 (arm_cdp_qualifiers): ... this. New.
28641 (arm_expand_builtin_args): Add case for 6 arguments.
28642 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28643 (arm_const_bounds): ... this.
28644 (arm_coproc_builtin_available): New.
28645 * config/arm/arm_acle.h (__arm_cdp): New.
28647 * config/arm/arm_acle_builtins.def (cdp): New.
28649 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28650 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28651 'arm_const_bounds'.
28652 * config/arm/types.md (coproc): New.
28653 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28654 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28655 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28656 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28658 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28660 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28661 (UBINOP_QUALIFIERS): New.
28663 (acle_builtin_data): New. Change comment.
28664 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28665 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28666 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28667 arm_acle_builtins.def.
28668 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28669 (arm_init_acle_builtins): New.
28670 (CRC32_BUILTIN): Remove.
28671 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28672 crc32cb, crc32ch and crc32cw.
28673 (arm_init_crc32_builtins): Remove.
28674 (arm_init_builtins): Use arm_init_acle_builtins rather
28675 than arm_init_crc32_builtins.
28676 (arm_expand_acle_builtin): New.
28677 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28678 * config/arm/arm_acle_builtins.def: New.
28680 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28682 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28683 (arm_builtin_datum): ... this.
28684 (arm_init_neon_builtin): Rename to ...
28685 (arm_init_builtin): ... this. Add a new parameters PREFIX
28686 and USE_SIG_IN_NAME.
28687 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28688 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28689 'arm_builtin_datum'.
28690 (arm_init_vfp_builtins): Likewise.
28691 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28692 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28693 (arm_expand_neon_args): Rename to ...
28694 (arm_expand_builtin_args): ... this. Rename builtin_arg
28695 enum values and differentiate between ARG_BUILTIN_MEMORY
28696 and ARG_BUILTIN_NEON_MEMORY.
28697 (arm_expand_neon_builtin_1): Rename to ...
28698 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28699 values, arm_expand_builtin_args and add bool parameter NEON.
28700 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28701 (arm_expand_vfp_builtin): Likewise.
28702 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28704 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28706 PR middle-end/77484
28707 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28708 * predict.c (tree_estimate_probability_bb): Reverse direction of
28709 polymorphic call predictor.
28711 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28713 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28714 (determine_pass_name_match): ...this new function and...
28715 (should_skip_pass_p): ...this new function.
28717 2017-01-06 Nathan Sidwell <nathan@acm.org>
28719 * ipa-visibility.c (function_and_variable_visibility): Reformat
28720 comments and long lines. Remove extrneous if.
28721 * symtab.c (symtab_node::make_decl_local): Fix code format.
28722 (symtab_node::set_section_for_node): Fix comment typo.
28724 2017-01-06 Martin Liska <mliska@suse.cz>
28727 * lra-constraints.c: Rename invariant to lra_invariant.
28728 * predict.c (set_even_probabilities): Initialize e to NULL.
28730 2017-01-05 Martin Sebor <msebor@redhat.com>
28732 PR tree-optimization/78910
28733 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28734 (format_integer): Correct off-by-one error in the handling
28735 of precision with negative numbers in signed conversions..
28737 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28739 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28741 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28743 PR tree-optimization/71016
28744 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28745 factor_out_conditional_conversion. Formatting fix.
28746 (factor_out_conditional_conversion): Add cond_stmt argument.
28747 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28748 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28751 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28753 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28754 read-rtl-function.o, and selftest-rtl.o.
28755 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28756 (selftest::aarch64_test_loading_full_dump): New function.
28757 (selftest::aarch64_run_selftests): New function.
28758 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28759 selftest::aarch64_run_selftests.
28760 * config/i386/i386.c
28761 (selftest::ix86_test_loading_dump_fragment_1): New function.
28762 (selftest::ix86_test_loading_call_insn): New function.
28763 (selftest::ix86_test_loading_full_dump): New function.
28764 (selftest::ix86_test_loading_unspec): New function.
28765 (selftest::ix86_run_selftests): Call the new functions.
28766 * emit-rtl.c (maybe_set_max_label_num): New function.
28767 * emit-rtl.h (maybe_set_max_label_num): New decl.
28768 * function.c (instantiate_decls): Guard call to
28769 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28770 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28772 * gensupport.c (gen_reader::gen_reader): Pass "false"
28773 for new "compact" param of rtx_reader.
28774 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28775 rather than an empty string for NULL strings.
28776 * read-md.c: Potentially include config.h rather than bconfig.h.
28777 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28778 (have_error): New global, copied from errors.c.
28779 (md_reader::read_name): Rename to...
28780 (md_reader::read_name_1): ...this, adding "out_loc" param,
28781 and converting "missing name or number" to returning false, rather
28783 (md_reader::read_name): Reimplement in terms of read_name_1.
28784 (md_reader::read_name_or_nil): New function.
28785 (md_reader::read_string): Handle "(nil)" by returning NULL.
28786 (md_reader::md_reader): Add new param "compact".
28787 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28788 (md_reader::read_file): New method.
28789 * read-md.h (md_reader::md_reader): Add new param "compact".
28790 (md_reader::read_file): New method.
28791 (md_reader::is_compact): New accessor.
28792 (md_reader::read_name): Convert return type from void to file_location.
28793 (md_reader::read_name_or_nil): New decl.
28794 (md_reader::read_name_1): New decl.
28795 (md_reader::m_compact): New field.
28796 (noop_reader::noop_reader): Pass "false" for new "compact" param
28798 (rtx_reader::rtx_reader): Add new "compact" param.
28799 (rtx_reader::read_rtx_operand): Make virtual and convert return
28800 type from void to rtx.
28801 (rtx_reader::read_until): New decl.
28802 (rtx_reader::handle_any_trailing_information): New virtual function.
28803 (rtx_reader::postprocess): New virtual function.
28804 (rtx_reader::finalize_string): New virtual function.
28805 (rtx_reader::m_in_call_function_usage): New field.
28806 (rtx_reader::m_reuse_rtx_by_id): New field.
28807 * read-rtl-function.c: New file.
28808 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28809 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28810 (selftest::verify_three_block_rtl_cfg): New decl.
28811 * read-rtl-function.h: New file.
28812 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28813 For host, include function.h, memmodel.h, and emit-rtl.h.
28814 (one_time_initialization): New function.
28815 (struct compact_insn_name): New struct.
28816 (compact_insn_names): New array.
28817 (find_code): Handle insn codes in compact dumps.
28818 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28819 (bind_subst_iter_and_attr): Likewise.
28820 (add_condition_to_string): Likewise.
28821 (add_condition_to_rtx): Likewise.
28822 (apply_attribute_uses): Likewise.
28823 (add_current_iterators): Likewise.
28824 (apply_iterators): Likewise.
28825 (initialize_iterators): Guard usage of apply_subst_iterator with
28826 #ifdef GENERATOR_FILE.
28827 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28828 (md_reader::read_mapping): Likewise.
28829 (add_define_attr_for_define_subst): Likewise.
28830 (add_define_subst_attr): Likewise.
28831 (read_subst_mapping): Likewise.
28832 (check_code_iterator): Likewise.
28833 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28835 (one_time_initialization): New function.
28836 (rtx_reader::read_until): New method.
28837 (read_flags): New function.
28838 (parse_reg_note_name): New function.
28839 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28840 Handle reuse_rtx ids.
28841 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28842 Add parsing support for RTL dumps, mirroring the special-cases in
28843 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28844 values, and calling handle_any_trailing_information.
28845 (rtx_reader::read_rtx_operand): Convert return type from void
28846 to rtx, returning return_rtx. Handle case 'e'. Call
28847 finalize_string on XSTR and XTMPL fields.
28848 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28849 "(nil)" values were omitted. Call the postprocess vfunc on the
28851 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28852 class ctor. Initialize m_in_call_function_usage. Call
28853 one_time_initialization.
28854 * rtl-tests.c (selftest::test_uncond_jump): Call
28855 set_new_first_and_last_insn.
28856 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28857 * selftest-rtl.c: New file.
28858 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28859 (selftest::get_insn_by_uid): New decl.
28860 * selftest-run-tests.c (selftest::run_tests): Call
28861 read_rtl_function_c_tests.
28862 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28863 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28866 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28868 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28869 operands in a special way. Assert that pos+len <= mode precision.
28871 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28873 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28874 3 argument Alias with unlimited for the negative form.
28875 (fno-vect-cost-model): Removed.
28877 2017-01-05 Martin Liska <mliska@suse.cz>
28879 * hsa-gen.c (gen_hsa_divmod): New function.
28880 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28882 2017-01-05 Martin Liska <mliska@suse.cz>
28885 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28888 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28890 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28891 small constant length operands.
28893 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28895 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28896 between loop iterations.
28898 2017-01-05 Martin Liska <mliska@suse.cz>
28901 * gimplify.c (gimplify_decl_expr): Compare to
28902 asan_poisoned_variables instread of checking flags.
28903 (gimplify_target_expr): Likewise.
28904 (gimplify_expr): Likewise.
28905 (gimplify_function_tree): Conditionally initialize
28906 asan_poisoned_variables.
28908 2017-01-04 Jeff Law <law@redhat.com>
28910 PR tree-optimizatin/78812
28911 * rtl.h (contains_mem_rtx_p): Prototype.
28912 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28913 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28914 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28915 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28916 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28918 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28920 * input.c (assert_char_at_range): Default-initialize actual_range.
28922 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28924 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28925 to match the caller.
28927 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28929 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28930 insns after final jump in test to emit dummy move.
28932 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28934 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28935 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28937 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28939 * multiple_target.c (create_dispatcher_calls): Init e_next.
28940 * tree-ssa-loop-split.c (split_loop): Init border.
28941 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28944 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28949 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28950 (altivec_register_operand): Do not return true if the operand
28951 contains a SUBREG mixing SImode and SFmode.
28952 (vsx_register_operand): Likewise.
28953 (vsx_reg_sfsubreg_ok): New predicate.
28954 (vfloat_operand): Do not return true if the operand contains a
28955 SUBREG mixing SImode and SFmode.
28956 (vint_operand): Likewise.
28957 (vlogical_operand): Likewise.
28958 (gpc_reg_operand): Likewise.
28959 (int_reg_operand): Likewise.
28960 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28961 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28962 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28964 (rs6000_emit_move_si_sf_subreg): New helper function.
28965 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28966 fixup SUBREGs involving SImode and SFmode.
28967 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28968 numbers for the new peephole2 optimization.
28969 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28970 the GLIBC math library that do AND/IOR/XOR operations on single
28971 precision floating point.
28972 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28973 target macros to say whether we need to avoid SUBREGs mixing
28975 (TARGET_ALLOW_SF_SUBREG): Likewise.
28976 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28977 (UNSPEC_SI_FROM_SF): Likewise.
28978 (iorxor): Change spacing.
28979 (and_ior_xor): New iterator for AND, IOR, and XOR.
28980 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28981 (movdi_from_sf_zero_ext): Likewise.
28982 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28983 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28984 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28985 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28986 (fms<mode>4): Likewise.
28987 (fnma<mode>4): Likewise.
28988 (fnms<mode>4): Likewise.
28989 (nfma<mode>4): Likewise.
28990 (nfms<mode>4): Likewise.
28992 2017-01-04 Marek Polacek <polacek@redhat.com>
28995 * doc/invoke.texi: Document -Wpointer-compare.
28997 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28999 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29002 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29003 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29006 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29009 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29010 documentation of the powerpc_popcntb_ok attribute.
29011 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29012 code to issue warning messages if a requested CPU configuration is
29013 not supported by the binary (assembler and loader) toolchain.
29014 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29015 made to define a built-in function that has been disabled.
29016 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29017 made to define a built-in function that has been disabled.
29018 (altivec_init_builtins): Add comment explaining why definition
29019 of the DST built-in functions is not preceded by an assertion
29020 check. Add assertions to prevent ICE if attempts are made to
29021 define an altivec predicate or an abs* built-in function that has
29023 (htm_init_builtins): Add comment explaining why definition of the
29024 htm built-in functions is not preceded by an assertion check.
29026 2017-01-04 Jeff Law <law@redhat.com>
29028 PR tree-optimizatin/67955
29029 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29030 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29031 the points-to solution does not include pt_null. Use DECL_PT_UID
29034 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29036 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29037 Use gen_int_mode instead of gen_lopwart for const_int operands.
29039 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29041 PR tree-optimization/71563
29042 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29043 out of range value - has low bits known to be zero.
29045 2017-01-04 Alan Modra <amodra@gmail.com>
29047 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29048 * configure: Regenerate.
29049 * config.in: Regenerate.
29051 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29054 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29055 a substring of the message, but strcmp with the whole message. Ifdef
29056 ENABLE_NLS, translate the message first using dgettext.
29058 2017-01-03 Jeff Law <law@redhat.com>
29060 PR tree-optimizatin/78856
29061 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29062 (mark_threaded_blocks): Remove code to truncate thread paths that
29063 cross multiple loop headers. Instead invalidate the cached loop
29064 iteration information and handle case of a thread path walking
29065 into an irreducible region.
29067 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29070 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29071 assertions. Add support for doing the signbit if the IEEE 128-bit
29072 floating point value is in a GPR.
29073 * config/rs6000/rs6000.md (Fsignbit): Delete.
29074 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29075 Update the length attribute if the value is in a GPR.
29076 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29077 the sign or zero extension instruction, since the value is always 0/1.
29078 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29081 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29082 extracting SImode to a GPR register so that we can generate a
29083 store, limit the vector to be in a traditional Altivec register
29084 for the vextuwrx instruction.
29086 2017-01-03 Ian Lance Taylor <iant@google.com>
29088 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29090 2017-01-03 Martin Sebor <msebor@redhat.com>
29092 PR tree-optimization/78696
29093 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29094 precision. Use MPFR for %f for greater fidelity. Correct handling
29096 (pass_sprintf_length::compute_format_length): Set width and precision
29097 specified by asrerisk to void_node for vararg functions.
29098 (try_substitute_return_value): Adjust dump output.
29100 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29102 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29104 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29106 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29107 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29108 -mlra/-mno-lra was passed to the compiler.
29110 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29112 PR rtl-optimization/65618
29113 * emit-rtl.c (try_split): Move initialization of "before" and
29114 "after" to just before the call to emit_insn_after_setloc.
29116 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29118 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29120 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29122 * dwarf2out.c (gen_enumeration_type_die): When
29123 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29125 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29127 PR tree-optimization/78965
29128 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29129 Change first argument from const call_info & to call_info &. For %n
29130 set info.nowrite to false.
29132 PR middle-end/78901
29133 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29134 possibly throwing calls.
29136 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29137 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29138 and fns handling, rather than in a separate case SSA_NAME.
29140 2017-01-02 Jeff Law <law@redhat.com>
29142 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29143 fixes for first_period and second_period variables.
29145 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29148 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29149 (*insvqi_1): New insn pattern.
29150 (*insvqi_1_mem_rex64): Ditto.
29151 (*insvqi_2): Ditto.
29152 (*insvqi_3): Rename from *insvqi.
29154 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29156 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29158 * doc/cfg.texi (Edges): Remove reference to Java.
29159 (Maintaining the CFG): Ditto.
29161 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29163 PR middle-end/77674
29164 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29165 transparent aliases.
29167 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29169 PR middle-end/77484
29170 * predict.def (PRED_CALL): Update hitrate.
29171 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29172 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29173 into direct/indirect/polymorphic variants.
29175 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29177 Update copyright years.
29179 * gcc.c (process_command): Update copyright notice dates.
29180 * gcov-dump.c (print_version): Ditto.
29181 * gcov.c (print_version): Ditto.
29182 * gcov-tool.c (print_version): Ditto.
29183 * gengtype.c (create_file): Ditto.
29184 * doc/cpp.texi: Bump @copying's copyright year.
29185 * doc/cppinternals.texi: Ditto.
29186 * doc/gcc.texi: Ditto.
29187 * doc/gccint.texi: Ditto.
29188 * doc/gcov.texi: Ditto.
29189 * doc/install.texi: Ditto.
29190 * doc/invoke.texi: Ditto.
29192 Copyright (C) 2017 Free Software Foundation, Inc.
29194 Copying and distribution of this file, with or without modification,
29195 are permitted in any medium without royalty provided the copyright
29196 notice and this notice are preserved.