1 2017-10-09 Martin Jambor <mjambor@suse.cz>
4 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8 (hsa_fixup_mov_insn_type): New function.
9 (hsa_op_with_type::get_in_type): Use it.
10 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11 immediates in an assert.
12 (hsa_op_with_type::extend_int_to_32bit): New method.
13 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
14 types. Convert to dest type if necessary.
15 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
16 (reg_for_gimple_ssa): Pass false as min32int to
17 hsa_type_for_scalar_tree_type.
18 (gen_hsa_addr): Fixup type when creating addresable temporary.
19 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
20 (gen_hsa_unary_operation): Extend operands and convert to dest type if
21 necessary. Call hsa_fixup_mov_insn_type.
22 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
23 extend operands and convert to dest type if necessary.
24 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
25 to dest type if necessary.
26 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
27 if conversion nt necessary and size matches.
28 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
29 to dest type if necessary.
30 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
31 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
33 (gen_hsa_clrsb): Likewise.
34 (gen_hsa_ffs): Likewise.
35 (gen_hsa_divmod): Extend operands and convert to dest type if
37 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
39 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
41 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
42 Remove empty default arguments. Use a brace block as output
44 (conditional return): Ditto.
46 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
47 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
49 (group_ending_nop): Ditto.
51 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
52 (splitters for those): Ditto.
54 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
56 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
57 a conditional jump (and the compare for it) so that pc_rtx is the
59 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
60 for the deleted and renamed ctr<mode>_internal[234] patterns.
61 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
62 Delete second conditional return pattern.
63 (ctr<mode>_internal2): Delete this second bdnz pattern.
64 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
65 (ctr<mode>_internal4): Delete this second bdz pattern.
67 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
69 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
70 (always_initialized_rtx_for_ssa_name_p): New predicate.
71 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
72 (finish_out_of_ssa): Free new field of SA.
73 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
74 * tree-ssa-coalesce.c: Include tree-ssa.h.
75 (get_parm_default_def_partitions): Remove extern keyword.
76 (get_undefined_value_partitions): New function.
77 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
78 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
80 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
82 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
84 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
86 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
88 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
91 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
93 * i386.c (has_dispatch): Disable for Ryzen.
95 2017-10-08 Olivier Hainque <hainque@adacore.com>
97 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
98 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
100 (thumb_set_return_address): Likewise.
102 2017-10-08 Olivier Hainque <hainque@adacore.com>
104 * common/config/arm/arm-common.c (arm_except_unwind_info):
105 Handle DWARF2_UNWIND_INFO.
107 2017-10-07 Michael Collison <michael.collison@arm.com>
109 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
112 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
114 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
115 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
116 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
117 defined, force the creation of a new block for a dispatch label.
119 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
121 * invoke.texi (Wsuggest-attribute=cold): Document.
122 * common.opt (Wsuggest-attribute=cold): New
123 * ipa-pure-const.c (warn_function_cold): New function.
124 * predict.c (compute_function_frequency): Use it.
125 * predict.h (warn_function_cold): Declare.
127 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
129 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
132 2017-10-06 Martin Liska <mliska@suse.cz>
134 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
135 keyword for member functions.
136 (struct sanopt_tree_couple): New struct.
137 (struct sanopt_tree_couple_hash): New function.
138 (struct sanopt_ctx): Add new hash_map.
139 (has_dominating_ubsan_ptr_check): New function.
140 (record_ubsan_ptr_check_stmt): Likewise.
141 (maybe_optimize_ubsan_ptr_ifn): Likewise.
142 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
143 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
145 2017-10-06 Sudakshina Das <sudi.das@arm.com>
148 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
149 aarch64_simd_valid_immediate on CONST_VECTORs.
150 (aarch64_reg_or_bic_imm): Likewise.
152 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
154 PR rtl-optimization/82396
155 * haifa-sched.c (ready_sort_real): Disable qsort checking.
157 2017-10-06 Sebastian Pop <sebpop@gmail.com>
159 * graphite-dependences.c (scop_get_reads): Move code to...
160 (scop_get_must_writes): Move code to...
161 (scop_get_may_writes): Move code to...
162 (scop_get_reads_and_writes): ... here.
163 (scop_get_dependences): Call scop_get_reads_and_writes.
165 2017-10-06 Jakub Jelinek <jakub@redhat.com>
167 PR tree-optimization/82434
168 * fold-const.h (can_native_encode_type_p,
169 can_native_encode_string_p): Remove.
170 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
171 don't encode anything, just return what would be otherwise returned.
172 (native_encode_fixed, native_encode_complex, native_encode_vector):
174 (native_encode_string): Likewise. Inline by hand
175 can_native_encode_string_p.
176 (can_native_encode_type_p): Remove.
177 (can_native_encode_string_p): Remove.
178 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
179 STRING_CSTs using can_native_encode_string_p, test all
180 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
181 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
182 argument from native_encode_expr.
183 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
184 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
187 2017-10-06 Richard Biener <rguenther@suse.de>
189 PR tree-optimization/82397
190 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
191 operand_equal_p but rely on data_ref_compare_tree for detecting
193 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
194 to match up with dr_group_sort_cmp.
196 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
199 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
201 * config/s390/s390-builtin-types.def: Regenerate.
203 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
206 * config/s390/s390-builtin-types.def: Regenerate.
207 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
208 Change flag from B_VXE to B_VX.
209 (s390_vec_min_dbl): Remove B_VXE flag.
211 2017-10-06 Richard Biener <rguenther@suse.de>
213 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
214 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
215 translate_isl_ast_to_gimple::is_valid_rename,
216 translate_isl_ast_to_gimple::get_rename,
217 translate_isl_ast_to_gimple::get_def_bb_for_const,
218 translate_isl_ast_to_gimple::get_new_name,
219 translate_isl_ast_to_gimple::collect_all_ssa_names,
220 translate_isl_ast_to_gimple::copy_loop_phi_args,
221 translate_isl_ast_to_gimple::collect_all_ssa_names,
222 translate_isl_ast_to_gimple::copy_loop_phi_args,
223 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
224 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
225 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
226 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
227 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
228 translate_isl_ast_to_gimple::copy_cond_phi_args,
229 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
230 translate_isl_ast_to_gimple::edge_for_new_close_phis,
231 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
232 translate_isl_ast_to_gimple::rename_uses,
233 translate_isl_ast_to_gimple::rename_all_uses): Remove.
234 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
235 (set_rename_for_each_def): Likewise.
236 (graphite_copy_stmts_from_block): Handle debug stmt resetting
237 here. Handle rewriting SCEV analyzable uses here.
238 (copy_bb_and_scalar_dependences): Generate code for PHI
240 (graphite_regenerate_ast_isl): Adjust.
241 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
242 (add_write, add_read): New functions.
243 (build_cross_bb_scalars_def): Use it and simplify.
244 (build_cross_bb_scalars_use): Likewise.
245 (graphite_find_cross_bb_scalar_vars): Inline into...
246 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
247 simulating out-of-SSA. Compute liveout and add dependencies.
248 (build_scops): Force an empty entry block.
249 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
251 (sese_build_liveouts): Declare.
252 (sese_trivially_empty_bb_p): Likewise.
253 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
254 compute liveout and debug_liveout.
255 (sese_bad_liveouts_use): Remove.
256 (sese_reset_debug_liveouts_bb): Likewise.
257 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
258 (sese_build_liveouts): Build liveout and debug_liveout and store
260 (new_sese_info): Adjust.
261 (free_sese_info): Likewise.
262 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
263 do not build liveout here.
264 (move_sese_in_condition): Adjust region entry.
265 (scev_analyzable_p): Match up with chrec_apply requirements.
266 (sese_trivially_empty_bb_p): New.
267 * tree-into-ssa.c (get_reaching_def): Properly support generating
268 default-defs for incremental rewrite of anonymous names.
270 2017-10-06 Richard Biener <rguenther@suse.de>
272 * graphite-sese-to-poly.c (extract_affine): For casts increasing
273 precision do not perform modulo reduction.
275 2017-10-06 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/82436
278 * tree-vect-slp.c (vect_supported_load_permutation_p): More
279 conservatively choose the vectorization factor when checking
280 whether we can perform the required load permutation.
281 (vect_transform_slp_perm_load): Assert when we may not fail.
283 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
285 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
286 message for incompatible -msdata=* and -mcall-* options.
288 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
290 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
291 for post-reload scheduling.
293 2017-10-05 Tamar Christina <tamar.christina@arm.com>
295 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
297 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
299 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
300 to improve monte carlo in scimark.
302 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
304 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
305 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
306 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
307 pentium4_cost, nocona_cost): Set reassociation width to 1.
308 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
309 width to 2 for fp operations and 1 otherwise.
310 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
312 (atom_cost): Set reassociation width to 2.
313 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
314 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
315 (core_cost): Set fp reassociation width to 4 and vector to 2.
316 (ix86_reassociation_width): Rewrite using cost table; special case
317 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
318 and TARGET_AVX128_OPTIMAL.
319 * i386.h (processor_costs): Add
320 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
321 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
322 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
323 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
324 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
325 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
327 2017-10-05 Nathan Sidwell <nathan@acm.org>
329 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
331 2017-10-05 Tamar Christina <tamar.christina@arm.com>
333 * config/arm/arm.c (arm_test_fpu_data): New.
334 (arm_run_selftests): Call arm_test_fpu_data.
336 2017-10-04 Nathan Sidwell <nathan@acm.org>
338 * toplev.c (toplev::main): Remove excess parens on pretty_printer
340 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
342 2017-10-04 Sudakshina Das <sudi.das@arm.com>
344 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
345 check type for aarch64_simd_valid_immediate.
346 (aarch64_output_simd_mov_immediate): Update prototype.
347 (aarch64_simd_valid_immediate): Update prototype.
348 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
349 support for ORR-immediate.
350 (and<mode>3): modified pattern to add support for BIC-immediate.
351 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
352 now checks for valid immediate for BIC and ORR based on new enum
354 (aarch64_output_simd_mov_immediate): Function now used to output
355 BIC/ORR imm as well based on new enum argument.
356 * config/aarch64/constraints.md (Do): New vector immediate constraint.
358 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
359 (aarch64_reg_or_bic_imm): Likewise.
361 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
363 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
364 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
366 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
370 PR rtl-optimization/82396
371 * haifa-sched.c (autopref_multipass_init): Simplify
373 (autopref_rank_data): Simplify sort order.
374 * sched-int.h (autopref_multipass_data_): Remove
375 multi_mem_insn_p, min_offset and max_offset.
377 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
379 * doc/sourcebuild.texi: Document vect_peeling_profitable.
381 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
383 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
386 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
388 * doc/sourcebuild.texi: Document vect_long_mult.
390 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
392 PR tree-optimization/82413
393 * fold-const.c (build_range_check): Use widest_int when comparing
394 the maximum ETYPE value with HIGH.
396 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
398 PR rtl-optimization/82396
399 * haifa-sched.c (autopref_multipass_init): Simplify
401 (autopref_rank_data): Simplify sort order.
402 * sched-int.h (autopref_multipass_data_): Remove
403 multi_mem_insn_p, min_offset and max_offset.
405 2017-10-04 Jakub Jelinek <jakub@redhat.com>
407 PR tree-optimization/82381
408 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
409 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
412 PR tree-optimization/82374
413 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
414 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
415 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
416 current_function_decl to the new decl.
418 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
420 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
421 helper macro for IEEE float128 hardware built-in functions.
422 (SQRTF128_ODD): Add built-in functions with the round-to-odd
424 (TRUNCF128_ODD): Likewise.
425 (ADDF128_ODD): Likewise.
426 (SUBF128_ODD): Likewise.
427 (MULF128_ODD): Likewise.
428 (DIVF128_ODD): Likewise.
429 (FMAF128_ODD): Likewise.
430 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
431 UNSPEC_TRUNC_ROUND_TO_ODD.
432 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
433 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
434 floating point round to odd instructions.
435 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
436 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
437 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
438 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
439 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
440 (trunc<mode>sf2_hw): Change the truncate with round to odd
441 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
442 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
443 to odd hardware instructions.
444 (sub<mode>3_odd): Likewise.
445 (mul<mode>3_odd): Likewise.
446 (div<mode>3_odd): Likewise.
447 (sqrt<mode>2_odd): Likewise.
448 (fma<mode>4_odd): Likewise.
449 (fms<mode>4_odd): Likewise.
450 (nfma<mode>4_odd): Likewise.
451 (nfms<mode>4_odd): Likewise.
452 (trunc<mode>df2_odd): Change the truncate with round to odd
453 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
455 * doc/extend.texi (PowerPC built-in functions): Update documentation
456 for existing IEEE float128-bit built-in functions. Add built-in
457 functions that generate the IEEE 128-bit floating point round to
460 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
462 PR rtl-optimization/77729
463 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
464 to (X&(C1&~C2))|C2 transformations.
466 2017-10-03 Martin Jambor <mjambor@suse.cz>
468 PR tree-optimization/82363
469 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
470 mismatch, mark lacc written regardless of racc.
472 2017-10-03 Jakub Jelinek <jakub@redhat.com>
474 PR tree-optimization/82381
475 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
476 stmt_to_insert nor wheather SSA_NAMEs are default defs.
477 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
478 fallthrough into reassoc_stmt_dominates_stmt_p.
481 * combine.c (combine_instructions): Don't combine in unreachable
484 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
487 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
488 function to not use the have_cpu variable. Do not set cpu_index,
489 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
491 (rs6000_valid_attribute_p): Remove duplicate initializations of
492 old_optimize and func_optimize.
493 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
494 (rs6000_activate_target_options): Make global.
495 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
498 2017-10-02 Jakub Jelinek <jakub@redhat.com>
500 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
501 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
502 Set *poffset to 0 and *psize and *pmax_size to -1 if
503 *poffset + *psize overflows in HOST_WIDE_INT.
505 PR tree-optimization/82387
506 PR tree-optimization/82388
507 PR tree-optimization/82389
508 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
509 instead of live_bytes non-NULL.
511 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
514 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
517 2017-10-02 Richard Biener <rguenther@suse.de>
519 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
520 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
521 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
523 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
525 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
526 requested precision matches the type's.
527 * calls.c (alloc_max_size): Calculate the new candidate size as
528 a widest_int and use wi::to_widest when comparing it with the
529 current candidate size.
530 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
531 zero rather than integer_zero_node.
532 * match.pd: Check for a no-op conversion before using wi::add
533 rather than after. Use tree_to_uhwi when summing small shift
534 counts into an unsigned int.
536 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
537 Alan Hayward <alan.hayward@arm.com>
538 David Sherwood <david.sherwood@arm.com>
541 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
542 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
543 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
546 2017-10-02 Richard Biener <rguenther@suse.de>
548 PR tree-optimization/82355
549 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
550 a mapping for the enclosing loop but avoid generating one for
552 (copy_bb_and_scalar_dependences): Remove premature codegen
553 error on PHIs in blocks duplicated into multiple places.
554 * graphite-scop-detection.c
555 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
556 in the region use it as loop and nest to analyze the DR in.
557 (try_generate_gimple_bb): Likewise.
558 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
559 (add_loop_constraints): For blocks in a loop not in the region
560 create a dimension with a single iteration.
561 * sese.h (gbb_loop_at_index): Remove assert.
563 2017-10-01 Kevin Buettner <kevinb@redhat.com>
565 * omp-expand.c (adjust_context_scope): New function.
566 (expand_parallel_call): Call adjust_context_scope.
568 2017-10-01 Jeff Law <law@redhat.com>
570 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
571 dom_opt_dom_walker class with direct access to private members.
572 Add comments. Call test_for_singularity.
573 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
574 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
575 m_dummy_cond anymore.
576 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
578 (pass_dominator:execute): Build the dummy_cond here and pass it
579 to the dom_opt_dom_walker ctor.
580 (test_for_singularity): New function.
582 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
583 Maya Rashish <coypu@sdf.org>
585 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
586 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
587 (alpha*-*-netbsd*) Use nbsd_tm_file.
588 (arm*-*-netbsdelf*) Likewise.
589 (i[34567]86-*-netbsdelf*) Likewise.
590 (x86_64-*-netbsd*) Likewise.
591 (mips*-*-netbsd*) Likewise.
592 (powerpc-*-netbsd*) Likewise.
593 (sh*-*-netbsd*) Likewise.
594 (sparc-*-netbsdelf*) Likewise.
595 (sparc64-*-netbsd*) Likewise.
596 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
598 (vax-*-netbsdelf*) Likewise.
599 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
600 (UINT_FAST8_TYPE) Likewise.
601 (INT_FAST16_TYPE) Check CHAR_FAST16.
602 (UINT_FAST16_TYPE) Likewise.
604 2017-09-30 Jakub Jelinek <jakub@redhat.com>
607 * config/i386/i386.md
608 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
609 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
610 *divmodsi4_zext_2): New define_insn_and_split.
611 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
612 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
613 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
614 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
615 New define_insn_and_split.
616 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
617 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
618 operands[1] having DImode when mode is SImode.
620 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
621 always SImode for DIV and MOD in REG_EQUAL notes.
623 2017-09-29 Yury Gribov <tetra2005@gmail.com>
626 * match.pd: Fix handling of NaNs in pattern.
628 2017-09-29 Jeff Law <law@redhat.com>
630 * sbitmap.c (bitmap_bit_in_range_p): New function.
631 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
632 * tree-ssa-dse.c (live_bytes_read): New function.
633 (dse_classify_store): Ignore reads of dead bytes.
635 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
636 typos and whitespace errors.
637 * config/i386/predicates.md (address_no_seg_operand): Likewise.
638 * config/s390/s390.c (s390_emit_prologue): Likewise.
640 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
643 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
644 with a symbol for LRA.
646 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
648 PR rtl-optimization/82338
649 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
651 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
653 * genmodes.c (calc_wider_mode): Suppress qsort macro.
654 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
655 (qsort_chk): Declare.
656 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
657 (qsort_chk): New function.
659 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
661 PR tree-optimization/82337
662 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
663 phi definition if the PHI result appears in an abnormal PHI.
664 (find_basis_for_base_expr): Don't record a basis if the LHS of the
665 basis appears in an abnormal PHI.
667 2017-09-29 Richard Biener <rguenther@suse.de>
669 * graphite-isl-ast-to-gimple.c
670 (translate_isl_ast_to_gimple::set_codegen_error): New function.
671 (binary_op_to_tree): Use it.
672 (get_rename_from_scev): Likewise.
673 (copy_loop_phi_nodes): Likewise.
674 (copy_bb_and_scalar_dependences): Likewise.
675 (translate_pending_phi_nodes): Likewise.
677 2017-09-29 Jakub Jelinek <jakub@redhat.com>
680 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
681 for movabsq $(i32 << shift), r64.
683 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
685 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
686 index when encoding %esp as %rsp to avoid 0x67 prefix.
688 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
690 * config/i386/i386.md (*movsf_internal, *movdf_internal):
691 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
693 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
695 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
696 Extensions with more than 16 double VFP registers.
697 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
698 to_clear_mask and all code related to it. Replace the remaining
699 entry by a sbitmap and adapt code accordingly.
701 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
703 * brig-builtins.def: Change pure attributes to const.
705 2017-09-28 Joseph Myers <joseph@codesourcery.com>
707 * config.gcc (default_gnu_indirect_function): Default to yes for
708 sparc*-*-linux* with glibc.
710 2017-09-28 Joseph Myers <joseph@codesourcery.com>
712 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
713 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
714 when creating .init_array and .fini_array sections with priority
717 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
720 * config/aarch64/aarch64.c
721 (aarch64_builtin_support_vector_misalignment): Always return false
722 when misalignment is unknown.
724 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
726 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
727 this function to return false if the definition used by the swap
728 instruction is artificial, or if the memory address from which the
729 constant value is loaded is not represented by a base address held
730 in a register or if the base address register is a frame or stack
731 pointer. Additionally, return false if the base address of the
732 loaded constant is a SYMBOL_REF but is not considered to be a
734 (replace_swapped_load_constant): New function.
735 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
736 loaded constant vector with a load of a swapped constant vector.
738 2017-09-27 Carl Love <cel@us.ibm.com>
740 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
741 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
742 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
745 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
747 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
748 first, always call autopref_rank_data otherwise.
750 2017-09-27 Richard Biener <rguenther@suse.de>
752 * graphite-scop-detection.c (find_scop_parameters): Move
753 loop bound handling ...
754 (gather_bbs::before_dom_children): ... here, avoiding the need
755 to build scop_info->loop_nest.
756 (record_loop_in_sese): Remove.
757 * sese.h (sese_info_t::loop_nest): Remove.
758 * sese.c (new_sese_info): Do not allocate loop_nest.
759 (free_sese_info): Do not free loop_nest.
761 2017-09-27 Jakub Jelinek <jakub@redhat.com>
764 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
765 lhs from calls if the lhs has addressable type.
767 2017-09-27 Richard Biener <rguenther@suse.de>
769 * graphite.h (scop::max_alias_set): New member.
770 * graphite-scop-detection.c: Remove references to non-existing
772 (build_alias_sets): Record the maximum alias set used for drs.
773 (build_scops): Support zero as unlimited for
774 --param graphite-max-arrays-per-scop.
775 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
777 (build_poly_sr_1): ... here. Compute alias set based on the
778 maximum alias set used for drs rather than
779 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
781 2017-09-27 Richard Biener <rguenther@suse.de>
783 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
784 --param loop-block-tile-size=0 to disable tiling.
786 2017-09-27 Richard Biener <rguenther@suse.de>
788 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
789 (graphite-max-nb-scop-params): Document special value zero.
790 * domwalk.h (dom_walker::STOP): New symbolical constant.
791 (dom_walker::dom_walker): Add optional parameter for bb to
793 (dom_walker::~dom_walker): Declare.
794 (dom_walker::before_dom_children): Document STOP return value.
795 (dom_walker::m_user_bb_to_rpo): New member.
796 (dom_walker::m_bb_to_rpo): Likewise.
797 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
798 mapping here if not provided by the user.
799 (dom_walker::~dom_walker): Free bb to RPO mapping if not
800 provided by the user.
801 (dom_walker::STOP): Define.
802 (dom_walker::walk): Do not compute bb to RPO mapping here.
803 Support STOP return value from before_dom_children to stop
805 * graphite-optimize-isl.c (optimize_isl): If the schedule
806 is the same still generate code if -fgraphite-identity
807 or -floop-parallelize-all are given.
808 * graphite-scop-detection.c: Include cfganal.h.
809 (gather_bbs::gather_bbs): Get and pass through bb to RPO
811 (gather_bbs::before_dom_children): Return STOP for BBs
813 (build_scops): Compute bb to RPO mapping and pass it to
814 the domwalk. Treat --param graphite-max-nb-scop-params=0
815 as not limiting the number of params.
816 * graphite.c (graphite_initialize): Remove limit on the
817 number of basic-blocks in a function.
818 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
819 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
822 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
824 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
825 Adjust code to eliminate needing to do the shift right 32-bits
826 operation after XSCVDPSPN.
828 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
830 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
831 ((X / Y) != 0 -> X >= Y): Likewise.
833 2017-09-26 Carl Love <cel@us.ibm.com>
835 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
836 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
837 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
838 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
839 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
840 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
841 definitions and overloading.
842 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
843 statement for P9V_BUILTIN_XST_LEN_R.
844 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
845 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
846 define_expand and define_insn for the instructions and builtins.
847 * doc/extend.texi: Update the built-in documentation file for the new
849 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
850 define_insn for the instructions
852 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
855 * gcc/config/netbsd-protos.h: New file.
856 * gcc/config/netbsd.c: New file.
857 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
858 * gcc/config/t-netbsd: New file.
859 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
860 (tmake_file) Add t-netbsd.
861 (extra_objs) Add netbsd.o.
863 2017-09-26 Janus Weil <janus@gcc.gnu.org>
867 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
869 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
871 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
872 sign extension from a vector register to a GPR by doing a 32-bit
873 direct move and then an EXTSW.
874 (extendsi<mode>2 splitter): Likewise.
875 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
876 right or vector extract after doing XSCVDPSPN. Use
877 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
879 (movdi_from_sf_zero_ext): Likewise.
880 (reload_gpr_from_vsxsf): Likewise.
881 (p8_mfvsrd_4_disf): Delete, no longer used.
882 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
883 then needing to move the SFmode to a GPR to use the XSCVDPSP
884 instruction instead of FRSP and XSCVDPSPN.
885 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
886 it is adjacent to the other XSCVSPDP insns.
887 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
888 SFmode to be in traditional Altivec registers.
889 (vsx_xscvdpspn): Eliminate useless alternative constraint.
890 (vsx_xscvspdpn): Likewise.
891 (vsx_xscvspdpn_scalar): Likewise.
893 2017-09-26 Martin Jambor <mjambor@suse.cz>
895 * tree-sra.c (compare_access_positions): Put integral types first,
896 stabilize sorting of integral types, remove conditions putting
897 non-full-precision integers last.
898 (sort_and_splice_var_accesses): Disable scalarization if a
899 non-integert would be represented by a non-full-precision integer.
901 2017-09-26 Joseph Myers <joseph@codesourcery.com>
903 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
904 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
905 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
906 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
907 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
908 conditionals inside the function instead of around it. Call
909 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
910 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
912 2017-09-26 Richard Biener <rguenther@suse.de>
914 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
916 (scop_detection::build_scop_breadth): ... this. Removed.
917 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
918 (scop_detection::harmful_stmt_in_bb): Likewise.
919 (scop_detection::graphite_can_represent_stmt): Likewise.
920 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
921 (scop_detection::can_represent_loop): Remove recursion, fold in ...
922 (scop_detection::can_represent_loop_1): ... this. Removed.
923 (scop_detection::harmful_loop_in_region): Simplify after inlining
924 the above and remove more quadraticness.
925 (build_scops): Adjust.
926 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
929 2017-09-26 Jakub Jelinek <jakub@redhat.com>
932 * config/i386/i386.c (ix86_print_operand_address_as): Only test
933 REGNO (base) == SP_REG if base is a REG.
936 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
937 if it is different SSA_NAME.
938 (optimize_range_tests_cmp_bitwise): New function.
939 (optimize_range_tests): Call it.
941 2017-09-26 Richard Biener <rguenther@suse.de>
943 PR tree-optimization/82321
944 * graphite.c (canonicalize_loop_closed_ssa): Properly check
945 for the def being inside the loop.
947 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
949 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
951 * config/s390/s390-builtins.def: Fix constraint on op4.
953 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
955 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
956 independent expanders.
957 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
958 ("vec_ordered", "vec_unordered"): New expanders.
960 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
962 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
965 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
967 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
968 vec_unpacks_lo_v16qi.
969 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
971 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
973 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
974 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
975 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
977 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
979 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
981 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
983 ("*vec_slb<mode>"): New insn pattern.
984 ("vec_shr_<mode>"): New expander.
985 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
986 and force the shift count operand to V16QImode.
987 ("vec_srb<mode>"): Set shift count mode to V16QI.
989 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
991 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
992 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
993 ("vec_widen_smult_hi_<mode>"): New expander definitions.
995 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
998 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1000 2017-09-26 Richard Biener <rguenther@suse.de>
1002 PR tree-optimization/82320
1003 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1006 2017-09-25 Jeff Law <law@redhat.com>
1008 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1009 prototype for new argument.
1010 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1011 mostly extracted from rs6000_emit_allocate_stack.
1012 (rs6000_emit_probe_stack_range_stack_clash): New function.
1013 (rs6000_emit_allocate_stack): Call
1014 rs6000_emit_probe_stack_range_stack_clash as needed.
1015 (rs6000_emit_probe_stack_range): Add additional argument
1016 to call to gen_probe_stack_range{si,di}.
1017 (output_probe_stack_range): New.
1018 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1019 (output_probe_stack_range_stack_clash): New.
1020 (rs6000_emit_prologue): Emit notes into dump file as requested.
1021 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1022 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1023 Add additional operand and pass it to output_probe_stack_range.
1025 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1027 PR tree-optimization/82163
1028 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1029 (checking_verify_loop_closed_ssa): New parameter.
1030 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1031 (check_loop_closed_ssa_stmt): Delete.
1032 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1033 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1034 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1037 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1039 * brig-builtins.def: Treat HSAIL barrier builtins as
1040 setjmp/longjump style functions.
1042 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1044 * target.def (constant_alignment): New hook.
1045 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1046 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1047 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1048 * doc/tm.texi: Regenerate.
1049 * targhooks.h (default_constant_alignment): Declare.
1050 (constant_alignment_word_strings): Likewise.
1051 * targhooks.c (default_constant_alignment): New function.
1052 (constant_alignment_word_strings): Likewise.
1053 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1054 instead of CONSTANT_ALIGNMENT.
1055 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1056 (force_const_mem): Likewise.
1057 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1058 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1059 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1060 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1061 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1063 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1064 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1065 constant_alignment_word_strings.
1066 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1067 (CONSTANT_ALIGNMENT): Likewise.
1068 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1069 (arm_constant_alignment): New function.
1070 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1071 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1072 constant_alignment_word_strings.
1073 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1074 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1075 constant_alignment_word_strings.
1076 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1077 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1078 (cris_constant_alignment): New function.
1079 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1080 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1081 (epiphany_constant_alignment): New function.
1082 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1083 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1084 constant_alignment_word_strings.
1085 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1086 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1087 constant_alignment_word_strings.
1088 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1089 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1090 constant_alignment_word_strings.
1091 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1092 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1093 * config/i386/i386.c (ix86_constant_alignment): Make static.
1094 Use the same interface as the target hook.
1095 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1096 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1097 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1098 constant_alignment_word_strings.
1099 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1100 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1101 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1102 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1103 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1104 constant_alignment_word_strings.
1105 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1106 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1107 constant_alignment_word_strings.
1108 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1109 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1110 constant_alignment_word_strings.
1111 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1112 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1114 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1115 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1116 * config/mips/mips.c (mips_constant_alignment): New function.
1117 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1118 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1119 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1120 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1121 (mmix_constant_alignment): Make static. Use the same interface
1123 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1124 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1125 constant_alignment_word_strings.
1126 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1127 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1128 constant_alignment_word_strings.
1129 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1130 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1131 constant_alignment_word_strings.
1132 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1133 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1134 (rs6000_constant_alignment): New function.
1135 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1136 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1137 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1138 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1139 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1140 (rs6000_constant_alignment): New function.
1141 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1142 * config/s390/s390.c (s390_constant_alignment): New function.
1143 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1144 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1145 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1146 constant_alignment_word_strings.
1147 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1148 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1149 (sparc_constant_alignment): New function.
1150 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1151 * config/spu/spu.c (spu_constant_alignment): New function.
1152 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1153 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1154 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1155 constant_alignment_word_strings.
1156 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1157 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1158 constant_alignment_word_strings.
1159 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1160 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1161 constant_alignment_word_strings.
1162 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1163 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1164 (visium_constant_alignment): New function.
1165 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1166 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1167 (xtensa_constant_alignment): New function.
1168 * system.h (CONSTANT_ALIGNMENT): Poison.
1170 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1172 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1173 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1174 (rs6000_builtin_valid_without_lhs): New helper function.
1175 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1176 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1178 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1180 * target.h (vec_perm_indices): Use unsigned short rather than
1182 (auto_vec_perm_indices): Likewise.
1183 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1184 Use unsigned int rather than unsigned char.
1185 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1187 2017-09-25 Richard Biener <rguenther@suse.de>
1189 * cfgloop.h (sort_sibling_loops): Declare.
1190 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1191 (sort_sibling_loops): New function sorting the sibling loop list
1193 * graphite.c (graphite_transform_loops): Sort sibling loops.
1195 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1197 * target.def (vec_perm_const_ok): Change sel parameter to
1199 * optabs-query.c (can_vec_perm_p): Update accordingly.
1200 * doc/tm.texi: Regenerate.
1201 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1202 auto_vec_perm_indices and remove separate nelt field.
1203 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1204 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1205 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1206 (aarch64_expand_vec_perm_const): Update accordingly.
1207 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1208 to vec_perm_indices.
1209 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1210 auto_vec_perm_indices and remove separate nelt field.
1211 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1212 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1213 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1215 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1216 to vec_perm_indices.
1217 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1218 sel to vec_perm_indices.
1219 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1220 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1221 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1223 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1226 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1229 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1230 on the FUNCTION_DECL function context if it has a DIE that is a
1233 2017-09-25 Richard Biener <rguenther@suse.de>
1235 PR tree-optimization/82285
1236 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1239 2017-09-25 Tom de Vries <tom@codesourcery.com>
1243 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1246 2017-09-25 Richard Biener <rguenther@suse.de>
1248 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1249 ISL errors other than isl_error_quota happen. Dump if the
1250 schedule is the same.
1251 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1252 errors instead of aborting inside ISL.
1254 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1257 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1258 of libgcc_eh for m64.
1259 * config/i386/darwin64.h: Likewise.
1261 2017-09-25 Richard Biener <rguenther@suse.de>
1264 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1265 attribute for incomplete types nor twice for complete ones.
1267 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1270 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1271 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1274 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1277 * config/i386/i386.opt (mprefer-avx256): Use
1278 ix86_target_flags variable.
1279 * config/i386/i386.c (ix86_target_string): Move
1280 -mprefer-avx256 to flag2_opts.
1282 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1285 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1286 and x != -1 | y != -1 into (x & y) != -1.
1288 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1290 * config.gcc: Add new case statement to set
1291 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1292 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1293 s390x-*-linux* case statements. Added aarch64 to the list of
1294 supported architectures.
1296 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1298 PR tree-optimization/82289
1299 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1300 STMT_VINFO_RELEVANT_P.
1302 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1303 Alan Hayward <alan.hayward@arm.com>
1304 David Sherwood <david.sherwood@arm.com>
1306 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1307 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1308 (extract_range_from_binary_expr_1): Don't call
1309 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1311 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1312 Alan Hayward <alan.hayward@arm.com>
1313 David Sherwood <david.sherwood@arm.com>
1315 * target.def (preferred_vector_alignment): New hook.
1316 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1318 * doc/tm.texi: Regenerate.
1319 * targhooks.h (default_preferred_vector_alignment): Declare.
1320 * targhooks.c (default_preferred_vector_alignment): New function.
1321 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1323 (DR_TARGET_ALIGNMENT): New macro.
1324 (aligned_access_p): Update commentary.
1325 (vect_known_alignment_in_bytes): New function.
1326 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1328 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1329 Calculate the misalignment based on the target alignment rather than
1331 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1332 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1333 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1334 the target alignment, rather than masking the element misalignment
1335 with the number of elements in a vector. Also use the target
1336 alignment when calculating the maximum number of peels.
1337 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1338 instead of TYPE_ALIGN_UNIT.
1339 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1340 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1341 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1342 (vect_create_data_ref_ptr): Likewise.
1343 (vect_setup_realignment): Realign by ANDing with
1344 -DR_TARGET_MISALIGNMENT.
1345 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1346 the number of peels based on DR_TARGET_ALIGNMENT.
1347 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1348 with the guaranteed alignment boundary when deciding whether
1350 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1351 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1352 (ensure_base_align): Remove stmt_info parameter. Get the
1353 target base alignment from DR_TARGET_ALIGNMENT.
1354 (vectorizable_store): Update call accordingly. Interpret
1355 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1357 (vectorizable_load): Likewise.
1359 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1364 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1365 (vect_enhance_data_refs_alignment): Likewise.
1367 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1369 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1370 error. Only quit immediately if parsing is complete.
1371 (BEGIN): Initialize fatal_err and parse_done.
1372 (begin fpu, end fpu): Check number of arguments.
1373 (begin arch, end arch): Likewise.
1374 (begin cpu, end cpu): Likewise.
1375 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1376 (optalias): Likewise.
1378 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1380 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1381 * config/arm/arm-isa.h: Delete. Move definitions to ...
1382 * arm-cpus.in: ... here. Use new feature and fgroup values.
1383 * config/arm/arm.c (arm_option_override): Use lower case for feature
1385 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1386 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1387 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1389 (print_isa_bits_for): New function.
1390 (gen_isa): New function.
1391 (gen_comm_data): Use print_isa_bits_for.
1392 (define feature): New keyword.
1393 (define fgroup): New keyword.
1394 * config/arm/t-arm (TM_H): Remove.
1395 (GTM_H): Add arm-isa.h.
1396 (arm-isa.h): Add rule to generate file.
1397 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1398 case for feature bit names.
1400 2017-09-22 Richard Biener <rguenther@suse.de>
1402 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1404 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1405 print of no dependency loops ...
1406 * graphite.c (graphite_transform_loops): ... here.
1407 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1409 (same_close_phi_node, remove_duplicate_close_phi,
1410 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1411 (canonicalize_loop_closed_ssa): ... here and simplify.
1412 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1413 (optimize_isl): Use dump_printf_loc to tell when we stopped
1414 optimizing because of an ISL timeout.
1416 2017-09-22 Richard Biener <rguenther@suse.de>
1418 PR tree-optimization/82291
1419 * tree-if-conv.c (predicate_mem_writes): Make sure to
1420 remove writes in blocks predicated with false.
1422 2017-09-22 Richard Biener <rguenther@suse.de>
1424 * sese.c: Include cfganal.h.
1425 (if_region_set_false_region): Remove.
1426 (create_if_region_on_edge): Likewise.
1427 (move_sese_in_condition): Re-implement without destroying
1430 2017-09-22 Richard Biener <rguenther@suse.de>
1432 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1433 Verify both BBs contain loop PHI nodes before dispatching to
1435 (graphite_regenerate_ast_isl): Do not recompute dominators,
1436 do not verify three times. Restructure for clarity.
1437 * graphite-scop-detection.c (same_close_phi_node,
1438 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1439 defined_in_loop_p, canonicalize_loop_closed_ssa,
1440 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1441 checking and SSA rewrite, move to ...
1442 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1443 (graphite_initialize): Do not pass in ctx, do not reset the
1444 SCEV cache, compute only dominators.
1445 (graphite_transform_loops): Allocate ISL ctx after
1446 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1447 Maintain post-dominators only around build_scops.
1448 * sese.c (if_region_set_false_region): Make static. Free
1449 and recompute dominators.
1450 (move_sese_in_condition): Assert we don't get called with
1451 post-dominators computed.
1452 * sese.h (if_region_set_false_region): Remove.
1454 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1456 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1457 mode attribute for TARGET_AVX512VL.
1459 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1461 * config/i386/i386.opt (mprefer-avx256): New option.
1462 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1464 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1465 for TARGET_PREFER_AVX256.
1466 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1468 2017-09-21 Jeff Law <law@redhat.com>
1470 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1471 Fix dump output if the only stack space is for pushed registers.
1473 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1475 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1478 2017-09-21 Martin Sebor <msebor@redhat.com>
1481 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1482 code (in C++) or code that triggers warnings.
1484 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1486 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1488 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1490 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1491 * sched-rgn.c: Ditto.
1492 * sel-sched-ir.c: Ditto.
1494 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1496 * toplev.h (set_random_seed): Adjust return type.
1497 * toplev.c (init_local_tick): Move eager initialization of random_seed
1498 to get_random_seed. Adjust comment.
1499 (init_random_seed): Inline to get_random_seed, delete.
1500 (get_random_seed): Initialize random_seed lazily.
1501 (set_random_seed): Do not return previous value.
1502 (print_switch_value): Do not call get_random_seed.
1504 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1506 * cgraph.c (delete_function_version): New, broken out from...
1507 (cgraph_node::delete_function_version): ...here. Rename to
1508 cgraph_node::delete_function_version_by_decl. Update all uses.
1509 (cgraph_node::remove): Call delete_function_version.
1511 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1514 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1515 VAR_DECLs to which addressable non-volatile parameters are mapped
1516 and for id->retvar after the return value assignment. Clear
1517 id->retval and id->retbnd after inlining.
1519 2017-09-21 Richard Biener <rguenther@suse.de>
1521 PR tree-optimization/82276
1522 PR tree-optimization/82244
1523 * tree-vrp.c (build_assert_expr_for): Set
1524 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1526 (remove_range_assertions): Revert earlier change.
1528 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1531 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1533 2017-09-21 Richard Biener <rguenther@suse.de>
1535 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1536 Restore valid IL after code generation errors.
1537 * graphite.c (graphite_transform_loops): Diagnose code
1538 generation issues as MSG_MISSED_OPTIMIZATION and continue
1539 with processing SCOPs.
1541 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1542 Alan Hayward <alan.hayward@arm.com>
1543 David Sherwood <david.sherwood@arm.com>
1545 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1546 rather than choosing between plus_constant and gen_rtx_<CODE>.
1547 * expr.c (emit_push_insn): Likewise.
1548 (expand_expr_real_2): Likewise.
1550 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1551 Alan Hayward <alan.hayward@arm.com>
1552 David Sherwood <david.sherwood@arm.com>
1554 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1556 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1557 Alan Hayward <alan.hayward@arm.com>
1558 David Sherwood <david.sherwood@arm.com>
1560 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1561 calling tree_to_uhwi.
1563 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1568 INTEGER_CST rather than a negative test for ADDR_EXPR.
1570 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1571 Alan Hayward <alan.hayward@arm.com>
1572 David Sherwood <david.sherwood@arm.com>
1574 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1575 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1577 2017-09-21 Richard Biener <rguenther@suse.de>
1579 PR tree-optimization/71351
1580 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1581 graphite_create_new_loop_guard): Remove, fold remaining parts
1583 (translate_isl_ast_node_for): ... here and simplify.
1585 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1588 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1589 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1590 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1591 alternative always use QI mode, for -Os imov (=R,R) alternative
1592 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1595 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1596 Jeff Law <law@redhat.com>
1598 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1599 (allocate_stack_space): New function, partially extracted from
1601 (s390_emit_prologue): Track offset to most recent stack probe.
1602 Code to allocate space moved into allocate_stack_space.
1603 Dump actions when no stack is allocated.
1604 (s390_prologue_plus_offset): New function.
1605 (s390_emit_stack_probe): Likewise.
1607 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1609 * common.opt (Wa, Wl, Wp, g, gz=): Add
1611 (gno-column-info): Remove.
1612 (gcolumn-info): Drop RejectNegative.
1614 (gno-record-gcc-switches): Remove.
1615 (grecord-gcc-switches): Drop RejectNegative.
1616 (gno-split-dwarf): Remove.
1617 (gsplit-dwarf): Drop RejectNegative.
1618 (gno-strict-dwarf): Remove.
1619 (gstrict-dwarf): Drop RejectNegative.
1620 * config/darwin.opt (gfull, gused): Add RejectNegative.
1621 * dwarf2out.c (gen_producer_string): Drop
1622 gno-record-gcc-switches handler.
1623 * optc-gen.awk: Add g to prefixes with negative forms.
1624 * opts-common.c (remapping_prefix_p): New.
1625 (find_opt): Check it.
1626 (generate_canonical_option): Test g prefix.
1627 (option_map): Add -gno- mapping.
1628 (add_misspelling_candidates): Check remapping_prefix_p.
1630 2017-09-20 Jeff Law <law@redhat.com>
1632 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1633 thinko in stack clash protection support.
1635 * explow.c (compute_stack_clash_protection_loop_data): Use
1636 CONST_INT_P instead of explicit test. Verify object is a
1637 CONST_INT_P before looking at INTVAL.
1638 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1639 instead of explicit test.
1641 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1644 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1645 address instead of to r1 and r11.
1647 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1649 * config.gcc: Support "knm".
1650 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1651 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1653 * config/i386/i386.c (m_KNM): Define.
1654 (processor_target_table): Add "knm".
1656 (ix86_option_override_internal): Add "knm".
1657 (ix86_issue_rate): Add PROCESSOR_KNM.
1658 (ix86_adjust_cost): Ditto.
1659 (ia32_multipass_dfa_lookahead): Ditto.
1660 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1661 (fold_builtin_cpu): Add M_INTEL_KNM.
1662 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1663 (processor_type): Add PROCESSOR_KNM.
1664 * config/i386/x86-tune.def: Add m_KNM.
1665 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1667 2017-09-20 Richard Biener <rguenther@suse.de>
1669 PR tree-optimization/80213
1670 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1671 are allowed in empty BBs as well.
1672 (canonicalize_loop_closed_ssa): Also look for other complex
1674 (scop_detection::get_sese): Include the loop-closed PHI block
1676 (scop_detection::merge_sese): Remove code adding extra blocks.
1677 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1678 (build_scops): Assert the final returned scop is invalid.
1680 2017-09-20 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/82264
1683 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1685 (vn_phi_lookup): Likewise.
1686 (vn_phi_insert): Likewise.
1688 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1690 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1691 that fits into uhwi or shwi, add DW_AT_const_value regardless
1692 of early_dwarf without going through RTL, using add_AT_unsigned
1695 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1696 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1697 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1698 (DEBUG_MACRO_SECTION): Likewise.
1699 (DEBUG_LTO_MACRO_SECTION): Likewise.
1700 (DEBUG_STR_DWO_SECTION): Likewise.
1701 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1702 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1703 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1704 (DEBUG_LTO_LINE_STR_SECTION): Define.
1705 (init_sections_and_labels): Initialize debug_line_str_section
1706 variable. Initialize debug_loc_section for -gdwarf-5 to
1707 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1709 2017-09-20 Richard Biener <rguenther@suse.de>
1711 * graphite-sese-to-poly.c (extract_affine): Properly handle
1712 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1714 2017-09-20 Richard Biener <rguenther@suse.de>
1716 PR tree-optimization/81373
1717 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1718 Force SESE live-out defs to be handled even if they are
1721 2017-09-19 Jeff Law <law@redhat.com>
1723 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1724 nothing for stack adjustments with REG_STACK_CHECK.
1725 * sched-deps.c (parse_add_or_inc): Reject insns with
1726 REG_STACK_CHECK from dependency breaking.
1727 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1728 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1729 * reg-notes.def (STACK_CHECK): New note.
1731 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1732 (ix86_expand_prologue): Dump stack clash info as needed.
1733 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1735 * function.c (dump_stack_clash_frame_info): New function.
1736 * function.h (dump_stack_clash_frame_info): Prototype.
1737 (enum stack_clash_probes): New enum.
1739 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1740 flag_stack_clash_protection.
1741 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1742 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1743 (arm_frame_pointer_required): Likewise.
1744 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1745 (ia64_expand_prologue): Likewise.
1746 * config/mips/mips.c (mips_expand_prologue): Likewise.
1747 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1748 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1749 (sparc_flat_expand_prologue): Likewise.
1750 * config/spu/spu.c (spu_expand_prologue): Likewise.
1752 * explow.c: Include "params.h".
1753 (anti_adjust_stack_and_probe_stack_clash): New function.
1754 (get_stack_check_protect): Likewise.
1755 (compute_stack_clash_protection_loop_data): Likewise.
1756 (emit_stack_clash_protection_loop_start): Likewise.
1757 (emit_stack_clash_protection_loop_end): Likewise.
1758 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1759 Use anti_adjust_stack_and_probe_stack_clash.
1760 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1761 (emit_stack_clash_protection_loop_start): Likewise.
1762 (emit_stack_clash_protection_loop_end): Likewise.
1763 * rtl.h (get_stack_check_protect): Prototype.
1764 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1765 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1766 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1768 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1770 * doc/tm.texi: Rebuilt.
1771 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1772 get_stack_check_protect.
1773 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1774 * config/arm/arm.c (arm_expand_prologue): Likewise.
1775 (arm_frame_pointer_required): Likewise.
1776 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1777 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1778 * config/mips/mips.c (mips_expand_prologue): Likewise.
1779 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1780 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1781 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1782 (sparc_flat_expand_prologue): Likewise.
1784 * common.opt (-fstack-clash-protection): New option.
1785 * flag-types.h (enum stack_check_type): Note difference between
1786 -fstack-check= and -fstack-clash-protection.
1787 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1788 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1789 * toplev.c (process_options): Issue warnings/errors for cases
1790 not handled with -fstack-clash-protection.
1791 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1792 (-fstack-check): Note additional problem with -fstack-check=generic.
1793 Note that -fstack-check is primarily for Ada and refer users
1794 to -fstack-clash-protection for stack-clash-protection.
1795 Document new params for stack clash protection.
1797 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1799 * config/i386/i386.c (ix86_split_long_move): Do not handle
1800 address used for LEA in a special way.
1802 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1804 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1806 2017-09-19 Martin Sebor <msebor@redhat.com>
1809 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1810 of incompatible types.
1812 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1814 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1815 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1816 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1817 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1819 2017-09-19 Richard Biener <rguenther@suse.de>
1821 PR tree-optimization/82244
1822 * tree-vrp.c (remove_range_assertions): Do not propagate
1823 a constant to abnormals but replace the assert with a copy.
1825 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1827 PR rtl-optimization/57878
1828 PR rtl-optimization/68988
1829 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1830 avoidance test involving non_reload_pseudos. Move frequency test
1831 below the general fragmentation avoidance test.
1833 2017-09-19 Richard Biener <rguenther@suse.de>
1835 PR tree-optimization/69728
1836 * graphite-sese-to-poly.c (schedule_error): New global.
1837 (add_loop_schedule): Handle empty domain by failing the
1839 (build_original_schedule): Handle schedule_error.
1841 2017-09-19 Richard Biener <rguenther@suse.de>
1843 * graphite-scop-detection.c (scop_detection::can_represent_loop):
1844 Do not iterate to sibling loops but only to siblings of inner
1847 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
1850 * config/m68k/m68k.md (moveq feeding equality comparison): Check
1851 that the registers are different.
1853 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
1855 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
1856 to processor_model and "amdfam17h" to arch_names_table.
1857 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
1859 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1862 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
1864 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1865 Alan Hayward <alan.hayward@arm.com>
1866 David Sherwood <david.sherwood@arm.com>
1868 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
1870 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
1872 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
1873 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
1874 vect_schedule_slp_instance.
1875 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
1876 Update call to vect_slp_analyze_node_operations. Simplify return
1878 (vect_slp_analyze_bb_1): Update call accordingly.
1879 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
1880 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
1881 (vect_schedule_slp): Update call accordingly.
1883 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1884 Alan Hayward <alan.hayward@arm.com>
1885 David Sherwood <david.sherwood@arm.com>
1887 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
1888 with types that aren't in fact scalar.
1890 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1892 * tree-vect-slp.c (vect_record_max_nunits): New function,
1894 (vect_build_slp_tree_1): ...here.
1895 (vect_build_slp_tree_2): Call it for phis too.
1897 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1899 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
1900 to vect_get_vec_def_for_operand when getting the mask operand.
1902 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1903 Alan Hayward <alan.hayward@arm.com>
1904 David Sherwood <david.sherwood@arm.com>
1906 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
1909 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1910 Alan Hayward <alan.hayward@arm.com>
1911 David Sherwood <david.sherwood@arm.com>
1913 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
1914 calculation for vector booleans.
1916 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
1917 Alan Hayward <alan.hayward@arm.com>
1918 David Sherwood <david.sherwood@arm.com>
1920 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
1922 (vect_transform_stmt): ...here.
1923 (vect_analyze_stmt): Use it instead of calling
1924 vectorizable_live_operation directly.
1926 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
1928 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
1929 non-SIMT targets in acc vector loops.
1931 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
1933 * configure.ac: Add arc and check if assembler supports gdwarf2.
1934 * configure: Regenerate.
1936 2017-09-18 Richard Biener <rguenther@suse.de>
1938 PR tree-optimization/82220
1939 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
1940 epilogue niters from the min_profitable_iters compute.
1942 2017-09-18 Jakub Jelinek <jakub@redhat.com>
1945 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
1946 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
1947 (ix86_init_pic_reg): Revert 2017-09-01 changes.
1949 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
1952 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
1953 switching to a new text section.
1955 2017-09-18 Richard Biener <rguenther@suse.de>
1957 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
1959 (build_alias_set): Reject aliases with no access function.
1961 2017-09-18 Richard Biener <rguenther@suse.de>
1963 PR tree-optimization/79622
1964 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
1966 (build_cross_bb_scalars_use): Likewise.
1968 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
1970 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
1972 2017-09-18 Alan Modra <amodra@gmail.com>
1975 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
1976 stack_pointer_rtx for count 0. Update comments. Break up
1977 large rtl expression.
1979 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
1981 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
1982 Increase to 20 bytes.
1983 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
1984 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
1985 or avx version of the stub.
1987 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
1990 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
1991 compute the minimum stack alignment. Also update preferred stack
1992 boundary for leaf functions.
1994 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
1996 PR tree-optimization/82228
1997 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2000 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2002 * common/config/nds32/nds32-common.c
2003 (nds32_option_optimization_table): Refine formatting.
2004 (nds32_option_optimization_table): Use -fsched-pressure and
2005 -fomit-frame-pointer for specific optimization level.
2007 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2009 * config/nds32/nds32.c: Refine formatting and comments.
2010 * config/nds32/nds32.h: Likewise.
2011 * config/nds32/nds32.md: Likewise.
2012 * config/nds32/nds32-cost.c: Likewise.
2013 * config/nds32/nds32-isr.c: Likewise.
2014 * config/nds32/nds32-md-auxiliary.c: Likewise.
2015 * config/nds32/nds32-multiple.md: Likewise.
2016 * config/nds32/nds32-predicates.c: Likewise.
2018 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2019 Jakub Jelinek <jakub@redhat.com>
2021 Add support for -std=c++2a.
2022 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2024 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2026 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2029 * doc/extend.texi (Common Function Attributes): Add
2030 references to ARM, AArch64, and S/390 specific attributes.
2031 (Function Specific Option Pragmas): Add AArch64 and S/390
2032 to list of back ends that support the target pragma.
2034 2017-09-15 Nathan Sidwell <nathan@acm.org>
2036 * doc/standards.texi: Fix C++17 description. Update URLs for
2039 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2041 * common.opt (Wcast-align=strict): New warning option.
2042 * doc/invoke.texi: Document -Wcast-align=strict.
2044 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2046 * cgraph.h (cgraph_thunk_info): Add comments.
2047 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2048 assert for VIRTUAL_* arguments stricter.
2050 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2052 PR tree-optimization/71026
2053 * match.pd: Move RDIV patterns from fold-const.c
2054 * fold-const.c (distribute_real_division): Removed.
2055 (fold_binary_loc): Remove calls to distribute_real_divison.
2057 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2059 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2060 c++1z and gnu++1z as deprecated. Change other references to
2061 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2062 Change -Wc++1z-compat to -Wc++17-compat.
2063 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2064 * dwarf2out.c (highest_c_language): Handle C++17.
2065 (gen_compile_unit_die): Likewise.
2067 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2069 PR rtl-optimization/82192
2070 * combine.c (make_extraction): Don't look through non-paradoxical
2071 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2074 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2075 Alan Hayward <alan.hayward@arm.com>
2076 David Sherwood <david.sherwood@arm.com>
2078 * target.def (function_arg_offset): New hook.
2079 * targhooks.h (default_function_arg_offset): Declare.
2080 * targhooks.c (default_function_arg_offset): New function.
2081 * function.c (locate_and_pad_parm): Use
2082 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2083 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2084 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2085 * doc/tm.texi: Regenerate.
2086 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2087 * config/spu/spu.c (spu_function_arg_offset): New function.
2088 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2089 * system.h (FUNCTION_ARG_OFFSET): Poison.
2091 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2092 Alan Hayard <alan.hayward@arm.com>
2093 David Sherwood <david.sherwood@arm.com>
2095 * target.def (truly_noop_truncation): New hook.
2096 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2097 than TRULY_NOOP_TRUNCATION.
2098 * hooks.h (hook_bool_uint_uint_true): Declare.
2099 * hooks.c (hook_bool_uint_uint_true): New function.
2100 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2101 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2102 * doc/tm.texi: Regenerate.
2103 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2104 rather than TRULY_NOOP_TRUNCATION in comments.
2105 (simplify_comparison): Likewise.
2106 (record_truncated_value): Likewise.
2107 * expmed.c (extract_bit_field_1): Likewise.
2108 (extract_split_bit_field): Likewise.
2109 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2110 instead of TRULY_NOOP_TRUNCATION.
2111 * function.c (assign_parm_setup_block): Likewise.
2112 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2113 * rtlhooks.c: Include target.h.
2114 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2115 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2116 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2117 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2118 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2119 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2120 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2121 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2122 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2123 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2124 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2125 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2126 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2127 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2128 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2129 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2130 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2131 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2132 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2133 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2134 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2135 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2136 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2137 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2138 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2139 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2140 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2141 rather than TRULY_NOOP_TRUNCATION in comments.
2142 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2143 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2144 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2145 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2146 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2147 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2148 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2149 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2150 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2151 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2152 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2153 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2154 rather than TRULY_NOOP_TRUNCATION in comments.
2155 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2156 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2157 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2158 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2159 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2160 TRULY_NOOP_TRUNCATION condition.
2161 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2162 (TRULY_NOOP_TRUNCATION): Delete.
2163 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2164 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2165 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2166 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2167 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2168 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2169 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2170 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2171 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2172 rather than TRULY_NOOP_TRUNCATION in comments.
2173 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2174 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2175 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2176 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2177 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2178 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2180 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2183 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2184 (*cmp_ior): Likewise.
2185 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2186 (*ior_scc_scc_cmp): Likewise.
2187 (*and_scc_scc): Likewise.
2188 (*and_scc_scc_cmp): Likewise.
2190 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2191 Alan Hayard <alan.hayward@arm.com>
2192 David Sherwood <david.sherwood@arm.com>
2194 * target.def (can_change_mode_class): New hook.
2195 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2196 (hard_regno_nregs): Likewise.
2197 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2198 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2199 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2200 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2201 (LOAD_EXTEND_OP): Update accordingly.
2202 * doc/tm.texi: Regenerate.
2203 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2204 CANNOT_CHANGE_MODE_CLASS.
2205 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2206 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2207 * combine.c (simplify_set): Update accordingly.
2208 * emit-rtl.c (validate_subreg): Likewise.
2209 * recog.c (general_operand): Likewise.
2210 * regcprop.c (mode_change_ok): Likewise.
2211 * reload1.c (choose_reload_regs): Likewise.
2212 (inherit_piecemeal_p): Likewise.
2213 * rtlanal.c (simplify_subreg_regno): Likewise.
2214 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2215 instead of CANNOT_CHANGE_MODE_CLASS.
2216 (reload_cse_simplify_operands): Likewise.
2217 * reload.c (push_reload): Use targetm.can_change_mode_class
2218 instead of CANNOT_CHANGE_MODE_CLASS.
2219 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2220 REG_CANNOT_CHANGE_MODE_P.
2221 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2222 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2223 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2224 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2225 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2226 (arm_can_change_mode_class): New function.
2227 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2228 than CANNOT_CHANGE_MODE_CLASS in comments.
2229 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2230 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2231 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2232 (ix86_can_change_mode_class): ...this new function, inverting the
2233 sense of the return value.
2234 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2235 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2236 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2237 (ia64_can_change_mode_class): New function.
2238 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2239 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2240 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2241 (m32c_can_change_mode_class): ...this new function, inverting the
2242 sense of the return value.
2243 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2244 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2245 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2246 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2247 (mips_can_change_mode_class): ...this new function, inverting the
2248 sense of the return value.
2249 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2250 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2251 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2252 (msp430_can_change_mode_class): New function.
2253 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2254 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2255 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2256 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2257 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2258 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2259 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2260 (pa_cannot_change_mode_class): Replace with...
2261 (pa_can_change_mode_class): ...this new function, inverting the
2262 sense of the return value.
2263 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2264 than CANNOT_CHANGE_MODE_CLASS in comments.
2265 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2266 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2267 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2268 (pdp11_cannot_change_mode_class): Replace with...
2269 (pdp11_can_change_mode_class): ...this new function, inverting the
2270 sense of the return value.
2271 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2272 * config/powerpcspe/powerpcspe-protos.h
2273 (rs6000_cannot_change_mode_class_ptr): Delete.
2274 * config/powerpcspe/powerpcspe.c
2275 (rs6000_cannot_change_mode_class_ptr): Delete.
2276 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2277 (rs6000_option_override_internal): Assign to
2278 targetm.can_change_mode_class instead of
2279 rs6000_cannot_change_mode_class_ptr.
2280 (rs6000_cannot_change_mode_class): Replace with...
2281 (rs6000_can_change_mode_class): ...this new function, inverting the
2282 sense of the return value.
2283 (rs6000_debug_cannot_change_mode_class): Replace with...
2284 (rs6000_debug_can_change_mode_class): ...this new function.
2285 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2286 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2287 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2288 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2289 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2291 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2292 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2293 (rs6000_option_override_internal): Assign to
2294 targetm.can_change_mode_class instead of
2295 rs6000_cannot_change_mode_class_ptr.
2296 (rs6000_cannot_change_mode_class): Replace with...
2297 (rs6000_can_change_mode_class): ...this new function, inverting the
2298 sense of the return value.
2299 (rs6000_debug_cannot_change_mode_class): Replace with...
2300 (rs6000_debug_can_change_mode_class): ...this new function.
2301 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2302 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2303 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2304 (s390_can_change_mode_class): ...this new function, inverting the
2305 sense of the return value.
2306 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2307 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2308 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2309 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2310 (sh_cannot_change_mode_class): Replace with...
2311 (sh_can_change_mode_class): ...this new function, inverting the
2312 sense of the return value.
2313 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2314 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2315 (sparc_can_change_mode_class): New function.
2316 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2317 * config/spu/spu.c (spu_can_change_mode_class): New function.
2318 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2319 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2320 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2321 (visium_can_change_mode_class): New function.
2322 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2324 2017-09-15 Richard Biener <rguenther@suse.de>
2326 PR tree-optimization/82217
2327 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2328 but not undefined case.
2330 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2333 * postreload.c (reload_cse_simplify_operands): Skip
2334 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2336 2017-09-15 Richard Biener <rguenther@suse.de>
2338 PR tree-optimization/68823
2339 * graphite-scop-detection.c (build_alias_set): If we have a
2340 possible dependence check whether we can handle them by just
2341 looking at the DRs DR_ACCESS_FNs.
2342 (build_scops): If build_alias_set fails, fail the SCOP.
2344 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2346 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2347 to support float128 built-in functions that require the ISA 3.0
2349 (BU_FLOAT128_3_HW): Likewise.
2350 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2352 (FMAF128): Likewise.
2354 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2355 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2356 floating point instructions.
2357 (rs6000_invalid_builtin): Likewise.
2358 (rs6000_builtin_mask_names): Likewise.
2359 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2360 (RS6000_BTM_FLOAT128_HW): Likewise.
2361 (RS6000_BTM_COMMON): Likewise.
2362 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2364 * doc/extend.texi (RS/6000 built-in functions): Document the
2365 IEEE 128-bit floating point square root and fused multiply-add
2368 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2370 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2371 reg (r2) isn't in the set of registers defined in the prologue.
2373 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2374 Alan Hayward <alan.hayward@arm.com>
2375 David Sherwood <david.sherwood@arm.com>
2377 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2378 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2379 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2380 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2381 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2383 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2384 max_vectorization_factor.
2385 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2387 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2388 Alan Hayward <alan.hayward@arm.com>
2389 David Sherwood <david.sherwood@arm.com>
2391 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2392 (vect_worthwhile_without_simd_p): Declare.
2393 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2394 (vectorizable_reduction): Use it.
2395 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2396 (vectorizable_operation): Likewise.
2398 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2399 Alan Hayward <alan.hayward@arm.com>
2400 David Sherwood <david.sherwood@arm.com>
2402 * tree-vectorizer.h (vect_get_num_copies): New function.
2403 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2404 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2405 (vectorizable_induction): Likewise.
2406 (vectorizable_live_operation): Likewise.
2407 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2408 (vectorizable_bswap): Likewise.
2409 (vectorizable_call): Likewise.
2410 (vectorizable_conversion): Likewise.
2411 (vectorizable_assignment): Likewise.
2412 (vectorizable_shift): Likewise.
2413 (vectorizable_operation): Likewise.
2414 (vectorizable_store): Likewise.
2415 (vectorizable_load): Likewise.
2416 (vectorizable_condition): Likewise.
2417 (vectorizable_comparison): Likewise.
2418 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2420 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2421 Alan Hayward <alan.hayward@arm.com>
2422 David Sherwood <david.sherwood@arm.com>
2424 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2425 of vect_init_vector.
2427 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2428 Alan Hayward <alan.hayward@arm.com>
2429 David Sherwood <david.sherwood@arm.com>
2431 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2432 an inline wrapper that provides a location.
2433 (gimple_build_vector): Likewise.
2434 * gimple-fold.c (gimple_build_vector_from_val): New function.
2435 (gimple_build_vector): Likewise.
2436 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2437 functions to build the initial value. Always return a gimple value.
2438 (get_initial_defs_for_reduction): Likewise. Only compute
2440 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2441 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2442 (vectorizable_induction): Use gimple_build_vector rather than
2445 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2446 Alan Hayward <alan.hayward@arm.com>
2447 David Sherwood <david.sherwood@arm.com>
2449 * target.h (vec_perm_indices): New typedef.
2450 (auto_vec_perm_indices): Likewise.
2451 * optabs-query.h: Include target.h
2452 (can_vec_perm_p): Take a vec_perm_indices *.
2453 * optabs-query.c (can_vec_perm_p): Likewise.
2454 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2455 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2456 * tree-vect-generic.c (lower_vec_perm): Likewise.
2457 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2458 (vect_grouped_load_supported): Likewise.
2459 (vect_shift_permute_load_chain): Likewise.
2460 (vect_permute_store_chain): Use auto_vec_perm_indices.
2461 (vect_permute_load_chain): Likewise.
2462 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2463 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2464 Update uses of can_vec_perm_p.
2465 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2466 mode with a number of elements. Take a vec_perm_indices *.
2467 (vect_create_epilog_for_reduction): Update accordingly.
2468 Use auto_vec_perm_indices.
2469 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2470 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2471 (vect_transform_slp_perm_load): Likewise.
2472 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2473 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2474 (vect_gen_perm_mask_checked): Likewise.
2475 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2476 (vect_gen_perm_mask_checked): Likewise.
2477 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2478 (vectorizable_store): Likewise.
2479 (vectorizable_load): Likewise.
2480 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2481 (vectorizable_bswap): Likewise.
2483 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2484 Alan Hayward <alan.hayward@arm.com>
2485 David Sherwood <david.sherwood@arm.com>
2487 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2488 * tree.c (build_vector): Likewise.
2489 (build_vector_from_ctor): Update accordingly.
2490 (build_vector_from_val): Likewise.
2491 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2492 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2493 * tree-vect-generic.c (add_rshift): Likewise.
2494 (expand_vector_divmod): Likewise.
2495 (optimize_vector_constructor): Likewise.
2496 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2497 (vect_transform_slp_perm_load): Likewise.
2498 (vect_schedule_slp_instance): Likewise.
2499 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2500 (vectorizable_call): Likewise.
2501 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2502 * expmed.c (make_tree): Likewise.
2503 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2504 a vector passed to build_vector.
2505 (fold_convert_const): Likewise.
2506 (exact_inverse): Likewise.
2507 (fold_ternary_loc): Likewise.
2508 (fold_relational_const): Likewise.
2509 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2510 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2511 (const_unop): Likewise. Store the reduction accumulator in a
2512 variable rather than an array.
2513 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2514 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2515 the new vector, rather than constructing it after the input arrays.
2516 (native_interpret_vector): Use auto_vec<tree> when building
2517 a vector passed to build_vector. Add elements in order.
2518 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2519 auto_vec<tree> when building a vector passed to build_vector.
2520 (vect_create_epilog_for_reduction): Likewise.
2521 (vectorizable_induction): Likewise.
2522 (get_initial_def_for_reduction): Likewise. Fix indentation of
2524 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2526 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2527 passed to build_vector.
2529 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2530 Alan Hayward <alan.hayward@arm.com>
2531 David Sherwood <david.sherwood@arm.com>
2533 * tree-core.h (tree_base::u): Add an "nelts" field.
2534 (tree_vector): Use VECTOR_CST_NELTS as the length.
2535 * tree.c (tree_size): Likewise.
2536 (make_vector): Initialize VECTOR_CST_NELTS.
2537 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2538 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2539 TYPE_VECTOR_SUBPARTS.
2540 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2541 as the number of units, setting it from VECTOR_CST_NELTS.
2542 (const_vector_from_tree): Likewise.
2543 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2544 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2545 (fold_negate_expr_1): Likewise.
2546 (fold_convert_const): Likewise.
2547 (const_binop): Likewise. Differentiate the number of output and
2549 (const_unop): Likewise.
2550 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2551 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2552 in cases that did the opposite.
2554 2017-09-14 Richard Biener <rguenther@suse.de>
2556 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2559 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2561 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2563 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2566 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2567 if and where to split a bb, except for splitting before debug insn
2568 sequences followed by non-label real insn. Delete debug insns
2569 in between basic blocks.
2571 * combine.c (make_compound_operation_int): Formatting fixes.
2573 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2574 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2575 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2576 * config/sol2.h (LINK_EH_SPEC): Likewise.
2577 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2578 * config/s390/linux.h (LINK_SPEC): Likewise.
2579 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2580 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2581 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2582 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2583 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2584 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2585 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2586 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2588 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2590 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2592 (ENDFILE_LINUX_SPEC): Likewise.
2593 (LINK_EH_SPEC): Likewise.
2594 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2595 (LINK_OS_LINUX_SPEC32): Likewise.
2596 (LINK_OS_LINUX_SPEC64): Likewise.
2597 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2598 (LINK_OS_LINUX_SPEC): Likewise.
2600 2017-09-13 Martin Liska <mliska@suse.cz>
2603 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2604 CASE_HIGH is NULL_TREE.
2606 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2607 Alan Hayward <alan.hayward@arm.com>
2608 David Sherwood <david.sherwood@arm.com>
2610 * target.def (secondary_memory_needed): New hook.
2611 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2612 instead of SECONDARY_MEMORY_NEEDED.
2613 (secondary_memory_needed_mode): Likewise.
2614 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2615 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2616 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2617 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2618 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2619 * doc/tm.texi: Regenerate.
2620 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2621 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2622 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2623 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2624 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2625 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2626 mode argument first and change the reg_class arguments to reg_class_t.
2627 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2628 Make static. Update the call to inline_secondary_memory_needed.
2629 (ix86_register_move_cost): Update the call to
2630 inline_secondary_memory_needed.
2631 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2632 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2634 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2635 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2637 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2638 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2639 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2640 and match hook interface. Add comment from mips.h.
2641 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2642 * config/mmix/mmix.md (truncdfsf2): Refer to
2643 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2645 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2646 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2647 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2648 (pa_secondary_memory_needed): New function.
2649 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2650 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2651 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2652 (pdp11_secondary_memory_needed): Make static and match hook interface.
2653 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2654 * config/powerpcspe/powerpcspe-protos.h
2655 (rs6000_secondary_memory_needed_ptr): Delete.
2656 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2658 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2659 (rs6000_option_override_internal): Assign to
2660 targetm.secondary_memory_needed rather than
2661 rs6000_secondary_memory_needed_ptr.
2662 (rs6000_secondary_memory_needed): Match hook interface.
2663 (rs6000_debug_secondary_memory_needed): Likewise.
2664 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2665 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2666 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2667 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2668 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2669 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2671 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2672 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2673 (rs6000_option_override_internal): Assign to
2674 targetm.secondary_memory_needed rather than
2675 rs6000_secondary_memory_needed_ptr.
2676 (rs6000_secondary_memory_needed): Match hook interface.
2677 (rs6000_debug_secondary_memory_needed): Likewise.
2678 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2679 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2680 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2681 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2682 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2683 (sparc_secondary_memory_needed): New function.
2684 * lra-constraints.c (check_and_process_move): Refer to
2685 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2687 (curr_insn_transform): Likewise.
2688 (process_alt_operands): Use targetm.secondary_memory_needed
2689 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2690 (check_secondary_memory_needed_p): Likewise.
2691 (choose_split_class): Likewise.
2692 * reload.c: Unconditionally include code that was previously
2693 conditional on SECONDARY_MEMORY_NEEDED.
2694 (push_secondary_reload): Use targetm.secondary_memory_needed
2695 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2696 (push_reload): Likewise.
2697 * reload1.c: Unconditionally include code that was previously
2698 conditional on SECONDARY_MEMORY_NEEDED.
2699 (choose_reload_regs): Use targetm.secondary_memory_needed
2700 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2701 (gen_reload): Likewise.
2702 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2704 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2705 Alan Hayward <alan.hayward@arm.com>
2706 David Sherwood <david.sherwood@arm.com>
2708 * target.def (secondary_memory_needed_mode): New hook:
2709 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2710 * targhooks.h (default_secondary_memory_needed_mode): New function.
2711 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2712 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2713 * doc/tm.texi: Regenerate.
2714 * lra-constraints.c (check_and_process_move): Use
2715 targetm.secondary_memory_needed_mode instead of
2716 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2717 (curr_insn_transform): Likewise.
2718 * reload.c (get_secondary_mem): Likewise.
2719 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2720 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2722 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2723 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2724 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2725 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2726 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2728 * config/powerpcspe/powerpcspe-protos.h
2729 (rs6000_secondary_memory_needed_mode): Delete.
2730 * config/powerpcspe/powerpcspe.c
2731 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2732 (rs6000_secondary_memory_needed_mode): Make static.
2733 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2734 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2736 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2738 (rs6000_secondary_memory_needed_mode): Make static.
2739 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2740 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2741 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2742 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2743 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2745 (sparc_secondary_memory_needed_mode): New function.
2746 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2748 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2750 * config/aarch64/constraints.md (Umq): New constraint.
2751 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2753 (mov<mode>): Update condition.
2755 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2757 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2758 when bitposition is the same.
2760 2017-09-13 Richard Biener <rguenther@suse.de>
2762 * dwarf2out.c (output_die_symbol): Remove.
2763 (output_die): Do not output a DIEs symbol.
2765 2017-09-13 Richard Biener <rguenther@suse.de>
2768 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2769 default-def to avoid breaking iterator update with the weird
2770 interaction with cgraph_update_edges_for_call_stmt_node.
2772 2017-09-13 Richard Biener <rguenther@suse.de>
2774 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2775 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2776 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2777 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2779 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2781 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2782 Disable pc relative literal load irrespective of
2783 TARGET_FIX_ERR_A53_84341 for default.
2785 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2787 * config/sparc/sparc.c (output_return): Output the source location of
2788 the insn in the delay slot, if any.
2789 (output_sibcall): Likewise.
2791 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2794 * common.opt (-static-pie): New alias.
2795 (shared): Negate static-pie.
2796 (-no-pie): Update help text.
2798 (static-pie): New option.
2799 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2800 -static-pie support.
2801 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2802 (LINK_EH_SPEC): Likewise.
2803 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2804 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2805 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2806 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2807 (init_gcc_specs): Likewise.
2808 (init_spec): Likewise.
2809 (display_help): Update help message for -pie.
2810 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2813 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2815 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2816 (movdi_aarch64): Likewise.
2817 (movti_aarch64): Likewise.
2819 2017-09-12 Simon Wright <simon@pushface.org>
2822 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2823 calculation of the minor version, always output as 0.
2825 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2828 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2829 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2830 on it early, rather than manual conversion late. For
2831 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
2832 instead of performing manual conversion.
2834 2017-09-12 Carl Love <cel@us.ibm.com>
2836 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
2837 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
2839 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2840 VMULOSW): Add definitions.
2841 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2842 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2843 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2844 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2845 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2847 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2849 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
2851 (movti_aarch64): Likewise.
2852 (movdf_aarch64): Likewise.
2853 (movtf_aarch64): Likewise.
2854 (load_pairdi): Likewise.
2855 (store_pairdi): Likewise.
2856 (load_pairdf): Likewise.
2857 (store_pairdf): Likewise.
2858 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
2859 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
2860 (ldr_got_small_<mode>): Likewise.
2861 (ldr_got_small_28k_<mode>): Likewise.
2862 (ldr_got_tiny): Likewise.
2863 * config/aarch64/iterators.md (ldst_sz): New.
2864 (ldpstp_sz): Likewise.
2865 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
2867 (thunderx_load): Split load_8 to load_16.
2868 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
2870 (thunderx2t99_storepair_basic): Split store_8 to store_16.
2871 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
2872 (xgene1_store_pair): Split store_8 to store_16.
2873 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
2874 (falkor_st_0_st_sd): Split store_8 to store_16.
2876 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
2878 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
2879 and store1/2/3/4 to store_4/8/12/16.
2880 * config/aarch64/aarch64.md: Update for rename.
2881 * config/arm/arm.md: Likewise.: Likewise.
2882 * config/arm/arm.c: Likewise.
2883 * config/arm/thumb1.md: Likewise.
2884 * config/arm/thumb2.md: Likewise.
2885 * config/arm/vfp.md: Likewise.
2886 * config/arm/arm-generic.md: Likewise.
2887 * config/arm/arm1020e.md: Likewise.
2888 * config/arm/arm1026ejs.md: Likewise.
2889 * config/arm/arm1136jfs.md: Likewise.
2890 * config/arm/arm926ejs.md: Likewise.
2891 * config/arm/cortex-a15.md: Likewise.
2892 * config/arm/cortex-a17.md: Likewise.
2893 * config/arm/cortex-a5.md: Likewise.
2894 * config/arm/cortex-a53.md: Likewise.
2895 * config/arm/cortex-a57.md: Likewise.
2896 * config/arm/cortex-a7.md: Likewise.
2897 * config/arm/cortex-a8.md: Likewise.
2898 * config/arm/cortex-a9.md: Likewise.
2899 * config/arm/cortex-m4.md: Likewise.
2900 * config/arm/cortex-m7.md: Likewise.
2901 * config/arm/cortex-r4.md: Likewise.
2902 * config/arm/exynos-m1.md: Likewise.
2903 * config/arm/fa526.md: Likewise.
2904 * config/arm/fa606te.md: Likewise.
2905 * config/arm/fa626te.md: Likewise.
2906 * config/arm/fa726te.md: Likewise.
2907 * config/arm/fmp626.md: Likewise.
2908 * config/arm/iwmmxt.md: Likewise.
2909 * config/arm/ldmstm.md: Likewise.
2910 * config/arm/marvell-pj4.md: Likewise.
2911 * config/arm/xgene1.md: Likewise.
2912 * config/aarch64/thunderx.md: Likewise.
2913 * config/aarch64/thunderx2t99.md: Likewise.
2914 * config/aarch64/falkor.md: Likewise.
2916 2017-09-12 Martin Liska <mliska@suse.cz>
2918 * attribs.c (private_lookup_attribute): New function.
2919 * attribs.h (private_lookup_attribute): Declared here.
2920 (lookup_attribute): Called from this place.
2922 2017-09-12 Richard Biener <rguenther@suse.de>
2924 PR tree-optimization/82157
2925 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
2926 stmts with side-effects.
2928 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
2929 Alan Hayward <alan.hayward@arm.com>
2930 David Sherwood <david.sherwood@arm.com>
2932 * target.def (hard_regno_nregs): New hook.
2933 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
2934 * targhooks.h (default_hard_regno_nregs): Declare.
2935 * targhooks.c (default_hard_regno_nregs): New function.
2936 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
2937 (TARGET_HARD_REGNO_NREGS): ...this hook.
2938 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
2939 (CLASS_MAX_NREGS): Likewise.
2940 * doc/tm.texi: Regenerate.
2941 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
2942 instead of HARD_REGNO_NREGS.
2943 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
2944 HARD_REGNO_NREGS in the comment.
2945 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
2946 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
2947 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
2948 Return an unsigned int.
2949 (TARGET_HARD_REGNO_NREGS): Redefine.
2950 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
2951 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
2952 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
2953 (arc_hard_regno_nregs): New function.
2954 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
2955 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
2956 (arm_hard_regno_nregs): New function.
2957 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
2958 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
2959 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
2960 (TARGET_HARD_REGNO_NREGS): Redefine.
2961 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
2962 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
2963 (HARD_REGNO_NREGS): Delete.
2964 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
2965 (cr16_hard_regno_nregs): New function.
2966 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
2967 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
2968 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
2969 (cris_hard_regno_nregs): New function.
2970 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
2971 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
2972 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
2973 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
2974 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
2975 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
2976 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
2977 (frv_hard_regno_nregs): Make static. Take and return an
2979 (frv_class_max_nregs): Remove outdated copy of documentation.
2980 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
2981 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
2982 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
2983 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
2984 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
2985 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
2986 (TARGET_HARD_REGNO_NREGS): Redefine.
2987 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
2988 (CLASS_MAX_NREGS): Update comment.
2989 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
2990 (ia64_hard_regno_nregs): New function.
2991 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
2992 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
2993 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
2994 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
2995 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
2997 (m32c_hard_regno_nregs): Likewise. Make static.
2998 (TARGET_HARD_REGNO_NREGS): Redefine.
2999 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3000 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3001 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3002 (m68k_hard_regno_nregs): New function.
3003 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3004 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3005 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3006 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3007 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3008 Take and return an unsigned int.
3009 (TARGET_HARD_REGNO_NREGS): Redefine.
3010 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3011 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3012 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3013 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3014 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3015 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3016 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3017 (msp430_hard_regno_nregs): Make static. Take and return an
3019 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3020 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3021 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3022 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3023 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3024 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3025 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3026 (TARGET_HARD_REGNO_NREGS): Redefine.
3027 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3028 (PA_HARD_REGNO_NREGS): ...this.
3029 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3030 (PA_HARD_REGNO_NREGS): ...this.
3031 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3032 (pa_hard_regno_nregs): New function.
3033 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3034 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3035 (pdp11_hard_regno_nregs): New function.
3036 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3037 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3038 (rs6000_hard_regno_nregs_hook): New function.
3039 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3040 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3041 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3042 Take and return an unsigned int. Move earlier in file.
3043 (TARGET_HARD_REGNO_NREGS): Redefine.
3044 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3045 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3046 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3047 (rl78_hard_regno_nregs): Make static. Take and return an
3049 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3050 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3051 (rs6000_hard_regno_nregs_hook): New function.
3052 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3053 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3054 (TARGET_HARD_REGNO_NREGS): Redefine.
3055 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3056 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3057 instead of HARD_REGNO_NREGS.
3058 (s390_hard_regno_nregs): New function.
3059 (s390_hard_regno_mode_ok): Add comment from s390.h.
3060 (TARGET_HARD_REGNO_NREGS): Redefine.
3061 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3062 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3063 (sh_hard_regno_nregs): New function.
3064 (sh_pass_in_reg_p): Use it.
3065 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3066 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3067 (sparc_hard_regno_nregs): New function.
3068 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3069 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3070 (spu_function_arg_advance): Use it, supplying a valid register number.
3071 (TARGET_HARD_REGNO_NREGS): Redefine.
3072 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3073 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3074 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3075 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3076 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3077 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3078 (CLASS_MAX_NREGS): Remove copy of old documentation.
3079 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3080 (visium_hard_regno_nregs): New function.
3081 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3082 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3083 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3084 xtensa_hard_regno_nregs): New function.
3085 * system.h (HARD_REGNO_NREGS): Poison.
3087 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3089 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3090 hard_regno_nregs instead of HARD_REGNO_NREGS.
3091 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3092 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3093 (c6x_expand_epilogue): Likewise.
3094 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3095 (frv_read_iacc_argument): Likewise.
3096 * config/sh/sh.c: Include regs.h.
3097 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3098 (regs_used): Likewise.
3099 (output_stack_adjust): Likewise.
3100 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3101 * expmed.c: Include regs.h.
3102 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3103 * ree.c: Include regs.h.
3104 (combine_reaching_defs): Use hard_regno_nregs instead of
3106 (add_removable_extension): Likewise.
3108 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3110 * regs.h (hard_regno_nregs): Turn into a function.
3111 (end_hard_regno): Update accordingly.
3112 * caller-save.c (setup_save_areas): Likewise.
3113 (save_call_clobbered_regs): Likewise.
3114 (replace_reg_with_saved_mem): Likewise.
3115 (insert_restore): Likewise.
3116 (insert_save): Likewise.
3117 * combine.c (can_change_dest_mode): Likewise.
3118 (move_deaths): Likewise.
3119 (distribute_notes): Likewise.
3120 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3121 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3122 (rs6000_split_multireg_move): Likewise.
3123 (rs6000_register_move_cost): Likewise.
3124 (rs6000_memory_move_cost): Likewise.
3125 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3126 (rs6000_split_multireg_move): Likewise.
3127 (rs6000_register_move_cost): Likewise.
3128 (rs6000_memory_move_cost): Likewise.
3129 * cselib.c (cselib_reset_table): Likewise.
3130 (cselib_lookup_1): Likewise.
3131 * emit-rtl.c (set_mode_and_regno): Likewise.
3132 * function.c (aggregate_value_p): Likewise.
3133 * ira-color.c (setup_profitable_hard_regs): Likewise.
3134 (check_hard_reg_p): Likewise.
3135 (calculate_saved_nregs): Likewise.
3136 (assign_hard_reg): Likewise.
3137 (improve_allocation): Likewise.
3138 (calculate_spill_cost): Likewise.
3139 * ira-emit.c (modify_move_list): Likewise.
3140 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3141 (ira_hard_reg_in_set_p): Likewise.
3142 * ira.c (setup_reg_mode_hard_regset): Likewise.
3143 (clarify_prohibited_class_mode_regs): Likewise.
3144 (check_allocation): Likewise.
3145 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3146 (lra_setup_reg_renumber): Likewise.
3147 (setup_try_hard_regno_pseudos): Likewise.
3148 (spill_for): Likewise.
3149 (assign_hard_regno): Likewise.
3150 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3151 * lra-constraints.c (in_class_p): Likewise.
3152 (lra_constraint_offset): Likewise.
3153 (simplify_operand_subreg): Likewise.
3154 (lra_constraints): Likewise.
3155 (split_reg): Likewise.
3156 (split_if_necessary): Likewise.
3157 (invariant_p): Likewise.
3158 (inherit_in_ebb): Likewise.
3159 * lra-lives.c (process_bb_lives): Likewise.
3160 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3161 (get_hard_regs): Likewise.
3162 (do_remat): Likewise.
3163 * lra-spills.c (assign_spill_hard_regs): Likewise.
3164 * mode-switching.c (create_pre_exit): Likewise.
3165 * postreload.c (reload_combine_recognize_pattern): Likewise.
3166 * recog.c (peep2_find_free_register): Likewise.
3167 * regcprop.c (kill_value_regno): Likewise.
3168 (set_value_regno): Likewise.
3169 (copy_value): Likewise.
3170 (maybe_mode_change): Likewise.
3171 (find_oldest_value_reg): Likewise.
3172 (copyprop_hardreg_forward_1): Likewise.
3173 * regrename.c (check_new_reg_p): Likewise.
3174 (regrename_do_replace): Likewise.
3175 * reload.c (push_reload): Likewise.
3176 (combine_reloads): Likewise.
3177 (find_dummy_reload): Likewise.
3178 (operands_match_p): Likewise.
3179 (find_reloads): Likewise.
3180 (find_equiv_reg): Likewise.
3181 (reload_adjust_reg_for_mode): Likewise.
3182 * reload1.c (count_pseudo): Likewise.
3183 (count_spilled_pseudo): Likewise.
3184 (find_reg): Likewise.
3185 (clear_reload_reg_in_use): Likewise.
3186 (free_for_value_p): Likewise.
3187 (allocate_reload_reg): Likewise.
3188 (choose_reload_regs): Likewise.
3189 (reload_adjust_reg_for_temp): Likewise.
3190 (emit_reload_insns): Likewise.
3191 (delete_output_reload): Likewise.
3192 * rtlanal.c (subreg_get_info): Likewise.
3193 * sched-deps.c (sched_analyze_reg): Likewise.
3194 * sel-sched.c (init_regs_for_mode): Likewise.
3195 (mark_unavailable_hard_regs): Likewise.
3196 (choose_best_reg_1): Likewise.
3197 (verify_target_availability): Likewise.
3198 * valtrack.c (dead_debug_insert_temp): Likewise.
3199 * var-tracking.c (track_loc_p): Likewise.
3200 (emit_note_insn_var_location): Likewise.
3201 * varasm.c (make_decl_rtl): Likewise.
3202 * reginfo.c (choose_hard_reg_mode): Likewise.
3203 (init_reg_modes_target): Refer directly to
3204 this_target_regs->x_hard_regno_nregs.
3206 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3208 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3209 instead of hard_regno_nregs.
3211 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3213 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3214 end_hard_regno instead of hard_regno_nregs.
3215 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3216 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3217 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3218 * ira-color.c (improve_allocation): Likewise.
3219 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3220 * lra-lives.c (mark_regno_live): Likewise.
3221 (mark_regno_dead): Likewise.
3222 * lra-remat.c (operand_to_remat): Likewise.
3223 * lra.c (collect_non_operand_hard_regs): Likewise.
3224 * postreload.c (reload_combine_note_store): Likewise.
3225 (move2add_valid_value_p): Likewise.
3226 * reload.c (regno_clobbered_p): Likewise.
3228 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3230 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3232 * config/v850/v850.c (v850_reorg): Likewise.
3233 * reload.c (refers_to_regno_for_reload_p): Likewise.
3234 (find_equiv_reg): Likewise.
3235 * reload1.c (reload_reg_reaches_end_p): Likewise.
3237 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3239 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3241 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3242 * config/arm/arm.c (output_move_neon): Likewise.
3243 (arm_attr_length_move_neon): Likewise.
3244 (neon_split_vcombine): Likewise.
3245 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3246 (c6x_mark_reg_written): Likewise.
3247 (c6x_dwarf_register_span): Likewise.
3248 * config/i386/i386.c (ix86_save_reg): Likewise.
3249 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3250 (rws_access_reg): Likewise.
3251 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3252 * mode-switching.c (create_pre_exit): Likewise.
3253 * ree.c (combine_reaching_defs): Likewise.
3254 (add_removable_extension): Likewise.
3255 * regcprop.c (find_oldest_value_reg): Likewise.
3256 (copyprop_hardreg_forward_1): Likewise.
3257 * reload.c (reload_inner_reg_of_subreg): Likewise.
3258 (push_reload): Likewise.
3259 (combine_reloads): Likewise.
3260 (find_dummy_reload): Likewise.
3261 (reload_adjust_reg_for_mode): Likewise.
3262 * reload1.c (find_reload_regs): Likewise.
3263 (forget_old_reloads_1): Likewise.
3264 (reload_reg_free_for_value_p): Likewise.
3265 (reload_adjust_reg_for_temp): Likewise.
3266 (emit_reload_insns): Likewise.
3267 (delete_output_reload): Likewise.
3268 * sel-sched.c (choose_best_reg_1): Likewise.
3269 (choose_best_pseudo_reg): Likewise.
3271 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3272 Alan Hayward <alan.hayward@arm.com>
3273 David Sherwood <david.sherwood@arm.com>
3275 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3276 * target.def (slow_unaligned_access): New hook.
3277 * targhooks.h (default_slow_unaligned_access): Declare.
3278 * targhooks.c (default_slow_unaligned_access): New function.
3279 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3280 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3281 * doc/tm.texi: Regenerate.
3282 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3283 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3284 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3286 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3287 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3289 (rs6000_slow_unaligned_access): New function.
3290 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3291 (expand_block_compare): Likewise.
3292 (expand_strn_compare): Likewise.
3293 (rs6000_rtx_costs): Likewise.
3294 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3295 (riscv_slow_unaligned_access): Likewise.
3296 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3297 (riscv_slow_unaligned_access_p): ...this and make static.
3298 (riscv_option_override): Update accordingly.
3299 (riscv_slow_unaligned_access): New function.
3300 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3301 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3302 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3303 (rs6000_slow_unaligned_access): New function.
3304 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3305 (rs6000_rtx_costs): Likewise.
3306 * config/rs6000/rs6000-string.c (expand_block_compare)
3307 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3308 of SLOW_UNALIGNED_ACCESS.
3309 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3310 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3311 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3312 of SLOW_UNALIGNED_ACCESS.
3313 * expmed.c (simple_mem_bitfield_p): Likewise.
3314 * expr.c (alignment_for_piecewise_move): Likewise.
3315 (emit_group_load_1): Likewise.
3316 (emit_group_store): Likewise.
3317 (copy_blkmode_from_reg): Likewise.
3318 (emit_push_insn): Likewise.
3319 (expand_assignment): Likewise.
3320 (store_field): Likewise.
3321 (expand_expr_real_1): Likewise.
3322 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3323 * lra-constraints.c (simplify_operand_subreg): Likewise.
3324 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3325 * gimple-ssa-store-merging.c: Likewise in block comment at start
3327 * tree-ssa-strlen.c: Include target.h.
3328 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3329 of SLOW_UNALIGNED_ACCESS.
3330 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3332 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3334 PR rtl-optimization/82185
3335 * expmed.c (emit_store_flag_int): Only test tem if it has been
3338 2017-09-12 Richard Biener <rguenther@suse.de>
3341 * match.pd ((FTYPE) N CMP CST): Fix typo.
3343 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3345 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3347 (mips_near_type_p): Add 'short_call' attribute as a synonym
3349 * doc/extend.texi (short_call): Document new function attribute.
3351 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3354 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3355 assertion check that in the condition.
3356 (get_atomic_generic_size): Likewise. Before testing if parameter
3357 has pointer type, if it has array type, call for C++
3358 default_conversion to perform array-to-pointer conversion.
3360 2017-09-12 Richard Biener <rguenther@suse.de>
3362 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3363 for operations we cannot scalarize.
3365 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3367 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3368 vectors heap vectors. Clean up comments.
3369 Make visited_bbs a reference.
3370 (profitable_jump_thread_path): Make GC
3371 vectors heap vectors. Clean up comments.
3373 (convert_and_register_jump_thread_path): Make GC vectors heap
3375 (check_subpath_and_update_thread_path): Same. Clean up comments.
3376 Make visited_bbs a reference.
3377 (handle_phi): Abstract common code to to
3378 register_jump_thread_path_if_profitable.
3379 Rename VAR_BB to DEF_BB.
3381 Make GC vectors heap vectors.
3382 Make visited_bbs a reference.
3383 (handle_assignment): Same.
3384 (register_jump_thread_path_if_profitable): New.
3385 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3387 Make GC vectors heap vectors. Clean up comments.
3388 Make visited_bbs a reference.
3389 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3390 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3393 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3396 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3397 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3399 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3401 Revert r251800 and r251799.
3403 2017-09-11 Martin Jambor <mjambor@suse.cz>
3406 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3407 arguments in advance.
3408 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3409 use it to find predecessor edges.
3410 (naive_outof_ssa): Collect vector of predecessors.
3412 2017-09-08 Jason Merrill <jason@redhat.com>
3414 PR c++/70029 - ICE with ref-qualifier and -flto
3415 * langhooks.h (struct lang_hooks_for_types): Add
3416 copy_lang_qualifiers.
3417 * attribs.c (build_type_attribute_qual_variant): Use it.
3418 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3420 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3421 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3423 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3426 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3427 (*mulsi3_sp64): New instruction.
3428 (mulsi3): New expander.
3430 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3432 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3434 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3436 * sancov.c: Include memmodel.h.
3438 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3441 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3444 2017-09-07 Carl Love <cel@us.ibm.com>
3446 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3447 the sldi instruction.
3449 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3451 * sancov.c: Include tm_p.h.
3453 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3456 * output.h (switch_to_other_text_partition): New declaration.
3457 * varasm.c (switch_to_other_text_partition): New function.
3458 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3459 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3460 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3461 to the other text partition before emitting LCL label and switch back
3462 after emitting the word after it.
3464 2017-09-07 Richard Biener <rguenther@suse.de>
3466 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3467 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3468 critical edges here, after loop init.
3469 (pass_data_pre): Remove PROP_no_crit_edges flags.
3470 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3471 for valueization of call args to avoid leaking VN_TOP.
3472 (visit_use): Assert we do not visit default defs.
3473 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3474 Use error_mark_node to more easily detect leaking VN_TOP.
3475 All default-defs are varying, not VN_TOP. Mark them visited.
3476 (run_scc_vn): Make code match comment.
3478 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3480 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3481 OPTION_MASK_FLOAT128_KEYWORD.
3482 (POWERPC_MASKS): Likewise.
3483 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3484 support for the -mfloat128-type option, and make -mfloat128
3485 default on PowerPC Linux systems. Define or undefine
3486 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3487 Define __float128 to be __ieee128 if IEEE 128-bit support is
3488 enabled, or undefine it.
3489 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3490 Delete defining __FLOAT128_TYPE__.
3491 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3492 -mfloat128-type option and make -mfloat128 default on PowerPC
3494 (TARGET_FLOAT128_TYPE): Likewise.
3495 (-mfloat128-type): Likewise.
3496 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3497 Delete the -mfloat128-type option and make -mfloat128 default on
3498 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3499 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3500 128-bit floating point is enabled. Change tests from using
3501 -mfloat128-type to -mfloat128.
3502 (rs6000_mangle_type): Use the correct mangling for the __float128
3503 type even if normal long double is restricted to 64-bits.
3504 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3506 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3507 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3508 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3510 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3511 documentation for -mfloat128.
3513 2017-09-06 Olivier Hainque <hainque@adacore.com>
3515 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3517 2017-09-06 Wish Wu <wishwu007@gmail.com>
3518 Jakub Jelinek <jakub@redhat.com>
3520 * asan.c (initialize_sanitizer_builtins): Add
3521 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3522 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3523 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3524 BT_FN_VOID_UINT64_PTR variables.
3525 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3526 (BT_FN_VOID_UINT16_UINT16): Likewise.
3527 (BT_FN_VOID_UINT32_UINT32): Likewise.
3528 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3529 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3530 (BT_FN_VOID_UINT64_PTR): Likewise.
3531 * common.opt (flag_sanitize_coverage): New variable.
3532 (fsanitize-coverage=trace-pc): Remove.
3533 (fsanitize-coverage=): Add.
3534 * flag-types.h (enum sanitize_coverage_code): New enum.
3535 * fold-const.c (fold_range_test): Disable non-short-circuit
3536 optimization if flag_sanitize_coverage.
3537 (fold_truth_andor): Likewise.
3538 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3539 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3540 (coverage_sanitizer_opts): New array.
3541 (get_closest_sanitizer_option): Add OPTS argument, handle also
3542 OPT_fsanitize_coverage_.
3543 (parse_sanitizer_options): Adjusted to also handle
3544 OPT_fsanitize_coverage_.
3545 (common_handle_option): Add OPT_fsanitize_coverage_.
3546 * sancov.c (instrument_comparison, instrument_switch): New function.
3547 (sancov_pass): Add trace-cmp support.
3548 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3549 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3550 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3551 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3552 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3553 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3554 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3555 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3556 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3557 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3559 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3561 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3562 error. Only quit immediately if parsing is complete.
3563 (BEGIN): Initialize fatal_err and parse_done.
3564 (begin fpu, end fpu): Check number of arguments.
3565 (begin arch, end arch): Likewise.
3566 (begin cpu, end cpu): Likewise.
3567 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3568 (optalias): Likewise.
3570 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3572 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3573 * config/arm/arm-isa.h: Delete. Move definitions to ...
3574 * arm-cpus.in: ... here. Use new feature and fgroup values.
3575 * config/arm/arm.c (arm_option_override): Use lower case for feature
3577 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3578 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3579 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3581 (print_isa_bits_for): New function.
3582 (gen_isa): New function.
3583 (gen_comm_data): Use print_isa_bits_for.
3584 (define feature): New keyword.
3585 (define fgroup): New keyword.
3586 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3587 (arm-isa.h): Add rule to generate file.
3588 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3589 case for feature bit names.
3591 2017-09-06 Richard Biener <rguenther@suse.de>
3593 * tree-ssa-pre.c (NECESSARY): Remove.
3594 (create_expression_by_pieces): Do not touch pass-local flags.
3595 (insert_into_preds_of_block): Likewise.
3596 (do_pre_regular_insertion): Likewise.
3597 (eliminate_insert): Likewise.
3598 (eliminate_dom_walker::before_dom_children): Likewise.
3599 (fini_eliminate): Do not look at inserted_exprs.
3600 (mark_operand_necessary): Remove.
3601 (remove_dead_inserted_code): Replace with simple work-list
3602 algorithm based on inserted_exprs and SSA uses.
3603 (pass_pre::execute): Re-order fini_eliminate and
3604 remove_dead_inserted_code.
3606 2017-09-06 Olivier Hainque <hainque@adacore.com>
3608 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3609 for VxWorks 7. Adjust surrounding comments.
3611 2017-09-06 Richard Biener <rguenther@suse.de>
3613 * gimple-ssa-strength-reduction.c
3614 (find_candidates_dom_walker::before_dom_children): Also allow
3617 2017-09-06 Richard Biener <rguenther@suse.de>
3619 PR tree-optimization/82108
3620 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3621 for gap in the non-permutation SLP case.
3623 2017-09-06 Martin Jambor <mjambor@suse.cz>
3625 PR tree-optimization/82078
3626 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3627 add_access_to_work_queue...
3628 (build_accesses_from_assign): ...here.
3629 (propagate_all_subaccesses): Make sure racc is the group
3630 representative, if there is one.
3632 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3635 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3638 2017-09-06 Richard Biener <rguenther@suse.de>
3640 * gimple-ssa-strength-reduction.c
3641 (find_candidates_dom_walker::before_doom_children): Use a
3642 type and not a mode check.
3644 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3647 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3649 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3651 2017-09-05 Jeff Law <law@redhat.com>
3653 PR tree-optimization/64910
3654 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3655 cases where we have 3 or more operands.
3657 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3660 * omp-low.c (lower_omp_for): Recompute tree invariant if
3661 gimple_omp_for_initial/final is ADDR_EXPR.
3664 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3665 into gimple val before gimplification fo the COND_EXPR.
3667 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3669 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3670 REGION_COPY argument.
3671 (thread_through_all_blocks): Remove unused argument to
3672 duplicate_thread_path.
3674 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3675 Alan Hayward <alan.hayward@arm.com>
3676 David Sherwood <david.sherwood@arm.com>
3678 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3679 Take a scalar_mode rather than a machine_mode.
3680 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3681 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3682 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3683 (aarch64_gen_adjusted_ldpstp): Likewise.
3684 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3686 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3687 Alan Hayward <alan.hayward@arm.com>
3688 David Sherwood <david.sherwood@arm.com>
3690 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3691 Take a scalar_int_mode instead of a machine_mode.
3692 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3693 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3694 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3695 (aarch64_simd_attr_length_rglist): Delete.
3696 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3697 a scalar_int_mode instead of a machine_mode.
3698 (aarch64_add_offset): Likewise.
3699 (aarch64_internal_mov_immediate): Likewise
3700 (aarch64_add_constant_internal): Likewise.
3701 (aarch64_add_constant): Likewise.
3702 (aarch64_movw_imm): Likewise.
3703 (aarch64_rtx_arith_op_extract_p): Likewise.
3704 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3705 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3706 Remove assert that the mode isn't a vector.
3707 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3708 (aarch64_expand_mov_immediate): Update calls after above changes.
3709 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3710 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3711 (aarch64_move_imm): Likewise.
3712 (aarch64_can_const_movi_rtx_p): Likewise.
3713 (aarch64_strip_extend): Likewise.
3714 (aarch64_extr_rtx_p): Likewise.
3715 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3716 a CONST_INT when the mode parameter is VOIDmode.
3717 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3719 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3721 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3722 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3723 (bitwise_type_for_mode): Update accordingly.
3725 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3727 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3728 * stor-layout.c (mode_for_size_tree): Likewise.
3729 (mode_for_array): Update accordingly.
3730 (layout_decl): Likewise.
3731 (compute_record_mode): Likewise. Only set the mode once.
3733 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3735 * target.def (get_mask_mode): Change return type to opt_mode.
3737 * doc/tm.texi: Regenerate.
3738 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3739 * targhooks.c (default_get_mask_mode): Likewise.
3740 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3741 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3742 targetm.get_mask_mode.
3743 * tree.c (build_truth_vector_type): Likewise.
3745 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3747 * machmode.h (mode_for_vector): Return an opt_mode.
3748 * stor-layout.c (mode_for_vector): Likewise.
3749 (mode_for_int_vector): Update accordingly.
3750 (layout_type): Likewise.
3751 * config/i386/i386.c (emit_memmov): Likewise.
3752 (ix86_expand_set_or_movmem): Likewise.
3753 (ix86_expand_vector_init): Likewise.
3754 (ix86_get_mask_mode): Likewise.
3755 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3757 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3758 * expmed.c (extract_bit_field_1): Likewise.
3759 * expr.c (expand_expr_real_2): Likewise.
3760 * optabs-query.c (can_vec_perm_p): Likewise.
3761 (can_vec_mask_load_store_p): Likewise.
3762 * optabs.c (expand_vec_perm): Likewise.
3763 * targhooks.c (default_get_mask_mode): Likewise.
3764 * tree-vect-stmts.c (vectorizable_store): Likewise.
3765 (vectorizable_load): Likewise.
3766 (get_vectype_for_scalar_type_and_size): Likewise.
3768 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3770 * machmode.h (mode_for_int_vector): New function.
3771 * stor-layout.c (mode_for_int_vector): Likewise.
3772 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3773 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3774 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3775 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3776 (s390_expand_vcond): Likewise.
3778 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3780 * machmode.h (opt_machine_mode): New type.
3781 (opt_mode<T>): Allow construction from anything that can be
3783 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3784 (mode_for_size): Return an opt_machine_mode.
3785 * stor-layout.c (mode_for_size): Likewise.
3786 (mode_for_size_tree): Update call accordingly.
3787 (bitwise_mode_for_mode): Likewise.
3788 (make_fract_type): Likewise.
3789 (make_accum_type): Likewise.
3790 * caller-save.c (replace_reg_with_saved_mem): Update call
3792 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3793 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3794 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3795 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3796 * expmed.c (extract_bit_field_1): Likewise.
3797 * reload.c (get_secondary_mem): Likewise.
3798 * varasm.c (assemble_integer): Likewise.
3799 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3802 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3804 * machmode.h (decimal_float_mode_for_size): New function.
3805 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3806 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3807 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3808 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3809 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3810 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3812 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3814 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3815 (get_builtin_sync_mode): Likewise.
3816 (expand_ifn_atomic_compare_exchange): Likewise.
3817 (expand_builtin_atomic_clear): Likewise.
3818 (expand_builtin_atomic_test_and_set): Likewise.
3819 (fold_builtin_atomic_always_lock_free): Likewise.
3820 * calls.c (compute_argument_addresses): Likewise.
3821 (emit_library_call_value_1): Likewise.
3822 (store_one_arg): Likewise.
3823 * combine.c (combine_instructions): Likewise.
3824 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3825 * config/arm/arm.c (arm_function_value): Likewise.
3826 (aapcs_allocate_return_reg): Likewise.
3827 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3828 * config/i386/i386.c (construct_container): Likewise.
3829 (ix86_gimplify_va_arg): Likewise.
3830 (ix86_expand_sse_cmp): Likewise.
3831 (emit_memmov): Likewise.
3832 (emit_memset): Likewise.
3833 (expand_small_movmem_or_setmem): Likewise.
3834 (ix86_expand_pextr): Likewise.
3835 (ix86_expand_pinsr): Likewise.
3836 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
3837 * config/microblaze/microblaze.c (microblaze_block_move_straight):
3839 * config/mips/mips.c (mips_function_value_1) Likewise.
3840 (mips_block_move_straight): Likewise.
3841 (mips_expand_ins_as_unaligned_store): Likewise.
3842 * config/powerpcspe/powerpcspe.c
3843 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3844 (rs6000_darwin64_record_arg_flush): Likewise.
3845 * config/rs6000/rs6000.c
3846 (rs6000_darwin64_record_arg_advance_flush): Likewise.
3847 (rs6000_darwin64_record_arg_flush): Likewise.
3848 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
3849 (sparc_function_value_1): Likewise.
3850 * config/spu/spu.c (adjust_operand): Likewise.
3851 (spu_emit_branch_or_set): Likewise.
3852 (arith_immediate_p): Likewise.
3853 * emit-rtl.c (gen_lowpart_common): Likewise.
3854 * expr.c (expand_expr_real_1): Likewise.
3855 * function.c (assign_parm_setup_block): Likewise.
3856 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3857 * reload1.c (alter_reg): Likewise.
3858 * stor-layout.c (mode_for_vector): Likewise.
3859 (layout_type): Likewise.
3861 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3863 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
3864 (spu_convert_move): Likewise.
3865 * lower-subreg.c (resolve_simple_move): Likewise.
3867 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3870 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
3871 define_insn to a define_expand.
3872 (altivec_vsum2sws_direct): New define_insn.
3873 (altivec_vsumsws): Convert from a define_insn to a define_expand.
3875 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3877 * config/arm/arm.c (arm_option_params_internal): Improve setting of
3880 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
3885 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
3886 realign stack if stack alignment needed is less than incoming
3889 2017-09-05 Marek Polacek <polacek@redhat.com>
3892 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
3895 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
3897 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
3899 2017-09-05 Richard Biener <rguenther@suse.de>
3901 PR tree-optimization/82084
3902 * fold-const.c (can_native_encode_string_p): Handle wide characters.
3904 2017-09-05 Richard Biener <rguenther@suse.de>
3906 PR tree-optimization/82102
3907 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
3909 2017-09-05 Martin Liska <mliska@suse.cz>
3911 PR tree-optimization/82032
3912 * tree-cfg.c (generate_range_test): New function.
3913 * tree-cfg.h (generate_range_test): Declared here.
3914 * tree-cfgcleanup.c (convert_single_case_switch): New function.
3915 (cleanup_control_expr_graph): Use it.
3916 * tree-switch-conversion.c (try_switch_expansion): Remove
3918 (emit_case_nodes): Use generate_range_test.
3920 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3923 * config/i386/i386.md (*<btsc><mode>_mask): Add
3924 TARGET_USE_BT to insn constraint.
3925 (*btr<mode>_mask): Ditto.
3927 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
3929 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
3930 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
3932 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3935 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
3936 TARGET_NEON and TARGET_IWMMXT.
3937 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
3938 TARGET_NEON and TARGET_IWMMXT.
3939 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
3941 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
3943 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
3944 (ix86_rewrite_tls_address): Ditto.
3945 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
3946 (ix86_rewrite_tls_address_1): Ditto.
3947 (ix86_rewrite_tls_address): Ditto.
3948 * config/i386/predicates.md (tls_address_pattern): New predicate.
3949 * config/i386/i386.md (TLS address splitter): New splitter.
3951 2017-09-04 Richard Biener <rguenther@suse.de>
3953 PR tree-optimization/82084
3954 * fold-const.h (can_native_encode_string_p): Declare.
3955 * fold-const.c (can_native_encode_string_p): Factor out from ...
3956 (native_encode_string): ... here.
3957 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
3958 vectorizing stores from constants we later cannot handle.
3960 2017-09-04 Marek Polacek <polacek@redhat.com>
3963 * doc/invoke.texi: Update -Wtautological-compare documentation.
3965 2017-09-04 Jeff Law <law@redhat.com>
3967 PR tree-optimization/64910
3968 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
3969 swap the first and last operand if the last is a constant.
3971 2017-09-04 Marek Polacek <polacek@redhat.com>
3974 * convert.c (do_narrow): When sanitizing signed integer overflows,
3975 bail out for signed types.
3976 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
3978 2017-09-04 Richard Biener <rguenther@suse.de>
3980 PR tree-optimization/82060
3981 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
3982 Move devirtualization after stmt folding and before EH/AB/noreturn
3983 cleanup to get the stmt refs canonicalized. Use a bool instead
3984 of gimple_modified_p since that doesn't work for NOPs. Schedule
3985 NOPs generated by folding for removal.
3987 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
3988 Alan Hayward <alan.hayward@arm.com>
3989 David Sherwood <david.sherwood@arm.com>
3991 * coretypes.h (pad_direction): New enum.
3992 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
3993 (FUNCTION_ARG_PADDING): Likewise.
3994 * target.def (function_arg_padding): New hook.
3995 * targhooks.h (default_function_arg_padding): Declare.
3996 * targhooks.c (default_function_arg_padding): New function.
3997 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
3998 (TARGET_FUNCTION_ARG_PADDING): ...this.
3999 * doc/tm.texi: Regenerate.
4000 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4001 instead of direction.
4002 (compute_argument_addresses): Likewise.
4003 (load_register_parameters): Likewise.
4004 (emit_library_call_value_1): Likewise.
4005 (store_one_arg): Use targetm.calls.function_arg_padding instead
4006 of FUNCTION_ARG_PADDING.
4007 (must_pass_in_stack_var_size_or_pad): Likewise.
4008 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4009 (emit_group_store): Likewise.
4010 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4011 instead of FUNCTION_ARG_PADDING.
4012 (emit_push_insn): Likewise, and propagate enum change throughout
4014 * function.h (direction): Delete.
4015 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4017 * function.c (assign_parm_find_stack_rtl): Likewise.
4018 (assign_parm_setup_block_p): Likewise.
4019 (assign_parm_setup_block): Likewise.
4020 (gimplify_parameters): Likewise.
4021 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4022 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4024 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4025 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4026 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4027 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4028 (aarch64_function_arg_padding): ...this new function.
4029 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4030 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4031 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4032 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4033 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4034 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4035 (arm_pad_arg_upward): Replace with...
4036 (arm_function_arg_padding): ...this new function.
4037 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4039 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4040 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4041 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4042 (ia64_hpux_function_arg_padding): Replace with...
4043 (ia64_function_arg_padding): ...this new function. Use pad_direction
4044 instead of direction. Check for TARGET_HPUX.
4045 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4046 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4047 (iq2000_function_arg_padding): New function.
4048 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4049 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4050 (mips_function_arg_padding): ...this new function.
4051 (mips_pad_reg_upward): Update accordingly.
4052 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4053 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4054 targetm.calls.function_arg_padding.
4055 (FUNCTION_ARG_PADDING): Delete.
4056 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4057 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4058 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4059 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4060 (nios2_block_reg_padding): Return pad_direction instead of direction.
4061 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4062 instead of direction.
4063 (nios2_function_arg_padding): Likewise. Make static.
4064 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4065 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4066 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4067 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4068 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4069 (pa_function_arg_padding): Make static. Return pad_direction instead
4071 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4072 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4073 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4074 instead of direction. Use targetm.calls.function_arg_padding.
4075 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4076 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4077 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4078 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4079 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4081 (function_arg_padding): Rename to...
4082 (rs6000_function_arg_padding): ...this. Make static. Return
4083 pad_direction instead of direction.
4084 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4085 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4086 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4087 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4088 instead of direction. Use targetm.calls.function_arg_padding.
4089 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4090 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4091 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4092 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4093 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4094 (function_arg_padding): Rename to...
4095 (rs6000_function_arg_padding): ...this. Make static. Return
4096 pad_direction instead of direction.
4097 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4098 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4099 * config/s390/s390.c (s390_function_arg_padding): New function.
4100 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4101 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4102 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4103 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4104 (function_arg_padding): Rename to...
4105 (sparc_function_arg_padding): ...this. Make static. Return
4106 pad_direction instead of direction.
4107 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4108 * config/spu/spu.c (spu_function_arg_padding): New function.
4109 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4110 * system.h (FUNCTION_ARG_PADDING): Poison.
4112 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4113 Alan Hayward <alan.hayward@arm.com>
4114 David Sherwood <david.sherwood@arm.com>
4116 * target.def (modes_tieable_p): New hook.
4117 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4118 (TARGET_MODES_TIEABLE_P): ...this.
4119 * doc/tm.texi.in: Regenerate.
4120 * hooks.h (hook_bool_mode_mode_true): Declare.
4121 * hooks.c (hook_bool_mode_mode_true): New function.
4122 * combine.c (subst): Use targetm.modes_tieable_p instead of
4124 * dse.c (find_shift_sequence): Likewise.
4125 * expmed.c (extract_low_bits): Likewise.
4126 * lower-subreg.c: Include target.h.
4127 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4129 * rtlanal.c (rtx_cost): Likewise.
4130 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4131 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4132 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4133 (TARGET_MODES_TIEABLE_P): Redefine.
4134 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4135 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4136 (TARGET_MODES_TIEABLE_P): Redefine.
4137 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4138 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4139 (arc_modes_tieable_p): New function.
4140 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4141 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4142 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4143 (arm_modes_tieable_p): Make static.
4144 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4145 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4146 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4147 (TARGET_MODES_TIEABLE_P): Redefine.
4148 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4149 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4150 (TARGET_MODES_TIEABLE_P): Redefine.
4151 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4152 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4153 (cr16_modes_tieable_p): New function.
4154 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4155 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4156 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4157 (TRULY_NOOP_TRUNCATION): Update comment.
4158 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4159 (TRULY_NOOP_TRUNCATION): Update comment.
4160 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4161 (frv_modes_tieable_p): New function.
4162 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4163 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4164 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4165 (TARGET_MODES_TIEABLE_P): Redefine.
4166 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4167 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4168 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4169 (TARGET_MODES_TIEABLE_P): Redefine.
4170 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4171 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4172 (ia64_modes_tieable_p): New function.
4173 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4174 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4175 (iq2000_modes_tieable_p): New function.
4176 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4177 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4178 (lm32_modes_tieable_p): New function.
4179 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4180 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4181 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4182 (TARGET_MODES_TIEABLE_P): Redefine.
4183 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4184 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4185 (m32r_modes_tieable_p): New function.
4186 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4187 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4188 (m68k_modes_tieable_p): New function.
4189 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4190 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4191 (mcore_modes_tieable_p): New function.
4192 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4193 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4195 (TARGET_MODES_TIEABLE_P): Redefine.
4196 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4197 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4198 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4199 (TARGET_MODES_TIEABLE_P): Redefine.
4200 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4201 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4202 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4203 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4204 (mn10300_modes_tieable_p): ...this and make static.
4205 (TARGET_MODES_TIEABLE_P): Redefine.
4206 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4207 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4208 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4209 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4210 (msp430_modes_tieable_p): Make static.
4211 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4212 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4213 (TARGET_MODES_TIEABLE_P): Redefine.
4214 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4215 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4216 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4217 (TARGET_MODES_TIEABLE_P): Redefine.
4218 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4219 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4220 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4221 (TARGET_MODES_TIEABLE_P): Redefine.
4222 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4223 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4224 (pdp11_modes_tieable_p): New function.
4225 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4226 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4227 (rs6000_modes_tieable_p): New function.
4228 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4229 * config/powerpcspe/powerpcspe.md: Update comment.
4230 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4231 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4232 (TARGET_MODES_TIEABLE_P): Redefine.
4233 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4234 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4235 (rl78_modes_tieable_p): New function.
4236 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4237 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4238 (rs6000_modes_tieable_p): New function.
4239 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4240 * config/rs6000/rs6000.md: Update comment.
4241 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4242 * config/rx/rx.c (rx_modes_tieable_p): New function.
4243 (TARGET_MODES_TIEABLE_P): Redefine.
4244 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4245 * config/s390/s390.c (s390_modes_tieable_p): New function.
4246 (TARGET_MODES_TIEABLE_P): Redefine.
4247 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4248 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4249 (sh_modes_tieable_p): New function.
4250 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4251 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4252 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4253 (sparc_modes_tieable_p): Make static.
4254 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4255 * config/spu/spu.c (spu_modes_tieable_p): New function.
4256 (TARGET_MODES_TIEABLE_P): Redefine.
4257 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4258 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4259 (TARGET_MODES_TIEABLE_P): Redefine.
4260 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4261 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4262 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4263 * config/v850/v850.c (v850_modes_tieable_p): New function.
4264 (TARGET_MODES_TIEABLE_P): Redefine.
4265 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4266 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4267 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4268 (visium_modes_tieable_p): New function.
4269 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4270 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4271 (xtensa_modes_tieable_p): New function.
4272 * system.h (MODES_TIEABLE_P): Poison.
4274 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4275 Alan Hayward <alan.hayward@arm.com>
4276 David Sherwood <david.sherwood@arm.com>
4278 * target.def (hard_regno_mode_ok): New hook.
4279 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4280 (TARGET_HARD_REGNO_MODE_OK): ...this.
4281 * doc/tm.texi.in: Regenerate.
4282 * hooks.h (hook_bool_uint_mode_true): Declare.
4283 * hooks.c (hook_bool_uint_mode_true): New function.
4284 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4286 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4287 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4288 instead of HARD_REGNO_MODE_OK.
4289 * caller-save.c: Include target.h.
4290 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4292 * combine.c (can_combine_p): Likewise.
4293 (combinable_i3pat): Likewise.
4294 (can_change_dest_mode): Likewise.
4295 * expr.c (init_expr_target): Likewise.
4296 (convert_move): Likewise.
4297 (convert_modes): Likewise.
4298 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4299 (setup_prohibited_mode_move_regs): Likewise.
4300 * ira.h (target_ira): Likewise.
4301 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4302 * lra-constraints.c (process_alt_operands): Likewise.
4303 (split_reg): Likewise.
4304 * recog.c (peep2_find_free_register): Likewise.
4305 * ree.c (combine_reaching_defs): Likewise.
4306 * regcprop.c (maybe_mode_change): Likewise.
4307 * reginfo.c (init_reg_sets_1): Likewise.
4308 (choose_hard_reg_mode): Likewise.
4309 (simplifiable_subregs): Likewise.
4310 * regrename.c (check_new_reg_p): Likewise.
4311 * reload.c (find_valid_class): Likewise.
4312 (find_valid_class_1): Likewise.
4313 (reload_inner_reg_of_subreg): Likewise.
4314 (push_reload): Likewise.
4315 (combine_reloads): Likewise.
4316 (find_dummy_reload): Likewise.
4317 (find_reloads): Likewise.
4318 * reload1.c (find_reg): Likewise.
4319 (set_reload_reg): Likewise.
4320 (allocate_reload_reg): Likewise.
4321 (choose_reload_regs): Likewise.
4322 (reload_adjust_reg_for_temp): Likewise.
4323 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4324 (simplify_subreg_regno): Likewise.
4325 * sel-sched.c (init_regs_for_mode): Likewise.
4326 * varasm.c (make_decl_rtl): Likewise.
4327 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4328 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4330 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4331 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4332 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4333 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4334 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4335 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4336 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4337 (arc_mode_class): Delete.
4338 (HARD_REGNO_MODE_OK): Delete.
4339 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4340 (arc_hard_regno_mode_ok): Rename old array to...
4341 (arc_hard_regno_mode_ok_modes): ...this.
4342 (arc_conditional_register_usage): Update accordingly.
4343 (arc_mode_class): Make static.
4344 (arc_hard_regno_mode_ok): New function.
4345 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4346 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4347 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4348 (arm_hard_regno_mode_ok): Make static.
4349 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4351 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4352 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4353 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4355 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4356 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4357 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4358 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4359 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4360 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4361 * config/bfin/predicates.md (valid_reg_operand): Use
4362 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4363 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4364 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4365 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4366 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4367 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4368 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4369 (cr16_hard_regno_mode_ok): Make static and return a bool.
4370 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4371 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4372 (cris_hard_regno_mode_ok): New function.
4373 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4374 (epiphany_mode_class): Delete.
4375 (HARD_REGNO_MODE_OK): Delete.
4376 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4377 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4378 (hard_regno_mode_ok): Rename to...
4379 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4380 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4381 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4383 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4384 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4385 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4386 (frv_hard_regno_mode_ok): Make static and return a bool.
4387 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4389 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4390 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4391 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4392 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4394 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4395 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4396 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4397 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4399 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4400 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4401 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4402 (ia64_hard_regno_mode_ok): New function.
4403 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4404 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4405 (iq2000_hard_regno_mode_ok): New function.
4406 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4407 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4408 (lm32_hard_regno_mode_ok): New function.
4409 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4410 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4411 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4412 instead of HARD_REGNO_MODE_OK.
4413 (m32c_hard_regno_ok): Rename to...
4414 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4415 (m32c_cannot_change_mode_class): Update accordingly.
4416 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4417 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4418 (m32r_mode_class): Delete.
4419 (HARD_REGNO_MODE_OK): Delete.
4420 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4421 (m32r_hard_regno_mode_ok): Rename to...
4422 (m32r_hard_regno_modes): ...this.
4423 (m32r_mode_class): Make static.
4424 (m32r_hard_regno_mode_ok): New function.
4425 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4426 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4427 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4428 (m68k_hard_regno_mode_ok): Make static.
4429 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4430 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4431 (mcore_hard_regno_mode_ok): New function.
4432 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4433 (HARD_REGNO_MODE_OK): Delete.
4434 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4436 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4437 (microblaze_hard_regno_mode_ok): New function.
4438 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4439 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4440 (mips_hard_regno_mode_ok): Delete.
4441 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4442 (mips_hard_regno_mode_ok_p): ...this and make static.
4443 (mips_hard_regno_mode_ok_p): Rename to...
4444 (mips_hard_regno_mode_ok_uncached): ...this.
4445 (mips_hard_regno_mode_ok): New function.
4446 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4447 of HARD_REGNO_MODE_OK.
4448 (mips_option_override): Update after above name changes.
4449 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4450 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4451 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4452 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4453 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4454 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4455 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4456 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4457 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4458 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4459 (msp430_hard_regno_mode_ok): Make static and return a bool.
4460 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4461 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4462 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4464 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4465 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4466 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4467 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4468 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4469 (PA_HARD_REGNO_MODE_OK): ...this
4470 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4471 (PA_HARD_REGNO_MODE_OK): ...this.
4472 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4473 (pa_hard_regno_mode_ok): New function.
4474 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4475 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4476 (pdp11_hard_regno_mode_ok): New function.
4477 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4478 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4480 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4482 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4483 (rs6000_hard_regno_mode_ok): Rename to...
4484 (rs6000_hard_regno_mode_ok_uncached): ...this.
4485 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4486 (rs6000_hard_regno_mode_ok): New function.
4487 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4488 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4489 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4490 (riscv_hard_regno_mode_ok): ...this and make static.
4491 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4492 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4493 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4494 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4495 (rl78_hard_regno_mode_ok): Make static and return bool.
4496 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4497 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4499 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4500 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4501 (rs6000_hard_regno_mode_ok): Rename to...
4502 (rs6000_hard_regno_mode_ok_uncached): ...this.
4503 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4504 (rs6000_hard_regno_mode_ok): New function.
4505 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4506 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4508 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4509 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4510 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4511 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4512 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4513 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4514 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4515 (sh_hard_regno_mode_ok): Make static.
4516 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4517 instead of HARD_REGNO_MODE_OK.
4518 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4519 (sparc_mode_class): Delete.
4520 (HARD_REGNO_MODE_OK): Delete.
4521 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4522 (hard_regno_mode_classes): Make static.
4523 (sparc_mode_class): Likewise.
4524 (sparc_hard_regno_mode_ok): New function.
4525 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4526 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4527 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4529 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4530 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4531 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4532 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4533 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4535 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4536 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4537 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4538 (visium_hard_regno_mode_ok): New function.
4539 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4540 instead of HARD_REGNO_MODE_OK.
4541 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4542 (HARD_REGNO_MODE_OK): Delete.
4543 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4544 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4545 (xtensa_option_override): Update accordingly.
4546 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4547 (xtensa_hard_regno_mode_ok): New function.
4548 * system.h (HARD_REGNO_MODE_OK): Poison.
4550 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4551 Alan Hayward <alan.hayward@arm.com>
4552 David Sherwood <david.sherwood@arm.com>
4554 * target.def (hard_regno_call_part_clobbered): New hook.
4555 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4556 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4557 * doc/tm.texi: Regenerate.
4558 * hooks.h (hook_bool_uint_mode_false): Declare.
4559 * hooks.c (hook_bool_uint_mode_false): New function.
4560 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4561 * cselib.c (cselib_process_insn): Use
4562 targetm.hard_regno_call_part_clobbered instead of
4563 HARD_REGNO_CALL_PART_CLOBBERED.
4564 * ira-conflicts.c (ira_build_conflicts): Likewise.
4565 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4566 * lra-constraints.c (need_for_call_save_p): Likewise.
4567 * lra-lives.c: Include target.h.
4568 (check_pseudos_live_through_calls): Use
4569 targetm.hard_regno_call_part_clobbered instead of
4570 HARD_REGNO_CALL_PART_CLOBBERED.
4571 * regcprop.c: Include target.h.
4572 (copyprop_hardreg_forward_1): Use
4573 targetm.hard_regno_call_part_clobbered instead of
4574 HARD_REGNO_CALL_PART_CLOBBERED.
4575 * reginfo.c (choose_hard_reg_mode): Likewise.
4576 * regrename.c (check_new_reg_p): Likewise.
4577 * reload.c (find_equiv_reg): Likewise.
4578 * reload1.c (emit_reload_insns): Likewise.
4579 * sched-deps.c (deps_analyze_insn): Likewise.
4580 * sel-sched.c (init_regs_for_mode): Likewise.
4581 (mark_unavailable_hard_regs): Likewise.
4582 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4583 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4584 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4586 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4587 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4588 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4590 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4592 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4593 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4594 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4596 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4597 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4598 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4600 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4601 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4603 * config/powerpcspe/powerpcspe.c
4604 (rs6000_hard_regno_call_part_clobbered): New function.
4605 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4606 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4607 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4609 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4610 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4611 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4613 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4614 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4615 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4617 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4618 Alan Hayward <alan.hayward@arm.com>
4619 David Sherwood <david.sherwood@arm.com>
4621 * rtl.h (subreg_memory_offset): Declare.
4622 * emit-rtl.c (subreg_memory_offset): New function.
4623 * expmed.c (store_bit_field_1): Use it.
4624 * expr.c (undefined_operand_subword_p): Likewise.
4625 * simplify-rtx.c (simplify_subreg): Likewise.
4627 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4629 PR rtl-optimization/57448
4632 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4633 using atomic_load pattern.
4634 (expand_atomic_store): Likewise.
4636 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4639 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4640 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4641 handling. Use replace_call_with_value with NULL instead of
4642 gsi_replace, unlink_stmt_vdef and release_defs.
4644 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4647 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4649 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4652 * rtl.h (emit_library_call_value_1): Declare.
4653 (emit_library_call): Replace declaration with a series of overloads.
4654 Remove the parameter count argument.
4655 (emit_library_call_value): Likewise.
4656 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4657 with an "rtx_mode_t *".
4658 (emit_library_call_value): Delete.
4659 (emit_library_call): Likewise.
4660 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4661 (asan_emit_allocas_unpoison): Likewise.
4662 * builtins.c (expand_builtin_powi): Likewise.
4663 (expand_asan_emit_allocas_unpoison): Likewise.
4664 * cfgexpand.c (expand_main_function): Likewise.
4665 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4666 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4667 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4668 * config/arm/arm.c (arm_trampoline_init): Likewise.
4669 (arm_call_tls_get_addr): Likewise.
4670 (arm_expand_divmod_libfunc): Likewise.
4671 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4672 (smulsi3_highpart): Likewise.
4673 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4674 (c6x_expand_compare): Likewise.
4675 (c6x_expand_movmem): Likewise.
4676 * config/frv/frv.c (frv_trampoline_init): Likewise.
4677 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4678 (ix86_expand_divmod_libfunc): Likewise.
4679 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4680 (ia64_expand_compare): Likewise.
4681 (ia64_profile_hook): Likewise.
4682 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4683 (nonlocal_goto): Likewise.
4684 (restore_stack_nonlocal): Likewise.
4685 * config/m32r/m32r.c (block_move_call): Likewise.
4686 (m32r_trampoline_init): Likewise.
4687 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4688 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4689 (m68k_call_m68k_read_tp): Likewise.
4690 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4691 (microblaze_expand_divide): Likewise.
4692 * config/mips/mips.h (mips_args): Likewise.
4693 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4694 (MIPS_ICACHE_SYNC): Likewise.
4695 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4696 (nios2_trampoline_init): Likewise.
4697 * config/pa/pa.c (hppa_tls_call): Likewise.
4698 (pa_trampoline_init): Likewise.
4699 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4700 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4701 (expand_strn_compare): Likewise.
4702 (rs6000_generate_compare): Likewise.
4703 (rs6000_expand_float128_convert): Likewise.
4704 (output_profile_hook): Likewise.
4705 (rs6000_trampoline_init): Likewise.
4706 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4707 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4708 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4709 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4710 (rs6000_generate_compare): Likewise.
4711 (rs6000_expand_float128_convert): Likewise.
4712 (output_profile_hook): Likewise.
4713 (rs6000_trampoline_init): Likewise.
4714 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4715 * config/sh/sh.c (sh_trampoline_init): Likewise.
4716 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4717 (sparc_emit_float_lib_cmp): Likewise.
4718 (sparc32_initialize_trampoline): Likewise.
4719 (sparc64_initialize_trampoline): Likewise.
4720 (sparc_profile_hook): Likewise.
4721 * config/spu/spu.c (ea_load_store): Likewise.
4722 * config/spu/spu.md (floatunssidf2): Likewise.
4723 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4724 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4725 * config/visium/visium.c (expand_block_move_4): Likewise.
4726 (expand_block_move_2): Likewise.
4727 (expand_block_move_1): Likewise.
4728 (expand_block_set_4): Likewise.
4729 (expand_block_set_2): Likewise.
4730 (expand_block_set_1): Likewise.
4731 (visium_trampoline_init): Likewise.
4732 (visium_profile_hook): Likewise.
4733 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4734 (xtensa_setup_frame_addresses): Likewise.
4735 (xtensa_trampoline_init): Likewise.
4736 * except.c (sjlj_emit_function_enter): Likewise.
4737 (sjlj_emit_function_exit): Likewise.
4738 * explow.c (allocate_dynamic_stack_space): Likewise.
4739 (probe_stack_range): Likewise.
4740 * expr.c (convert_mode_scalar): Likewise.
4741 * optabs.c (expand_binop): Likewise.
4742 (expand_twoval_binop_libfunc): Likewise.
4743 (expand_unop): Likewise.
4744 (prepare_cmp_insn): Likewise.
4745 (prepare_float_lib_cmp): Likewise.
4746 (expand_float): Likewise.
4747 (expand_fix): Likewise.
4748 (expand_fixed_convert): Likewise.
4749 (maybe_emit_sync_lock_test_and_set): Likewise.
4750 (expand_atomic_compare_and_swap): Likewise.
4751 (expand_mem_thread_fence): Likewise.
4752 (expand_atomic_fetch_op): Likewise.
4754 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4756 * doc/generic.texi (OpenACC): Adjust URL.
4757 * doc/invoke.texi (C Dialect Options): Ditto.
4759 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4761 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4762 predicate for operand 1. Add (m,<S>) constraint.
4763 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4764 Prevent memory operand 1 with register operand 2.
4766 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4768 PR rtl-optimization/82024
4769 * combine.c (try_combine): If the combination result is a PARALLEL,
4770 and we only need to retain the SET in there that would be placed
4771 at I2, check that we can place that at I3 instead, before doing so.
4773 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4776 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4778 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4779 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4782 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4783 Jeff Law <law@redhat.com>
4785 * varasm.c (bss_initializer_p): Do not put constants into .bss
4786 (categorize_decl_for_section): Handle bss_initializer_p returning
4787 false when DECL_INITIAL is NULL.
4789 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4792 * config/s390/s390.c (s390_can_inline_p): New function.
4794 2017-09-01 Jeff Law <law@redhat.com>
4796 PR tree-optimization/82052
4797 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4798 Always initialize the returned slot after a hash table miss
4799 when INSERT is true.
4801 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4803 * config/s390/s390.md (mem_signal_fence): Remove.
4804 * doc/md.texi (mem_signal_fence): Remove.
4805 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4807 * target-insns.def (mem_signal_fence): Remove.
4809 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4812 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4815 * asan.c (create_odr_indicator): Strip name encoding from assembler
4816 name before appending it after __odr_asan_.
4818 2017-09-01 Martin Liska <mliska@suse.cz>
4820 PR tree-optimization/82059
4821 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4822 frequency only when an edge is redirected.
4824 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4826 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4827 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4828 (arc_conditional_register_usage): Remove ARC600 lp_count
4830 (arc_file_start): Emit Tag_ARC_CPU_variation.
4831 (arc_can_use_doloop_p): New conditions to use ZOLs.
4832 (hwloop_fail): New function.
4833 (hwloop_optimize): Likewise.
4834 (hwloop_pattern_reg): Likewise.
4835 (arc_doloop_hooks): New struct, to be used with reorg_loops.
4836 (arc_reorg_loops): New function, calls reorg_loops.
4837 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
4838 (arc600_corereg_hazard): Remove ZOL checking, case handled by
4840 (arc_loop_hazard): Remove function, functionality moved into
4842 (arc_hazard): Remove arc_loop_hazard call.
4843 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
4844 into hwloop_optimize.
4845 (arc_label_align): Remove ZOL handling.
4846 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
4847 * config/arc/arc.md (doloop_begin): Remove pattern.
4848 (doloop_begin_i): Likewise.
4849 (doloop_end_i): Likewise.
4850 (doloop_fallback): Likewise.
4851 (doloop_fallback_m): Likewise.
4852 (doloop_end): Reimplement expand.
4853 (arc_lp): New pattern for LP instruction.
4854 (loop_end): New pattern.
4855 (loop_fail): Likewise.
4856 (decrement_and_branch_until_zero): Likewise.
4857 * config/arc/arc.opt (mlpc-width): New option.
4858 * doc/invoke.texi (mlpc-width): Document option.
4860 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4862 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
4863 (arc_ccfsm_advance): Fix checking for delay slots.
4864 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
4866 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4868 * config/arc/arc.md (movqi_insn): Add stores to save constant long
4870 (movhi_insn): Update store instruction constraint which are saving
4871 6-bit short immediates.
4872 (movsi_insn): Consider also short scaled load operations.
4873 (zero_extendhisi2_i): Use Usd constraint instead of T.
4874 (extendhisi2_i): Add q constraint.
4875 (arc_clzsi2): Add type and length attributes.
4876 (arc_ctzsi2): Likewise.
4877 * config/arc/constraints.md (Usc): Update constraint, the
4878 assembler can parse two relocations for a single instruction.
4880 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4882 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
4883 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
4885 2017-08-31 Olivier Hainque <hainque@adacore.com>
4887 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
4888 match as powerpc-wrs-vxworks*.
4890 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
4892 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
4893 register constraint for by-element operand.
4894 (aarch64_mls_elt_merge<mode>): Likewise.
4896 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4898 * config/arc/arc.c (arc_can_follow_jump): Check for short
4901 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4903 * config.gcc: Use g.opt for arc.
4904 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
4905 functionality moved to ...
4906 (legitimate_scaled_address_p): New function, ...here.
4907 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
4908 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
4909 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
4911 (arc_override_options): Handle G option.
4912 (arc_output_pic_addr_const): Correct function definition.
4913 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
4914 (arc_decl_anon_ns_mem_p): Delete.
4915 (arc_in_small_data_p): Overhaul this function to take into
4916 consideration the value given via G option.
4917 (arc_rewrite_small_data_1): Renamed and corrected old
4918 arc_rewrite_small_data function.
4919 (arc_rewrite_small_data): New function.
4920 (small_data_pattern): Don't use pic_offset_table_rtx.
4921 * config/arc/arc.h (CC1_SPEC): Recognize G option.
4922 * config/arc/simdext.md (movmisalignv2hi): Use
4923 prepare_move_operands function.
4925 (movmisalign*): Likewise.
4926 * doc/invoke.texi (ARC options): Document -G option.
4928 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
4930 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
4932 * config/arc/arc.c (arc_print_operand): Output scalled address for
4933 sdata whenever is possible.
4934 (arc_in_small_data_p): Allow sdata for 64bit datum when double
4935 load/stores are available.
4936 (compact_sda_memory_operand): Check for the alignment required by
4937 code density instructions.
4938 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
4940 * config/arc/constraints.md (Usd): Update constraint.
4941 (Us0): New constraint.
4942 (Usc): Update constraint.
4944 2017-08-31 Richard Biener <rguenther@suse.de>
4947 * dwarf2out.c (dwarf2out_early_global_decl): Process each
4950 2017-08-31 Tamar Christina <tamar.christina@arm.com>
4952 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
4953 Resize type_signature.
4955 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4956 Alan Hayward <alan.hayward@arm.com>
4957 David Sherwood <david.sherwood@arm.com>
4959 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
4960 subregs whose inner modes can be stored in GPRs.
4961 (aarch64_classify_index): Likewise.
4963 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4964 Alan Hayward <alan.hayward@arm.com>
4965 David Sherwood <david.sherwood@arm.com>
4967 * config/aarch64/iterators.md (V_cmp_result): Rename to...
4968 (V_INT_EQUIV): ...this.
4969 (v_cmp_result): Rename to...
4970 (v_int_equiv): ...this.
4971 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
4972 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
4973 (copysign<mode>3): Likewise.
4974 (aarch64_simd_bsl<mode>_internal): Likewise.
4975 (aarch64_simd_bsl<mode>): Likewise.
4976 (vec_cmp<mode><mode>): Likewise.
4977 (vcond<mode><mode>): Likewise.
4978 (vcond<v_cmp_mixed><mode>): Likewise.
4979 (vcondu<mode><v_cmp_mixed>): Likewise.
4980 (aarch64_cm<optab><mode>): Likewise.
4981 (aarch64_cmtst<mode>): Likewise.
4982 (aarch64_fac<optab><mode>): Likewise.
4983 (vec_perm_const<mode>): Likewise.
4984 (vcond_mask_<mode><v_cmp_result>): Rename to...
4985 (vcond_mask_<mode><v_int_equiv>): ...this.
4986 (vec_cmp<mode><v_cmp_result>): Rename to...
4987 (vec_cmp<mode><v_int_equiv>): ...this.
4989 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
4990 Alan Hayward <alan.hayward@arm.com>
4991 David Sherwood <david.sherwood@arm.com>
4993 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
4995 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
4996 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
4997 (UNSPEC_LD4_DREG): New unspecs.
4998 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
4999 (aarch64_ld2<mode>_dreg_be): Replace with...
5000 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5002 (aarch64_ld3<mode>_dreg_le)
5003 (aarch64_ld3<mode>_dreg_be): Replace with...
5004 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5006 (aarch64_ld4<mode>_dreg_le)
5007 (aarch64_ld4<mode>_dreg_be): Replace with...
5008 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5011 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5013 PR tree-optimization/81987
5014 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5015 insert an initializer in a location not dominated by the stride
5018 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5020 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5021 on the entire header of the finally block in the fallthru case.
5023 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5025 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5027 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5029 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5030 rs6000_emit_move_from_cr and call renamed function.
5031 (rs6000_emit_prologue): Call renamed functions.
5032 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5033 movesi_from_cr, remove volatile CRs.
5035 2017-08-30 Jon Beniston <jon@beniston.com>
5036 Richard Biener <rguenther@suse.de>
5038 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5039 of VECTOR_MODE_P check.
5040 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5041 element vector types.
5043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5045 * df.h (df_read_modify_subreg_p): Remove in favor of...
5046 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5047 const_rtx instead of an rtx.
5048 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5049 * df-problems.c (df_word_lr_mark_ref): Likewise.
5050 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5051 (mark_pseudo_reg_dead): Likewise.
5052 (mark_ref_dead): Likewise.
5053 * reginfo.c (init_subregs_of_mode): Likewise.
5054 * sched-deps.c (sched_analyze_1): Likewise.
5055 * df-scan.c (df_def_record_1): Likewise.
5056 (df_uses_record): Likewise.
5057 (df_read_modify_subreg_p): Remove in favor of...
5058 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5059 const_rtx instead of an rtx.
5061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5062 Alan Hayward <alan.hayward@arm.com>
5063 David Sherwood <david.sherwood@arm.com>
5065 * rtl.h (partial_subreg_p): New function.
5066 * caller-save.c (save_call_clobbered_regs): Use it.
5067 * calls.c (expand_call): Likewise.
5068 * combine.c (combinable_i3pat): Likewise.
5069 (simplify_set): Likewise.
5070 (make_extraction): Likewise.
5071 (make_compound_operation_int): Likewise.
5072 (gen_lowpart_or_truncate): Likewise.
5073 (force_to_mode): Likewise.
5074 (make_field_assignment): Likewise.
5075 (reg_truncated_to_mode): Likewise.
5076 (record_truncated_value): Likewise.
5077 (move_deaths): Likewise.
5078 * cse.c (record_jump_cond): Likewise.
5079 (cse_insn): Likewise.
5080 * cselib.c (cselib_lookup_1): Likewise.
5081 * expmed.c (extract_bit_field_using_extv): Likewise.
5082 * function.c (assign_parm_setup_reg): Likewise.
5083 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5084 * ira-build.c (create_insn_allocnos): Likewise.
5085 * lra-coalesce.c (merge_pseudos): Likewise.
5086 * lra-constraints.c (match_reload): Likewise.
5087 (simplify_operand_subreg): Likewise.
5088 (curr_insn_transform): Likewise.
5089 * lra-lives.c (process_bb_lives): Likewise.
5090 * lra.c (new_insn_reg): Likewise.
5091 (lra_substitute_pseudo): Likewise.
5092 * regcprop.c (mode_change_ok): Likewise.
5093 (maybe_mode_change): Likewise.
5094 (copyprop_hardreg_forward_1): Likewise.
5095 * reload.c (push_reload): Likewise.
5096 (find_reloads): Likewise.
5097 (find_reloads_subreg_address): Likewise.
5098 * reload1.c (alter_reg): Likewise.
5099 (eliminate_regs_1): Likewise.
5100 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5102 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5104 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5105 back to if statements, including unpack.
5107 2017-08-30 Martin Liska <mliska@suse.cz>
5110 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5112 (func_checker::compare_asm_inputs_outputs): ... this function.
5113 (func_checker::compare_gimple_asm): Use the function to compare
5114 also ASM constrains.
5115 * ipa-icf-gimple.h: Rename the function.
5117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5118 Alan Hayward <alan.hayward@arm.com>
5119 David Sherwood <david.sherwood@arm.com>
5121 * coretypes.h (complex_mode): New type.
5122 * gdbhooks.py (build_pretty_printer): Handle it.
5123 * machmode.h (complex_mode): New class.
5124 (complex_mode::includes_p): New function.
5125 (is_complex_int_mode): Likewise.
5126 (is_complex_float_mode): Likewise.
5127 * genmodes.c (get_mode_class): Handle complex mode classes.
5128 * function.c (expand_function_end): Use is_complex_int_mode.
5130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5131 Alan Hayward <alan.hayward@arm.com>
5132 David Sherwood <david.sherwood@arm.com>
5134 * coretypes.h (scalar_mode_pod): New typedef.
5135 * gdbhooks.py (build_pretty_printer): Handle it.
5136 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5137 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5138 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5139 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5143 Alan Hayward <alan.hayward@arm.com>
5144 David Sherwood <david.sherwood@arm.com>
5146 * machmode.h (mode_for_vector): Take a scalar_mode instead
5148 * stor-layout.c (mode_for_vector): Likewise.
5149 * explow.c (promote_mode): Use as_a <scalar_mode>.
5150 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5153 Alan Hayward <alan.hayward@arm.com>
5154 David Sherwood <david.sherwood@arm.com>
5156 * target.def (preferred_simd_mode): Take a scalar_mode
5157 instead of a machine_mode.
5158 * targhooks.h (default_preferred_simd_mode): Likewise.
5159 * targhooks.c (default_preferred_simd_mode): Likewise.
5160 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5161 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5162 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5163 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5164 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5165 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5166 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5167 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5169 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5170 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5171 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5172 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5173 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5174 * doc/tm.texi: Regenerate.
5175 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5179 Alan Hayward <alan.hayward@arm.com>
5180 David Sherwood <david.sherwood@arm.com>
5182 * target.def (scalar_mode_supported_p): Take a scalar_mode
5183 instead of a machine_mode.
5184 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5185 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5186 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5187 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5188 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5189 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5190 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5191 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5192 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5193 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5194 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5195 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5196 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5197 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5199 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5200 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5201 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5202 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5203 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5205 * doc/tm.texi: Regenerate.
5207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5208 Alan Hayward <alan.hayward@arm.com>
5209 David Sherwood <david.sherwood@arm.com>
5211 * coretypes.h (opt_scalar_mode): New typedef.
5212 * gdbhooks.py (build_pretty_printers): Handle it.
5213 * machmode.h (mode_iterator::get_2xwider): Add overload for
5215 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5217 * expr.c (convert_mode_scalar): Likewise.
5218 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5219 * optabs.c (expand_float): Likewise.
5220 (expand_fix): Likewise.
5221 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5224 Alan Hayward <alan.hayward@arm.com>
5225 David Sherwood <david.sherwood@arm.com>
5227 * optabs.c (expand_float): Explicitly check for scalars before
5228 using a branching expansion.
5229 (expand_fix): Likewise.
5231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5232 Alan Hayward <alan.hayward@arm.com>
5233 David Sherwood <david.sherwood@arm.com>
5235 * expr.c (convert_mode): Split scalar handling out into...
5236 (convert_mode_scalar): ...this new function. Treat the modes
5239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5240 Alan Hayward <alan.hayward@arm.com>
5241 David Sherwood <david.sherwood@arm.com>
5243 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5245 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5248 Alan Hayward <alan.hayward@arm.com>
5249 David Sherwood <david.sherwood@arm.com>
5251 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5252 rather than a machine_mode.
5253 (fixed_from_string): Likewise.
5254 (fixed_convert): Likewise.
5255 (fixed_convert_from_int): Likewise.
5256 (fixed_convert_from_real): Likewise.
5257 (real_convert_from_fixed): Likewise.
5258 * fixed-value.c (fixed_from_double_int): Likewise.
5259 (fixed_from_string): Likewise.
5260 (fixed_convert): Likewise.
5261 (fixed_convert_from_int): Likewise.
5262 (fixed_convert_from_real): Likewise.
5263 (real_convert_from_fixed): Likewise.
5264 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5267 Alan Hayward <alan.hayward@arm.com>
5268 David Sherwood <david.sherwood@arm.com>
5270 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5271 of separate mode class checks. Do not allow vector modes here.
5272 (immed_wide_int_const): Use as_a <scalar_mode>.
5273 * explow.c (trunc_int_for_mode): Likewise.
5274 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5275 (wi::shwi): Likewise.
5276 (wi::min_value): Likewise.
5277 (wi::max_value): Likewise.
5278 * dwarf2out.c (loc_descriptor): Likewise.
5279 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5283 Alan Hayward <alan.hayward@arm.com>
5284 David Sherwood <david.sherwood@arm.com>
5286 * tree.h (SCALAR_TYPE_MODE): New macro.
5287 * expr.c (expand_expr_addr_expr_1): Use it.
5288 (expand_expr_real_2): Likewise.
5289 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5290 (fold_convert_const_fixed_from_int): Likewise.
5291 (fold_convert_const_fixed_from_real): Likewise.
5292 (native_encode_fixed): Likewise
5293 (native_encode_complex): Likewise
5294 (native_encode_vector): Likewise.
5295 (native_interpret_fixed): Likewise.
5296 (native_interpret_real): Likewise.
5297 (native_interpret_complex): Likewise.
5298 (native_interpret_vector): Likewise.
5299 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5300 (simd_clone_adjust_argument_types): Likewise.
5301 (simd_clone_init_simd_arrays): Likewise.
5302 (simd_clone_adjust): Likewise.
5303 * stor-layout.c (layout_type): Likewise.
5304 * tree.c (build_minus_one_cst): Likewise.
5305 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5306 * tree-inline.c (estimate_move_cost): Likewise.
5307 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5308 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5309 (vectorizable_reduction): Likewise.
5310 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5311 (vect_recog_mixed_size_cond_pattern): Likewise.
5312 (check_bool_pattern): Likewise.
5313 (adjust_bool_pattern): Likewise.
5314 (search_type_for_mask_1): Likewise.
5315 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5316 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5317 (vectorizable_load): Likewise.
5318 (vectorizable_store): Likewise.
5319 * ubsan.c (ubsan_encode_value): Likewise.
5320 * varasm.c (output_constant): Likewise.
5322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5323 Alan Hayward <alan.hayward@arm.com>
5324 David Sherwood <david.sherwood@arm.com>
5326 * coretypes.h (scalar_mode): New class.
5327 * machmode.h (scalar_mode): Likewise.
5328 (scalar_mode::includes_p): New function.
5329 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5330 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5331 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5332 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5333 * expmed.c (store_bit_field_1): Likewise.
5334 (extract_bit_field_1): Likewise.
5335 * expr.c (write_complex_part): Likewise.
5336 (read_complex_part): Likewise.
5337 (emit_move_complex_push): Likewise.
5338 (expand_expr_real_2): Likewise.
5339 * function.c (assign_parm_setup_reg): Likewise.
5340 (assign_parms_unsplit_complex): Likewise.
5341 * optabs.c (expand_binop): Likewise.
5342 * rtlanal.c (subreg_get_info): Likewise.
5343 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5344 * varasm.c (output_constant_pool_2): Likewise.
5346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5347 Alan Hayward <alan.hayward@arm.com>
5348 David Sherwood <david.sherwood@arm.com>
5350 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5352 (expmed_mult_highpart_optab): Likewise.
5353 (expmed_mult_highpart): Likewise.
5355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5356 Alan Hayward <alan.hayward@arm.com>
5357 David Sherwood <david.sherwood@arm.com>
5359 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5360 instead of a machine_mode.
5361 (builtin_memset_read_str): Likewise.
5362 * builtins.c (c_readstr): Likewise.
5363 (builtin_memcpy_read_str): Likewise.
5364 (builtin_strncpy_read_str): Likewise.
5365 (builtin_memset_read_str): Likewise.
5366 (builtin_memset_gen_str): Likewise.
5367 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5368 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5369 instead of a machine_mode.
5370 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5372 (make_extraction): Likewise.
5373 (try_widen_shift_mode): Take and return scalar_int_modes instead
5375 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5376 a scalar_int_mode instead of a machine_mode.
5377 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5378 (avr_addr_space_pointer_mode): Likewise.
5379 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5380 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5381 (msp430_unwind_word_mode): Likewise.
5382 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5383 (spu_addr_space_pointer_mode): Likewise.
5384 (spu_addr_space_address_mode): Likewise.
5385 (spu_libgcc_cmp_return_mode): Likewise.
5386 (spu_libgcc_shift_count_mode): Likewise.
5387 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5388 (rl78_addr_space_pointer_mode): Likewise.
5389 (fl78_unwind_word_mode): Likewise.
5390 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5392 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5393 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5394 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5395 (mips_valid_pointer_mode): Likewise.
5396 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5397 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5398 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5400 (ft32_addr_space_address_mode): Likewise.
5401 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5402 scalar_int_mode instead of a machine_mode.
5403 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5405 (m32c_addr_space_address_mode): Likewise.
5406 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5407 (rs6000_eh_return_filter_mode): Likewise.
5408 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5409 (rs6000_eh_return_filter_mode): Likewise.
5410 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5411 (s390_libgcc_shift_count_mode): Likewise.
5412 (s390_unwind_word_mode): Likewise.
5413 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5415 * target.def (mode_rep_extended): Likewise.
5416 (valid_pointer_mode): Likewise.
5417 (addr_space.valid_pointer_mode): Likewise.
5418 (eh_return_filter_mode): Return a scalar_int_mode rather than
5420 (libgcc_cmp_return_mode): Likewise.
5421 (libgcc_shift_count_mode): Likewise.
5422 (unwind_word_mode): Likewise.
5423 (addr_space.pointer_mode): Likewise.
5424 (addr_space.address_mode): Likewise.
5425 * doc/tm.texi: Regenerate.
5426 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5428 (do_jump): Use scalar_int_mode for local variables.
5429 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5430 rather than a machine_mode.
5431 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5432 (scompare_loc_descriptor_wide): Likewise.
5433 (scompare_loc_descriptor_narrow): Likewise.
5434 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5436 * except.c (sjlj_emit_dispatch_table): Likewise.
5437 (expand_builtin_eh_copy_values): Likewise.
5438 * explow.c (convert_memory_address_addr_space_1): Likewise.
5439 Take a scalar_int_mode rather than a machine_mode.
5440 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5441 than a machine_mode.
5442 (memory_address_addr_space): Use scalar_int_mode for local variables.
5443 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5444 rather than a machine_mode.
5445 * expmed.c (mask_rtx): Likewise.
5446 (init_expmed_one_conv): Likewise.
5447 (expand_mult_highpart_adjust): Likewise.
5448 (extract_high_half): Likewise.
5449 (expmed_mult_highpart_optab): Likewise.
5450 (expmed_mult_highpart): Likewise.
5451 (expand_smod_pow2): Likewise.
5452 (expand_sdiv_pow2): Likewise.
5453 (emit_store_flag_int): Likewise.
5454 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5456 (extract_low_bits): Likewise.
5457 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5459 * expr.c (pieces_addr::adjust): Likewise.
5460 (can_store_by_pieces): Likewise.
5461 (store_by_pieces): Likewise.
5462 (clear_by_pieces_1): Likewise.
5463 (expand_expr_addr_expr_1): Likewise.
5464 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5465 (expand_expr_real_1): Likewise.
5466 (try_casesi): Likewise.
5467 * final.c (shorten_branches): Likewise.
5468 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5469 type of "mode" to machine_mode.
5470 * internal-fn.c (expand_arith_overflow_result_store): Take a
5471 scalar_int_mode rather than a machine_mode.
5472 (expand_mul_overflow): Use scalar_int_mode for local variables.
5473 * loop-doloop.c (doloop_modify): Likewise.
5474 (doloop_optimize): Likewise.
5475 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5476 than a machine_mode.
5477 (expand_doubleword_shift_condmove): Likewise.
5478 (expand_doubleword_shift): Likewise.
5479 (expand_doubleword_clz): Likewise.
5480 (expand_doubleword_popcount): Likewise.
5481 (expand_doubleword_parity): Likewise.
5482 (expand_absneg_bit): Use scalar_int_mode for local variables.
5483 (prepare_float_lib_cmp): Likewise.
5484 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5485 rather than a machine_mode.
5486 (convert_memory_address_addr_space): Likewise.
5487 (get_mode_bounds): Likewise.
5488 (get_address_mode): Return a scalar_int_mode rather than a
5490 * rtlanal.c (get_address_mode): Likewise.
5491 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5492 than a machine_mode.
5493 * targhooks.c (default_mode_rep_extended): Likewise.
5494 (default_valid_pointer_mode): Likewise.
5495 (default_addr_space_valid_pointer_mode): Likewise.
5496 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5497 than a machine_mode.
5498 (default_libgcc_cmp_return_mode): Likewise.
5499 (default_libgcc_shift_count_mode): Likewise.
5500 (default_unwind_word_mode): Likewise.
5501 (default_addr_space_pointer_mode): Likewise.
5502 (default_addr_space_address_mode): Likewise.
5503 * targhooks.h (default_eh_return_filter_mode): Likewise.
5504 (default_libgcc_cmp_return_mode): Likewise.
5505 (default_libgcc_shift_count_mode): Likewise.
5506 (default_unwind_word_mode): Likewise.
5507 (default_addr_space_pointer_mode): Likewise.
5508 (default_addr_space_address_mode): Likewise.
5509 (default_mode_rep_extended): Take a scalar_int_mode rather than
5511 (default_valid_pointer_mode): Likewise.
5512 (default_addr_space_valid_pointer_mode): Likewise.
5513 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5515 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5516 rather than a machine_mode.
5517 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5518 for local variables.
5519 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5520 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5521 than a machine_mode.
5523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5524 Alan Hayward <alan.hayward@arm.com>
5525 David Sherwood <david.sherwood@arm.com>
5527 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5528 the mode argument to scalar_int_mode.
5529 (do_jump_by_parts_zero_rtx): Likewise.
5530 (do_jump_by_parts_equality_rtx): Likewise.
5531 (do_jump_by_parts_greater): Take a mode argument.
5532 (do_jump_by_parts_equality): Likewise.
5533 (do_jump_1): Update calls accordingly.
5535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5536 Alan Hayward <alan.hayward@arm.com>
5537 David Sherwood <david.sherwood@arm.com>
5539 * is-a.h (safe_dyn_cast): New function.
5540 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5541 (jump_table_for_label): Likewise.
5542 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5543 instead of an rtx_insn *.
5544 (shorten_branches): Use dyn_cast instead of LABEL_P and
5545 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5546 rtx_jump_table_data::get_data_mode.
5547 (final_scan_insn): Likewise.
5549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5550 Alan Hayward <alan.hayward@arm.com>
5551 David Sherwood <david.sherwood@arm.com>
5553 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5554 trying to combine a full-register integer set with a subreg
5557 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5558 Alan Hayward <alan.hayward@arm.com>
5559 David Sherwood <david.sherwood@arm.com>
5561 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5562 that is always either address_mode or pointer_mode.
5564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5565 Alan Hayward <alan.hayward@arm.com>
5566 David Sherwood <david.sherwood@arm.com>
5568 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5569 when the two are known to be equal.
5571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5572 Alan Hayward <alan.hayward@arm.com>
5573 David Sherwood <david.sherwood@arm.com>
5575 * simplify-rtx.c (simplify_const_unary_operation): Use
5576 is_a <scalar_int_mode> instead of checking for a nonzero
5577 precision. Forcibly convert op_mode to a scalar_int_mode
5578 in that case. More clearly differentiate the operand and
5579 result modes and use the former when deciding what the value
5580 of a count-bits operation should be. Use is_int_mode instead
5581 of checking for a MODE_INT. Remove redundant check for whether
5582 this mode has a zero precision.
5584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5585 Alan Hayward <alan.hayward@arm.com>
5586 David Sherwood <david.sherwood@arm.com>
5588 * optabs.c (widen_leading): Change the type of the mode argument
5589 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5590 (widen_bswap): Likewise.
5591 (expand_parity): Likewise.
5592 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5593 (expand_ffs): Likewise.
5594 (epand_unop): Check for scalar integer modes before calling the
5597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5598 Alan Hayward <alan.hayward@arm.com>
5599 David Sherwood <david.sherwood@arm.com>
5601 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5603 (expand_expr_real_1): Update call accordingly.
5605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5606 Alan Hayward <alan.hayward@arm.com>
5607 David Sherwood <david.sherwood@arm.com>
5609 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5610 value_mode arguments. Use scalar_int_mode internally.
5611 (store_bit_field_1): Rename the new integer mode from imode
5612 to op0_mode and use it instead of GET_MODE (op0). Update calls
5613 to store_split_bit_field, store_bit_field_using_insv and
5614 store_fixed_bit_field.
5615 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5616 Use scalar_int_mode internally. Use a bit count rather than a mode
5617 when calculating the largest bit size for get_best_mode.
5618 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5619 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5620 Remove assertion that OP0 has a scalar integer mode.
5621 (store_split_bit_field): Add op0_mode and value_mode arguments.
5622 Update calls to extract_fixed_bit_field.
5623 (extract_bit_field_using_extv): Add an op0_mode argument.
5624 Use scalar_int_mode internally.
5625 (extract_bit_field_1): Rename the new integer mode from imode to
5626 op0_mode and use it instead of GET_MODE (op0). Update calls to
5627 extract_split_bit_field, extract_bit_field_using_extv and
5628 extract_fixed_bit_field.
5629 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5630 to extract_split_bit_field and extract_fixed_bit_field_1.
5631 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5632 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5634 (extract_split_bit_field): Add an op0_mode argument. Update call
5635 to extract_fixed_bit_field.
5637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5638 Alan Hayward <alan.hayward@arm.com>
5639 David Sherwood <david.sherwood@arm.com>
5641 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5642 * explow.c (hard_function_value): Likewise.
5643 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5644 convert_to_mode call outside the loop.
5645 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5646 for the mode iterator. Require the mode specified by max_pieces
5648 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5650 (copy_blkmode_to_reg): Likewise.
5651 (set_storage_via_setmem): Likewise.
5652 * optabs.c (prepare_cmp_insn): Likewise.
5653 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5654 * stor-layout.c (finish_bitfield_representative): Likewise.
5656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5657 Alan Hayward <alan.hayward@arm.com>
5658 David Sherwood <david.sherwood@arm.com>
5660 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5661 * expr.c (convert_move): Use them.
5662 (convert_modes): Likewise.
5663 (store_expr_with_bounds): Likewise.
5665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5666 Alan Hayward <alan.hayward@arm.com>
5667 David Sherwood <david.sherwood@arm.com>
5669 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5670 parameter for the mode of "x". Remove the "known_x", "known_mode"
5671 and "known_ret" arguments. Change the type of the mode argument
5673 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5674 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5675 (reg_num_sign_bit_copies_for_combine): Likewise.
5676 * rtlanal.c (nonzero_bits1): Likewise.
5677 (num_sign_bit_copies1): Likewise.
5678 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5679 (reg_num_sign_bit_copies_general): Likewise.
5680 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5681 (reg_nonzero_bits_general): Likewise.
5683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5684 Alan Hayward <alan.hayward@arm.com>
5685 David Sherwood <david.sherwood@arm.com>
5687 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5688 than in subroutines. Return 1 for non-integer modes.
5689 (cached_num_sign_bit_copies): Change the type of the mode parameter
5691 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5692 classes. Handle CONST_INT_P first and then check whether X also
5693 has a scalar integer mode. Check the same thing for inner registers
5694 of a SUBREG and for values that are being extended or truncated.
5696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5697 Alan Hayward <alan.hayward@arm.com>
5698 David Sherwood <david.sherwood@arm.com>
5700 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5701 in subroutines. Return the mode mask for non-integer modes.
5702 (cached_nonzero_bits): Change the type of the mode parameter
5704 (nonzero_bits1): Likewise. Remove early exit for other mode
5705 classes. Handle CONST_INT_P first and then check whether X
5706 also has a scalar integer mode.
5708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5709 Alan Hayward <alan.hayward@arm.com>
5710 David Sherwood <david.sherwood@arm.com>
5712 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5713 Return a scalar_int_mode and assert that the size is greater than
5715 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5716 (op_by_pieces_d::op_by_pieces_d): Likewise.
5717 (op_by_pieces_d::run): Likewise.
5718 (can_store_by_pieces): Likewise.
5720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5721 Alan Hayward <alan.hayward@arm.com>
5722 David Sherwood <david.sherwood@arm.com>
5724 * combine.c (extract_left_shift): Add a mode argument and update
5726 (make_compound_operation_int): Change the type of the mode parameter
5727 to scalar_int_mode and update the call to extract_left_shift.
5729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5730 Alan Hayward <alan.hayward@arm.com>
5731 David Sherwood <david.sherwood@arm.com>
5733 * combine.c (simplify_and_const_int): Change the type of the mode
5734 parameter to scalar_int_mode.
5735 (simplify_and_const_int_1): Likewise. Update recursive call.
5737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5738 Alan Hayward <alan.hayward@arm.com>
5739 David Sherwood <david.sherwood@arm.com>
5741 * combine.c (simplify_compare_const): Check that the mode is a
5742 scalar_int_mode (rather than VOIDmode) before testing its
5744 (simplify_comparison): Move COMPARISON_P handling out of the
5745 loop and restrict the latter part of the loop to scalar_int_modes.
5746 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5747 and when considering SUBREG_REGs. Use is_int_mode instead of
5748 checking GET_MODE_CLASS against MODE_INT.
5750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5751 Alan Hayward <alan.hayward@arm.com>
5752 David Sherwood <david.sherwood@arm.com>
5754 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5755 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5756 shift_unit_mode and for modes involved in scalar shifts.
5758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5759 Alan Hayward <alan.hayward@arm.com>
5760 David Sherwood <david.sherwood@arm.com>
5762 * combine.c (force_int_to_mode): New function, split out from...
5763 (force_to_mode): ...here. Keep xmode up-to-date and use it
5764 instead of GET_MODE (x).
5766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5767 Alan Hayward <alan.hayward@arm.com>
5768 David Sherwood <david.sherwood@arm.com>
5770 * optabs-query.h (extraction_insn::struct_mode): Change type to
5771 opt_scalar_int_mode and update comment.
5772 (extraction_insn::field_mode): Change type to scalar_int_mode.
5773 (extraction_insn::pos_mode): Likewise.
5774 * combine.c (make_extraction): Update accordingly.
5775 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5776 (get_optab_extraction_insn): Likewise.
5777 * recog.c (simplify_while_replacing): Likewise.
5778 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5779 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5786 to a scalar_int_mode instead of a machine_mode.
5787 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5788 (get_best_mode): Return a boolean and use a pointer argument to store
5789 the selected mode. Replace the limit mode parameter with a bit limit.
5790 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5791 for the values returned by bit_field_mode_iterator::next_mode.
5792 (store_bit_field): Update call to get_best_mode.
5793 (store_fixed_bit_field): Likewise.
5794 (extract_fixed_bit_field): Likewise.
5795 * expr.c (optimize_bitfield_assignment_op): Likewise.
5796 * fold-const.c (optimize_bit_field_compare): Likewise.
5797 (fold_truth_andor_1): Likewise.
5798 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5799 Update for new type of m_mode.
5800 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5807 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5808 (store_bit_field): Check is_a <scalar_int_mode> before calling
5809 strict_volatile_bitfield_p.
5810 (extract_bit_field): Likewise.
5812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5813 Alan Hayward <alan.hayward@arm.com>
5814 David Sherwood <david.sherwood@arm.com>
5816 * target.def (cstore_mode): Return a scalar_int_mode.
5817 * doc/tm.texi: Regenerate.
5818 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5819 * targhooks.h (default_cstore_mode): Likewise.
5820 * targhooks.c (default_cstore_mode): Likewise, using a forced
5822 * expmed.c (emit_cstore): Expect the target of the cstore to be
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 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
5831 (niter_desc): Likewise mode.
5832 (iv_analyze): Add a mode parameter.
5834 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
5835 and change its type to scalar_int_mode.
5836 * loop-iv.c: Update commentary at head of file.
5837 (iv_constant): Pass the mode paraeter before the rtx it describes
5838 and change its type to scalar_int_mode. Remove VOIDmode handling.
5839 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
5840 (iv_extend): Likewise.
5841 (shorten_into_mode): Likewise.
5842 (iv_add): Use scalar_int_mode.
5843 (iv_mult): Likewise.
5844 (iv_shift): Likewise.
5845 (canonicalize_iv_subregs): Likewise.
5846 (get_biv_step_1): Pass the outer_mode parameter before the rtx
5847 it describes and change its mode to scalar_int_mode. Also change
5848 the type of the returned inner_mode to scalar_int_mode.
5849 (get_biv_step): Likewise, turning outer_mode from a pointer
5850 into a direct parameter. Update call to get_biv_step_1.
5851 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
5852 iv_constant and get_biv_step.
5853 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
5854 and change its type to scalar_int_mode. Don't initialise iv->mode
5855 to VOIDmode and remove later checks for its still being VOIDmode.
5856 Update calls to iv_analyze_op and iv_analyze_expr. Check
5857 is_a <scalar_int_mode> when changing the mode under consideration.
5858 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
5859 Update call to iv_analyze_expr.
5860 (iv_analyze_op): Add a mode parameter. Reject subregs whose
5861 inner register is not also a scalar_int_mode. Update call to
5863 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
5864 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
5865 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
5866 separate mode class checks. Update calls to iv_analyze. Remove
5867 fix-up of VOIDmodes after iv_analyze_biv.
5868 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
5869 don't have a scalar_int_mode. Update call to biv_p.
5871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5872 Alan Hayward <alan.hayward@arm.com>
5873 David Sherwood <david.sherwood@arm.com>
5875 * cfgexpand.c (convert_debug_memory_address): Use
5876 as_a <scalar_int_mode>.
5877 * combine.c (expand_compound_operation): Likewise.
5878 (make_extraction): Likewise.
5879 (change_zero_ext): Likewise.
5880 (simplify_comparison): Likewise.
5881 * cse.c (cse_insn): Likewise.
5882 * dwarf2out.c (minmax_loc_descriptor): Likewise.
5883 (mem_loc_descriptor): Likewise.
5884 (loc_descriptor): Likewise.
5885 * expmed.c (init_expmed_one_mode): Likewise.
5886 (synth_mult): Likewise.
5887 (emit_store_flag_1): Likewise.
5888 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
5889 of a comparison with size.
5890 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
5891 (reduce_to_bit_field_precision): Likewise.
5892 * function.c (expand_function_end): Likewise.
5893 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
5894 * loop-doloop.c (doloop_modify): Likewise.
5895 * optabs.c (expand_binop): Likewise.
5896 (expand_unop): Likewise.
5897 (expand_copysign_absneg): Likewise.
5898 (prepare_cmp_insn): Likewise.
5899 (maybe_legitimize_operand): Likewise.
5900 * recog.c (const_scalar_int_operand): Likewise.
5901 * rtlanal.c (get_address_mode): Likewise.
5902 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5903 (simplify_cond_clz_ctz): Likewise.
5904 * tree-nested.c (get_nl_goto_field): Likewise.
5905 * tree.c (build_vector_type_for_mode): Likewise.
5906 * var-tracking.c (use_narrower_mode): Likewise.
5908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5909 Alan Hayward <alan.hayward@arm.com>
5910 David Sherwood <david.sherwood@arm.com>
5912 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
5913 * builtins.c (expand_builtin_signbit): Use it.
5914 * cfgexpand.c (expand_debug_expr): Likewise.
5915 * dojump.c (do_jump): Likewise.
5916 (do_compare_and_jump): Likewise.
5917 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
5918 * expmed.c (make_tree): Likewise.
5919 * expr.c (expand_expr_real_2): Likewise.
5920 (expand_expr_real_1): Likewise.
5921 (try_casesi): Likewise.
5922 * fold-const-call.c (fold_const_call_ss): Likewise.
5923 * fold-const.c (unextend): Likewise.
5924 (extract_muldiv_1): Likewise.
5925 (fold_single_bit_test): Likewise.
5926 (native_encode_int): Likewise.
5927 (native_encode_string): Likewise.
5928 (native_interpret_int): Likewise.
5929 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
5930 * internal-fn.c (expand_addsub_overflow): Likewise.
5931 (expand_neg_overflow): Likewise.
5932 (expand_mul_overflow): Likewise.
5933 (expand_arith_overflow): Likewise.
5934 * match.pd: Likewise.
5935 * stor-layout.c (layout_type): Likewise.
5936 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5937 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
5938 * tree-ssanames.c (get_range_info): Likewise.
5939 * tree-switch-conversion.c (array_value_type) Likewise.
5940 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
5941 (vect_recog_divmod_pattern): Likewise.
5942 (vect_recog_mixed_size_cond_pattern): Likewise.
5943 * tree-vrp.c (extract_range_basic): Likewise.
5944 (simplify_float_conversion_using_ranges): Likewise.
5945 * tree.c (int_fits_type_p): Likewise.
5946 * ubsan.c (instrument_bool_enum_load): Likewise.
5947 * varasm.c (mergeable_string_section): Likewise.
5948 (narrowing_initializer_constant_valid_p): Likewise.
5949 (output_constant): Likewise.
5951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5952 Alan Hayward <alan.hayward@arm.com>
5953 David Sherwood <david.sherwood@arm.com>
5955 * machmode.h (NARROWEST_INT_MODE): New macro.
5956 * expr.c (alignment_for_piecewise_move): Use it instead of
5957 GET_CLASS_NARROWEST_MODE (MODE_INT).
5958 (push_block): Likewise.
5959 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
5961 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5964 Alan Hayward <alan.hayward@arm.com>
5965 David Sherwood <david.sherwood@arm.com>
5967 * postreload.c (move2add_valid_value_p): Change the type of the
5968 mode parameter to scalar_int_mode.
5969 (move2add_use_add2_insn): Add a mode parameter and use it instead
5971 (move2add_use_add3_insn): Likewise.
5972 (reload_cse_move2add): Update accordingly.
5974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5975 Alan Hayward <alan.hayward@arm.com>
5976 David Sherwood <david.sherwood@arm.com>
5978 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
5980 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
5981 * optabs.c (expand_unop): Likewise.
5983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5984 Alan Hayward <alan.hayward@arm.com>
5985 David Sherwood <david.sherwood@arm.com>
5987 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
5988 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
5989 (popcount_loc_descriptor): Likewise.
5990 (bswap_loc_descriptor): Likewise.
5991 (rotate_loc_descriptor): Likewise.
5992 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
5993 calling the functions above.
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 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6001 (try_combine): Likewise.
6002 (simplify_if_then_else): Likewise.
6003 * cse.c (cse_insn): Likewise.
6004 * dwarf2out.c (mem_loc_descriptor): Likewise.
6005 * emit-rtl.c (gen_lowpart_common): Likewise.
6006 * simplify-rtx.c (simplify_truncation): Likewise.
6007 (simplify_binary_operation_1): Likewise.
6008 (simplify_const_relational_operation): Likewise.
6009 (simplify_ternary_operation): Likewise.
6010 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6013 Alan Hayward <alan.hayward@arm.com>
6014 David Sherwood <david.sherwood@arm.com>
6016 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6017 * reload.c (push_reload): Likewise.
6018 (find_reloads): Likewise.
6020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6021 Alan Hayward <alan.hayward@arm.com>
6022 David Sherwood <david.sherwood@arm.com>
6024 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6025 (make_compound_operation_int): Likewise.
6026 (change_zero_ext): Likewise.
6027 * expr.c (convert_move): Likewise.
6028 (convert_modes): Likewise.
6029 * fwprop.c (forward_propagate_subreg): Likewise.
6030 * loop-iv.c (get_biv_step_1): Likewise.
6031 * optabs.c (widen_operand): Likewise.
6032 * postreload.c (move2add_valid_value_p): Likewise.
6033 * recog.c (simplify_while_replacing): Likewise.
6034 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6035 (simplify_binary_operation_1): Likewise. Remove redundant
6036 mode equality check.
6038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6039 Alan Hayward <alan.hayward@arm.com>
6040 David Sherwood <david.sherwood@arm.com>
6042 * combine.c (combine_simplify_rtx): Add checks for
6043 is_a <scalar_int_mode>.
6044 (simplify_if_then_else): Likewise.
6045 (make_field_assignment): Likewise.
6046 (simplify_comparison): Likewise.
6047 * ifcvt.c (noce_try_bitop): Likewise.
6048 * loop-invariant.c (canonicalize_address_mult): Likewise.
6049 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6052 Alan Hayward <alan.hayward@arm.com>
6053 David Sherwood <david.sherwood@arm.com>
6055 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6056 is_a <scalar_int_mode> instead of != BLKmode.
6058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6059 Alan Hayward <alan.hayward@arm.com>
6060 David Sherwood <david.sherwood@arm.com>
6062 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6063 instead of != VOIDmode.
6064 * combine.c (if_then_else_cond): Likewise.
6065 (change_zero_ext): Likewise.
6066 * dwarf2out.c (mem_loc_descriptor): Likewise.
6067 (loc_descriptor): Likewise.
6068 * rtlanal.c (canonicalize_condition): Likewise.
6069 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6072 Alan Hayward <alan.hayward@arm.com>
6073 David Sherwood <david.sherwood@arm.com>
6075 * simplify-rtx.c (simplify_binary_operation_1): Use
6076 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6079 Alan Hayward <alan.hayward@arm.com>
6080 David Sherwood <david.sherwood@arm.com>
6082 * wide-int.h (int_traits<unsigned char>) New class.
6083 (int_traits<unsigned short>) Likewise.
6084 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6085 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6087 * combine.c (set_nonzero_bits_and_sign_copies): Use
6088 is_a <scalar_int_mode>.
6089 (find_split_point): Likewise.
6090 (combine_simplify_rtx): Likewise.
6091 (simplify_logical): Likewise.
6092 (expand_compound_operation): Likewise.
6093 (expand_field_assignment): Likewise.
6094 (make_compound_operation): Likewise.
6095 (extended_count): Likewise.
6096 (change_zero_ext): Likewise.
6097 (simplify_comparison): Likewise.
6098 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6099 (ucompare_loc_descriptor): Likewise.
6100 (minmax_loc_descriptor): Likewise.
6101 (mem_loc_descriptor): Likewise.
6102 (loc_descriptor): Likewise.
6103 * expmed.c (init_expmed_one_mode): Likewise.
6104 * lra-constraints.c (lra_constraint_offset): Likewise.
6105 * optabs.c (prepare_libcall_arg): Likewise.
6106 * postreload.c (move2add_note_store): Likewise.
6107 * reload.c (operands_match_p): Likewise.
6108 * rtl.h (load_extend_op): Likewise.
6109 * rtlhooks.c (gen_lowpart_general): Likewise.
6110 * simplify-rtx.c (simplify_truncation): Likewise.
6111 (simplify_unary_operation_1): Likewise.
6112 (simplify_binary_operation_1): Likewise.
6113 (simplify_const_binary_operation): Likewise.
6114 (simplify_const_relational_operation): Likewise.
6115 (simplify_subreg): Likewise.
6116 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6117 * var-tracking.c (adjust_mems): Likewise.
6118 (prepare_call_arguments): Likewise.
6120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6121 Alan Hayward <alan.hayward@arm.com>
6122 David Sherwood <david.sherwood@arm.com>
6124 * machmode.h (is_int_mode): New fuction.
6125 * combine.c (find_split_point): Use it.
6126 (combine_simplify_rtx): Likewise.
6127 (simplify_if_then_else): Likewise.
6128 (simplify_set): Likewise.
6129 (simplify_shift_const_1): Likewise.
6130 (simplify_comparison): Likewise.
6131 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6132 * cse.c (notreg_cost): Likewise.
6133 (cse_insn): Likewise.
6134 * cselib.c (cselib_lookup_1): Likewise.
6135 * dojump.c (do_jump_1): Likewise.
6136 (do_compare_rtx_and_jump): Likewise.
6137 * dse.c (get_call_args): Likewise.
6138 * dwarf2out.c (rtl_for_decl_init): Likewise.
6139 (native_encode_initializer): Likewise.
6140 * expmed.c (emit_store_flag_1): Likewise.
6141 (emit_store_flag): Likewise.
6142 * expr.c (convert_modes): Likewise.
6143 (store_field): Likewise.
6144 (expand_expr_real_1): Likewise.
6145 * fold-const.c (fold_read_from_constant_string): Likewise.
6146 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6147 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6148 * optabs.c (expand_binop): Likewise.
6149 (expand_unop): Likewise.
6150 (expand_abs_nojump): Likewise.
6151 (expand_one_cmpl_abs_nojump): Likewise.
6152 * simplify-rtx.c (mode_signbit_p): Likewise.
6153 (val_signbit_p): Likewise.
6154 (val_signbit_known_set_p): Likewise.
6155 (val_signbit_known_clear_p): Likewise.
6156 (simplify_relational_operation_1): Likewise.
6157 * tree.c (vector_type_mode): Likewise.
6159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6160 Alan Hayward <alan.hayward@arm.com>
6161 David Sherwood <david.sherwood@arm.com>
6163 * machmode.h (smallest_mode_for_size): Fix formatting.
6164 (smallest_int_mode_for_size): New function.
6165 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6166 instead of smallest_mode_for_size.
6167 * combine.c (make_extraction): Likewise.
6168 * config/arc/arc.c (arc_expand_movmem): Likewise.
6169 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6170 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6171 * config/s390/s390.c (s390_expand_insv): Likewise.
6172 * config/sparc/sparc.c (assign_int_registers): Likewise.
6173 * config/spu/spu.c (spu_function_value): Likewise.
6174 (spu_function_arg): Likewise.
6175 * coverage.c (get_gcov_type): Likewise.
6176 (get_gcov_unsigned_t): Likewise.
6177 * dse.c (find_shift_sequence): Likewise.
6178 * expmed.c (store_bit_field_1): Likewise.
6179 * expr.c (convert_move): Likewise.
6180 (store_field): Likewise.
6181 * internal-fn.c (expand_arith_overflow): Likewise.
6182 * optabs-query.c (get_best_extraction_insn): Likewise.
6183 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6184 * stor-layout.c (layout_type): Likewise.
6185 (initialize_sizetypes): Likewise.
6186 * targhooks.c (default_get_mask_mode): Likewise.
6187 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6190 Alan Hayward <alan.hayward@arm.com>
6191 David Sherwood <david.sherwood@arm.com>
6193 * machmode.h (opt_mode::else_blk): New function.
6194 (int_mode_for_mode): Declare.
6195 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6196 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6198 * cfgexpand.c (expand_debug_expr): Likewise.
6199 * combine.c (gen_lowpart_or_truncate): Likewise.
6200 (gen_lowpart_for_combine): Likewise.
6201 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6202 * config/avr/avr.c (avr_to_int_mode): Likewise.
6203 (avr_out_plus_1): Likewise.
6204 (avr_out_plus): Likewise.
6205 (avr_out_round): Likewise.
6206 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6207 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6208 (s390_expand_vcond): Likewise.
6209 * config/spu/spu.c (spu_split_immediate): Likewise.
6210 (spu_expand_mov): Likewise.
6211 * dse.c (get_stored_val): Likewise.
6212 * expmed.c (store_bit_field_1): Likewise.
6213 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6215 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6216 (extract_low_bits): Likewise.
6217 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6218 handling rather than repeating the check.
6219 (emit_group_store): Likewise.
6220 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6221 * optabs.c (expand_absneg_bit): Likewise.
6222 (expand_copysign_absneg): Likewise.
6223 (expand_copysign_bit): Likewise.
6224 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6225 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6226 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6227 * var-tracking.c (prepare_call_arguments): Likewise.
6228 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6229 int_mode_for_mode instead of mode_for_size.
6230 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6233 Alan Hayward <alan.hayward@arm.com>
6234 David Sherwood <david.sherwood@arm.com>
6236 * machmode.h (int_mode_for_size): New function.
6237 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6238 instead of mode_for_size.
6239 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6241 * combine.c (expand_field_assignment): Use int_mode_for_size
6242 instead of mode_for_size.
6243 (make_extraction): Likewise.
6244 (simplify_shift_const_1): Likewise.
6245 (simplify_comparison): Likewise.
6246 * dojump.c (do_jump): Likewise.
6247 * dwarf2out.c (mem_loc_descriptor): Likewise.
6248 * emit-rtl.c (init_derived_machine_modes): Likewise.
6249 * expmed.c (flip_storage_order): Likewise.
6250 (convert_extracted_bit_field): Likewise.
6251 * expr.c (copy_blkmode_from_reg): Likewise.
6252 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6253 * internal-fn.c (expand_mul_overflow): Likewise.
6254 * lower-subreg.c (simple_move): Likewise.
6255 * optabs-libfuncs.c (init_optabs): Likewise.
6256 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6257 * tree.c (vector_type_mode): Likewise.
6258 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6259 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6260 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6261 * tree-vect-stmts.c (vectorizable_load): Likewise.
6262 (vectorizable_store): Likewise.
6264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6265 Alan Hayward <alan.hayward@arm.com>
6266 David Sherwood <david.sherwood@arm.com>
6268 * coretypes.h (pod_mode): New type.
6269 (scalar_int_mode_pod): New typedef.
6270 * machmode.h (pod_mode): New class.
6271 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6272 * genmodes.c (emit_mode_int_n): Update accordingly.
6273 * lower-subreg.h (target_lower_subreg): Change type to
6274 scalar_int_mode_pod.
6275 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6276 scalar_int_mode_pod.
6278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6279 Alan Hayward <alan.hayward@arm.com>
6280 David Sherwood <david.sherwood@arm.com>
6282 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6283 machine_mode to scalar_int_mode.
6284 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6285 (rs6000_option_override_internal): Remove cast to int.
6286 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6287 machine_mode to scalar_int_mode.
6288 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6289 (rs6000_option_override_internal): Remove cast to int.
6290 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6291 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6293 * config/s390/s390.c (s390_expand_builtin): Likewise.
6294 * coretypes.h (scalar_int_mode): New type.
6295 (opt_scalar_int_mode): New typedef.
6296 * machmode.h (scalar_int_mode): New class.
6297 (scalar_int_mode::includes_p): New function.
6298 (byte_mode): Change type to scalar_int_mode.
6299 (word_mode): Likewise.
6300 (ptr_mode): Likewise.
6301 * emit-rtl.c (byte_mode): Likewise.
6302 (word_mode): Likewise.
6303 (ptr_mode): Likewise.
6304 (init_derived_machine_modes): Update accordingly.
6305 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6306 and MODE_PARTIAL_INT.
6307 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6308 opt_scalar_int_mode.
6310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6311 Alan Hayward <alan.hayward@arm.com>
6312 David Sherwood <david.sherwood@arm.com>
6314 * target.def (libgcc_floating_mode_supported_p): Take a
6316 * doc/tm.texi: Regenerate.
6317 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6319 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6320 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6324 Alan Hayward <alan.hayward@arm.com>
6325 David Sherwood <david.sherwood@arm.com>
6327 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6328 * doc/tm.texi: Regenerate.
6329 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6330 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6331 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6332 * targhooks.h (default_floatn_mode): Likewise.
6333 * targhooks.c (default_floatn_mode): Likewise.
6334 * tree.c (build_common_tree_nodes): Update accordingly.
6336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6337 Alan Hayward <alan.hayward@arm.com>
6338 David Sherwood <david.sherwood@arm.com>
6340 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6341 (mode_iterator::iterate_p): Likewise.
6342 (mode_iterator::get_wider): Likewise.
6343 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6346 Alan Hayward <alan.hayward@arm.com>
6347 David Sherwood <david.sherwood@arm.com>
6349 * coretypes.h (opt_scalar_float_mode): New typedef.
6350 * machmode.h (float_mode_for_size): New function.
6351 * emit-rtl.c (double_mode): Delete.
6352 (init_emit_once): Use float_mode_for_size.
6353 * stor-layout.c (layout_type): Likewise.
6354 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6356 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6357 Alan Hayward <alan.hayward@arm.com>
6358 David Sherwood <david.sherwood@arm.com>
6360 * output.h (assemble_real): Take a scalar_float_mode.
6361 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6362 * config/arm/arm.md (consttable_4): Likewise.
6363 (consttable_8): Likewise.
6364 (consttable_16): Likewise.
6365 * config/mips/mips.md (consttable_float): Likewise.
6366 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6367 * varasm.c (assemble_real): Take a scalar_float_mode.
6368 (output_constant_pool_2): Update accordingly.
6369 (output_constant): Likewise.
6371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6372 Alan Hayward <alan.hayward@arm.com>
6373 David Sherwood <david.sherwood@arm.com>
6375 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6376 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6377 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6378 (native_encode_real): Likewise.
6379 (native_interpret_real): Likewise.
6380 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6381 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6384 Alan Hayward <alan.hayward@arm.com>
6385 David Sherwood <david.sherwood@arm.com>
6387 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6388 <scalar_float_mode>. Simplify.
6389 (gen_extend_conv_libfunc): Likewise.
6391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6392 Alan Hayward <alan.hayward@arm.com>
6393 David Sherwood <david.sherwood@arm.com>
6395 * coretypes.h (scalar_float_mode): New type.
6396 * machmode.h (mode_traits::from_int): Use machine_mode if
6397 USE_ENUM_MODES is defined.
6398 (is_a): New function.
6400 (dyn_cast): Likewise.
6401 (scalar_float_mode): New class.
6402 (scalar_float_mode::includes_p): New function.
6403 (is_float_mode): Likewise.
6404 * gdbhooks.py (MachineModePrinter): New class.
6405 (build_pretty_printer): Use it for scalar_float_mode.
6406 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6407 (format_helper::format_helper): Turn into a template.
6408 * genmodes.c (get_mode_class): New function.
6409 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6410 or machine_mode if none.
6411 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6412 as_a <scalar_float_mode>.
6413 * dwarf2out.c (mem_loc_descriptor): Likewise.
6414 (insert_float): Likewise.
6415 (add_const_value_attribute): Likewise.
6416 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6417 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6418 (expand_unop): Update accordingly.
6419 (expand_abs_nojump): Likewise.
6420 (expand_copysign_absneg): Take a scalar_float_mode.
6421 (expand_copysign_bit): Likewise.
6422 (expand_copysign): Update accordingly.
6424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6425 Alan Hayward <alan.hayward@arm.com>
6426 David Sherwood <david.sherwood@arm.com>
6428 * coretypes.h (opt_mode): New class.
6429 * machmode.h (opt_mode): Likewise.
6430 (opt_mode::else_void): New function.
6431 (opt_mode::require): Likewise.
6432 (opt_mode::exists): Likewise.
6433 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6434 (GET_MODE_2XWIDER_MODE): Likewise.
6435 (mode_iterator::get_wider): Update accordingly.
6436 (mode_iterator::get_2xwider): Likewise.
6437 (mode_iterator::get_known_wider): Likewise, turning into a template.
6438 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6439 forcing a wider mode to exist.
6440 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6441 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6442 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6443 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6444 * lower-subreg.c (init_lower_subreg): Likewise.
6445 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6446 on the final iteration.
6447 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6448 a wider mode exists before asking for a move pattern.
6449 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6450 forcing a wider mode to exist.
6451 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6452 returning false if no such mode exists.
6453 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6454 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6455 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6456 Avoid checking for a MODE_INT if we already know the mode is not a
6458 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6459 forcing a wider mode to exist.
6460 (expmed_mult_highpart_optab): Likewise.
6461 (expmed_mult_highpart): Likewise.
6462 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6464 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6465 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6466 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6467 * internal-fn.c (expand_mul_overflow): Update use of
6468 GET_MODE_2XWIDER_MODE.
6469 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6470 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6471 GET_MODE_WIDER_MODE.
6472 (convert_plusminus_to_widen): Likewise.
6473 * tree-switch-conversion.c (array_value_type): Likewise.
6474 * var-tracking.c (emit_note_insn_var_location): Likewise.
6475 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6476 Return false inside rather than outside the loop if no wider mode
6478 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6479 and GET_MODE_2XWIDER_MODE
6480 (can_compare_p): Use else_void.
6481 * gdbhooks.py (OptMachineModePrinter): New class.
6482 (build_pretty_printer): Use it for opt_mode.
6484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6485 Alan Hayward <alan.hayward@arm.com>
6486 David Sherwood <david.sherwood@arm.com>
6488 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6489 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6492 Alan Hayward <alan.hayward@arm.com>
6493 David Sherwood <david.sherwood@arm.com>
6495 * machmode.h (mode_traits): New structure.
6496 (get_narrowest_mode): New function.
6497 (mode_iterator::start): Likewise.
6498 (mode_iterator::iterate_p): Likewise.
6499 (mode_iterator::get_wider): Likewise.
6500 (mode_iterator::get_known_wider): Likewise.
6501 (mode_iterator::get_2xwider): Likewise.
6502 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6503 (FOR_EACH_MODE): Likewise.
6504 (FOR_EACH_MODE_FROM): Likewise.
6505 (FOR_EACH_MODE_UNTIL): Likewise.
6506 (FOR_EACH_WIDER_MODE): Likewise.
6507 (FOR_EACH_2XWIDER_MODE): Likewise.
6508 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6509 * combine.c (simplify_comparison): Likewise
6510 * config/i386/i386.c (type_natural_mode): Likewise.
6511 * cse.c (cse_insn): Likewise.
6512 * dse.c (find_shift_sequence): Likewise.
6513 * emit-rtl.c (init_derived_machine_modes): Likewise.
6514 (init_emit_once): Likewise.
6515 * explow.c (hard_function_value): Likewise.
6516 * expmed.c (extract_fixed_bit_field_1): Likewise.
6517 (extract_bit_field_1): Likewise.
6518 (expand_divmod): Likewise.
6519 (emit_store_flag_1): Likewise.
6520 * expr.c (init_expr_target): Likewise.
6521 (convert_move): Likewise.
6522 (alignment_for_piecewise_move): Likewise.
6523 (widest_int_mode_for_size): Likewise.
6524 (emit_block_move_via_movmem): Likewise.
6525 (copy_blkmode_to_reg): Likewise.
6526 (set_storage_via_setmem): Likewise.
6527 (compress_float_constant): Likewise.
6528 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6529 * optabs-query.c (get_best_extraction_insn): Likewise.
6530 * optabs.c (expand_binop): Likewise.
6531 (expand_twoval_unop): Likewise.
6532 (expand_twoval_binop): Likewise.
6533 (widen_leading): Likewise.
6534 (widen_bswap): Likewise.
6535 (expand_parity): Likewise.
6536 (expand_unop): Likewise.
6537 (prepare_cmp_insn): Likewise.
6538 (prepare_float_lib_cmp): Likewise.
6539 (expand_float): Likewise.
6540 (expand_fix): Likewise.
6541 (expand_sfix_optab): Likewise.
6542 * postreload.c (move2add_use_add2_insn): Likewise.
6543 * reg-stack.c (reg_to_stack): Likewise.
6544 * reginfo.c (choose_hard_reg_mode): Likewise.
6545 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6546 * stor-layout.c (mode_for_size): Likewise.
6547 (smallest_mode_for_size): Likewise.
6548 (mode_for_vector): Likewise.
6549 (finish_bitfield_representative): Likewise.
6550 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6551 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6552 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6553 * var-tracking.c (prepare_call_arguments): Likewise.
6555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6556 Alan Hayward <alan.hayward@arm.com>
6557 David Sherwood <david.sherwood@arm.com>
6559 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6560 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6561 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6562 * machmode.h (mode_size): Move earlier in file.
6563 (mode_precision): Likewise.
6564 (mode_inner): Likewise.
6565 (mode_nunits): Likewise.
6566 (mode_unit_size): Likewise.
6567 (unit_unit_precision): Likewise.
6568 (mode_wider): Likewise.
6569 (mode_2xwider): Likewise.
6570 (machine_mode): New class.
6571 (mode_to_bytes): New function.
6572 (mode_to_bits): Likewise.
6573 (mode_to_precision): Likewise.
6574 (mode_to_inner): Likewise.
6575 (mode_to_unit_size): Likewise.
6576 (mode_to_unit_precision): Likewise.
6577 (mode_to_nunits): Likewise.
6578 (GET_MODE_SIZE): Use mode_to_bytes.
6579 (GET_MODE_BITSIZE): Use mode_to_bits.
6580 (GET_MODE_PRECISION): Use mode_to_precision.
6581 (GET_MODE_INNER): Use mode_to_inner.
6582 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6583 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6584 (GET_MODE_NUNITS): Use mode_to_nunits.
6585 * system.h (ALWAYS_INLINE): New macro.
6586 * config/powerpcspe/powerpcspe-c.c
6587 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6588 int for arg1_mode and arg2_mode.
6590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6591 Alan Hayward <alan.hayward@arm.com>
6592 David Sherwood <david.sherwood@arm.com>
6594 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6595 Prefix mode names with E_ in case statements.
6596 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6597 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6598 (aarch64_split_simd_move): Likewise.
6599 (aarch64_gen_storewb_pair): Likewise.
6600 (aarch64_gen_loadwb_pair): Likewise.
6601 (aarch64_gen_store_pair): Likewise.
6602 (aarch64_gen_load_pair): Likewise.
6603 (aarch64_get_condition_code_1): Likewise.
6604 (aarch64_constant_pool_reload_icode): Likewise.
6605 (get_rsqrte_type): Likewise.
6606 (get_rsqrts_type): Likewise.
6607 (get_recpe_type): Likewise.
6608 (get_recps_type): Likewise.
6609 (aarch64_gimplify_va_arg_expr): Likewise.
6610 (aarch64_simd_container_mode): Likewise.
6611 (aarch64_emit_load_exclusive): Likewise.
6612 (aarch64_emit_store_exclusive): Likewise.
6613 (aarch64_expand_compare_and_swap): Likewise.
6614 (aarch64_gen_atomic_cas): Likewise.
6615 (aarch64_emit_bic): Likewise.
6616 (aarch64_emit_atomic_swap): Likewise.
6617 (aarch64_emit_atomic_load_op): Likewise.
6618 (aarch64_evpc_trn): Likewise.
6619 (aarch64_evpc_uzp): Likewise.
6620 (aarch64_evpc_zip): Likewise.
6621 (aarch64_evpc_ext): Likewise.
6622 (aarch64_evpc_rev): Likewise.
6623 (aarch64_evpc_dup): Likewise.
6624 (aarch64_gen_ccmp_first): Likewise.
6625 (aarch64_gen_ccmp_next): Likewise.
6626 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6627 (alpha_emit_xfloating_libcall): Likewise.
6628 (emit_insxl): Likewise.
6629 (alpha_arg_type): Likewise.
6630 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6631 (arc_preferred_simd_mode): Likewise.
6632 (arc_secondary_reload): Likewise.
6633 (get_arc_condition_code): Likewise.
6634 (arc_print_operand): Likewise.
6635 (arc_legitimate_constant_p): Likewise.
6636 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6637 * config/arc/arc.md (casesi_load): Likewise.
6638 (casesi_compact_jump): Likewise.
6639 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6640 (cc_use_register): Likewise.
6641 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6642 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6643 (arm_init_iwmmxt_builtins): Likewise.
6644 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6645 (neon_expand_vector_init): Likewise.
6646 (arm_attr_length_move_neon): Likewise.
6647 (maybe_get_arm_condition_code): Likewise.
6648 (arm_emit_vector_const): Likewise.
6649 (arm_preferred_simd_mode): Likewise.
6650 (arm_output_iwmmxt_tinsr): Likewise.
6651 (thumb1_output_casesi): Likewise.
6652 (thumb2_output_casesi): Likewise.
6653 (arm_emit_load_exclusive): Likewise.
6654 (arm_emit_store_exclusive): Likewise.
6655 (arm_expand_compare_and_swap): Likewise.
6656 (arm_evpc_neon_vuzp): Likewise.
6657 (arm_evpc_neon_vzip): Likewise.
6658 (arm_evpc_neon_vrev): Likewise.
6659 (arm_evpc_neon_vtrn): Likewise.
6660 (arm_evpc_neon_vext): Likewise.
6661 (arm_validize_comparison): Likewise.
6662 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6663 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6664 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6665 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6666 (c6x_preferred_simd_mode): Likewise.
6667 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6668 (epiphany_rtx_costs): Likewise.
6669 * config/epiphany/predicates.md (proper_comparison_operator):
6671 * config/frv/frv.c (condexec_memory_operand): Likewise.
6672 (frv_emit_move): Likewise.
6673 (output_move_single): Likewise.
6674 (output_condmove_single): Likewise.
6675 (frv_hard_regno_mode_ok): Likewise.
6676 (frv_matching_accg_mode): Likewise.
6677 * config/h8300/h8300.c (split_adds_subs): Likewise.
6678 (h8300_rtx_costs): Likewise.
6679 (h8300_print_operand): Likewise.
6680 (compute_mov_length): Likewise.
6681 (output_logical_op): Likewise.
6682 (compute_logical_op_length): Likewise.
6683 (compute_logical_op_cc): Likewise.
6684 (h8300_shift_needs_scratch_p): Likewise.
6685 (output_a_shift): Likewise.
6686 (compute_a_shift_length): Likewise.
6687 (compute_a_shift_cc): Likewise.
6688 (expand_a_rotate): Likewise.
6689 (output_a_rotate): Likewise.
6690 * config/i386/i386.c (classify_argument): Likewise.
6691 (function_arg_advance_32): Likewise.
6692 (function_arg_32): Likewise.
6693 (function_arg_64): Likewise.
6694 (function_value_64): Likewise.
6695 (ix86_gimplify_va_arg): Likewise.
6696 (ix86_legitimate_constant_p): Likewise.
6697 (put_condition_code): Likewise.
6698 (split_double_mode): Likewise.
6699 (ix86_avx256_split_vector_move_misalign): Likewise.
6700 (ix86_expand_vector_logical_operator): Likewise.
6701 (ix86_split_idivmod): Likewise.
6702 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6703 (ix86_build_const_vector): Likewise.
6704 (ix86_build_signbit_mask): Likewise.
6705 (ix86_match_ccmode): Likewise.
6706 (ix86_cc_modes_compatible): Likewise.
6707 (ix86_expand_branch): Likewise.
6708 (ix86_expand_sse_cmp): Likewise.
6709 (ix86_expand_sse_movcc): Likewise.
6710 (ix86_expand_int_sse_cmp): Likewise.
6711 (ix86_expand_vec_perm_vpermi2): Likewise.
6712 (ix86_expand_vec_perm): Likewise.
6713 (ix86_expand_sse_unpack): Likewise.
6714 (ix86_expand_int_addcc): Likewise.
6715 (ix86_split_to_parts): Likewise.
6716 (ix86_vectorize_builtin_gather): Likewise.
6717 (ix86_vectorize_builtin_scatter): Likewise.
6718 (avx_vpermilp_parallel): Likewise.
6719 (inline_memory_move_cost): Likewise.
6720 (ix86_tieable_integer_mode_p): Likewise.
6721 (x86_maybe_negate_const_int): Likewise.
6722 (ix86_expand_vector_init_duplicate): Likewise.
6723 (ix86_expand_vector_init_one_nonzero): Likewise.
6724 (ix86_expand_vector_init_one_var): Likewise.
6725 (ix86_expand_vector_init_concat): Likewise.
6726 (ix86_expand_vector_init_interleave): Likewise.
6727 (ix86_expand_vector_init_general): Likewise.
6728 (ix86_expand_vector_set): Likewise.
6729 (ix86_expand_vector_extract): Likewise.
6730 (emit_reduc_half): Likewise.
6731 (ix86_emit_i387_round): Likewise.
6732 (ix86_mangle_type): Likewise.
6733 (ix86_expand_round_sse4): Likewise.
6734 (expand_vec_perm_blend): Likewise.
6735 (canonicalize_vector_int_perm): Likewise.
6736 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6737 (expand_vec_perm_1): Likewise.
6738 (expand_vec_perm_interleave3): Likewise.
6739 (expand_vec_perm_even_odd_pack): Likewise.
6740 (expand_vec_perm_even_odd_1): Likewise.
6741 (expand_vec_perm_broadcast_1): Likewise.
6742 (ix86_vectorize_vec_perm_const_ok): Likewise.
6743 (ix86_expand_vecop_qihi): Likewise.
6744 (ix86_expand_mul_widen_hilo): Likewise.
6745 (ix86_expand_sse2_abs): Likewise.
6746 (ix86_expand_pextr): Likewise.
6747 (ix86_expand_pinsr): Likewise.
6748 (ix86_preferred_simd_mode): Likewise.
6749 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6750 * config/i386/sse.md (*andnot<mode>3): Likewise.
6751 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6752 (*<code><mode>3): Likewise.
6753 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6754 (ia64_expand_atomic_op): Likewise.
6755 (ia64_arg_type): Likewise.
6756 (ia64_mode_to_int): Likewise.
6757 (ia64_scalar_mode_supported_p): Likewise.
6758 (ia64_vector_mode_supported_p): Likewise.
6759 (expand_vec_perm_broadcast): Likewise.
6760 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6761 (iq2000_function_arg_advance): Likewise.
6762 (iq2000_function_arg): Likewise.
6763 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6764 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6765 (m68k_libcall_value): Likewise.
6766 (m68k_function_value): Likewise.
6767 (sched_attr_op_type): Likewise.
6768 * config/mcore/mcore.c (mcore_output_move): Likewise.
6769 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6771 (microblaze_function_arg): Likewise.
6772 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6773 (mips_print_operand): Likewise.
6774 (mips_mode_ok_for_mov_fmt_p): Likewise.
6775 (mips_vector_mode_supported_p): Likewise.
6776 (mips_preferred_simd_mode): Likewise.
6777 (mips_expand_vpc_loongson_even_odd): Likewise.
6778 (mips_expand_vec_unpack): Likewise.
6779 (mips_expand_vi_broadcast): Likewise.
6780 (mips_expand_vector_init): Likewise.
6781 (mips_expand_vec_reduc): Likewise.
6782 (mips_expand_msa_cmp): Likewise.
6783 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6784 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6785 (cc_flags_for_mode): Likewise.
6786 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6787 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6788 (nds32_output_casesi_pc_relative): Likewise.
6789 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6790 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6791 (nvptx_gen_unpack): Likewise.
6792 (nvptx_gen_pack): Likewise.
6793 (nvptx_gen_shuffle): Likewise.
6794 (nvptx_gen_wcast): Likewise.
6795 (nvptx_preferred_simd_mode): Likewise.
6796 * config/pa/pa.c (pa_secondary_reload): Likewise.
6797 * config/pa/predicates.md (base14_operand): Likewise.
6798 * config/powerpcspe/powerpcspe-c.c
6799 (altivec_resolve_overloaded_builtin): Likewise.
6800 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6802 (rs6000_preferred_simd_mode): Likewise.
6803 (output_vec_const_move): Likewise.
6804 (rs6000_expand_vector_extract): Likewise.
6805 (rs6000_split_vec_extract_var): Likewise.
6806 (reg_offset_addressing_ok_p): Likewise.
6807 (rs6000_legitimate_offset_address_p): Likewise.
6808 (rs6000_legitimize_address): Likewise.
6809 (rs6000_emit_set_const): Likewise.
6810 (rs6000_const_vec): Likewise.
6811 (rs6000_emit_move): Likewise.
6812 (spe_build_register_parallel): Likewise.
6813 (rs6000_darwin64_record_arg_recurse): Likewise.
6814 (swap_selector_for_mode): Likewise.
6815 (spe_init_builtins): Likewise.
6816 (paired_init_builtins): Likewise.
6817 (altivec_init_builtins): Likewise.
6818 (do_load_for_compare): Likewise.
6819 (rs6000_generate_compare): Likewise.
6820 (rs6000_expand_float128_convert): Likewise.
6821 (emit_load_locked): Likewise.
6822 (emit_store_conditional): Likewise.
6823 (rs6000_output_function_epilogue): Likewise.
6824 (rs6000_handle_altivec_attribute): Likewise.
6825 (rs6000_function_value): Likewise.
6826 (emit_fusion_gpr_load): Likewise.
6827 (emit_fusion_p9_load): Likewise.
6828 (emit_fusion_p9_store): Likewise.
6829 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6830 (fusion_gpr_mem_load): Likewise.
6831 (fusion_addis_mem_combo_load): Likewise.
6832 (fusion_addis_mem_combo_store): Likewise.
6833 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
6834 (fusion_gpr_mem_load): Likewise.
6835 (fusion_addis_mem_combo_load): Likewise.
6836 (fusion_addis_mem_combo_store): Likewise.
6837 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6839 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
6840 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6841 (rs6000_preferred_simd_mode): Likewise.
6842 (output_vec_const_move): Likewise.
6843 (rs6000_expand_vector_extract): Likewise.
6844 (rs6000_split_vec_extract_var): Likewise.
6845 (reg_offset_addressing_ok_p): Likewise.
6846 (rs6000_legitimate_offset_address_p): Likewise.
6847 (rs6000_legitimize_address): Likewise.
6848 (rs6000_emit_set_const): Likewise.
6849 (rs6000_const_vec): Likewise.
6850 (rs6000_emit_move): Likewise.
6851 (rs6000_darwin64_record_arg_recurse): Likewise.
6852 (swap_selector_for_mode): Likewise.
6853 (paired_init_builtins): Likewise.
6854 (altivec_init_builtins): Likewise.
6855 (rs6000_expand_float128_convert): Likewise.
6856 (emit_load_locked): Likewise.
6857 (emit_store_conditional): Likewise.
6858 (rs6000_output_function_epilogue): Likewise.
6859 (rs6000_handle_altivec_attribute): Likewise.
6860 (rs6000_function_value): Likewise.
6861 (emit_fusion_gpr_load): Likewise.
6862 (emit_fusion_p9_load): Likewise.
6863 (emit_fusion_p9_store): Likewise.
6864 * config/rx/rx.c (rx_gen_move_template): Likewise.
6865 (flags_from_mode): Likewise.
6866 * config/s390/predicates.md (s390_alc_comparison): Likewise.
6867 (s390_slb_comparison): Likewise.
6868 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
6869 (s390_vector_mode_supported_p): Likewise.
6870 (s390_cc_modes_compatible): Likewise.
6871 (s390_match_ccmode_set): Likewise.
6872 (s390_canonicalize_comparison): Likewise.
6873 (s390_emit_compare_and_swap): Likewise.
6874 (s390_branch_condition_mask): Likewise.
6875 (s390_rtx_costs): Likewise.
6876 (s390_secondary_reload): Likewise.
6877 (__SECONDARY_RELOAD_CASE): Likewise.
6878 (s390_expand_cs): Likewise.
6879 (s390_preferred_simd_mode): Likewise.
6880 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
6881 * config/sh/sh.c (sh_print_operand): Likewise.
6882 (dump_table): Likewise.
6883 (sh_secondary_reload): Likewise.
6884 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6885 * config/sh/sh.md (casesi_worker_1): Likewise.
6886 (casesi_worker_2): Likewise.
6887 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
6888 (fcc_comparison_operator): Likewise.
6889 * config/sparc/sparc.c (sparc_expand_move): Likewise.
6890 (emit_soft_tfmode_cvt): Likewise.
6891 (sparc_preferred_simd_mode): Likewise.
6892 (output_cbranch): Likewise.
6893 (sparc_print_operand): Likewise.
6894 (sparc_expand_vec_perm_bmask): Likewise.
6895 (vector_init_bshuffle): Likewise.
6896 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6897 (spu_vector_mode_supported_p): Likewise.
6898 (spu_expand_insv): Likewise.
6899 (spu_emit_branch_or_set): Likewise.
6900 (spu_handle_vector_attribute): Likewise.
6901 (spu_builtin_splats): Likewise.
6902 (spu_builtin_extract): Likewise.
6903 (spu_builtin_promote): Likewise.
6904 (spu_expand_sign_extend): Likewise.
6905 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6906 (tilegx_simd_int): Likewise.
6907 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
6908 (tilepro_simd_int): Likewise.
6909 * config/v850/v850.c (const_double_split): Likewise.
6910 (v850_print_operand): Likewise.
6911 (ep_memory_offset): Likewise.
6912 * config/vax/vax.c (vax_rtx_costs): Likewise.
6913 (vax_output_int_move): Likewise.
6914 (vax_output_int_add): Likewise.
6915 (vax_output_int_subtract): Likewise.
6916 * config/visium/predicates.md (visium_branch_operator): Likewise.
6917 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
6918 (visium_print_operand_address): Likewise.
6919 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6920 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
6921 (xtensa_expand_conditional_branch): Likewise.
6922 (xtensa_copy_incoming_a7): Likewise.
6923 (xtensa_output_literal): Likewise.
6924 * dfp.c (decimal_real_maxval): Likewise.
6925 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6928 Alan Hayward <alan.hayward@arm.com>
6929 David Sherwood <david.sherwood@arm.com>
6931 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
6932 (mode_nunits_inline): Likewise.
6933 (mode_inner_inline): Likewise.
6934 (mode_unit_size_inline): Likewise.
6935 (mode_unit_precision_inline): Likewise.
6936 (emit_insn_modes_h): Likewise. Also emit a #define of the
6938 (emit_mode_wider): Add an E_ prefix to mode names.
6939 (emit_mode_complex): Likewise.
6940 (emit_mode_inner): Likewise.
6941 (emit_mode_adjustments): Likewise.
6942 (emit_mode_int_n): Likewise.
6943 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
6944 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
6945 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
6946 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
6947 (CRC32_BUILTIN, ENTRY): Likewise.
6948 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
6949 (aarch64_pop_regs): Likewise.
6950 (aarch64_process_components): Likewise.
6951 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
6952 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
6953 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
6954 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
6955 * config/arm/arm.c (arm_init_libfuncs): Likewise.
6956 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
6958 * config/i386/i386-builtin.def (pcmpestr): Likewise.
6959 (pcmpistr): Likewise.
6960 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
6961 * config/mmix/mmix.c (mmix_output_condition): Likewise.
6962 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6964 * config/rl78/rl78.c (mduc_regs): Likewise.
6965 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6966 (htm_expand_builtin): Likewise.
6967 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
6968 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
6969 * config/xtensa/xtensa.c (print_operand): Likewise.
6970 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
6971 (NUM_MODE_VECTOR_INT): Likewise.
6972 * genoutput.c (null_operand): Likewise.
6973 (output_operand_data): Likewise.
6974 * genrecog.c (print_parameter_value): Likewise.
6975 * lra.c (debug_operand_data): Likewise.
6977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6979 * dwarf2out.c (scompare_loc_descriptor_wide)
6980 (scompare_loc_descriptor_narrow): New functions, split out from...
6981 (scompare_loc_descriptor): ...here.
6982 * expmed.c (emit_store_flag_int): New function, split out from...
6983 (emit_store_flag): ...here.
6985 2017-08-30 Richard Biener <rguenther@suse.de>
6987 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
6988 (dwarf2out_early_finish): Move setting of AT_pubnames from
6989 early debug output to early finish.
6991 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6993 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
6994 and -mdata-region to the assembler.
6996 2017-08-30 Richard Biener <rguenther@suse.de>
6998 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7000 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7001 (dwarf2out_early_global_decl): Only generate a DIE for the
7002 abstract origin if it doesn't already exist or is a declaration DIE.
7003 (resolve_addr): Do not add the linkage name twice when
7004 generating a stub DIE for the DW_TAG_GNU_call_site target.
7006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7008 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7009 Use machine_mode rather than int for arg1_mode.
7011 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7014 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7015 that the second argument of the built-in functions to unpack
7016 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7017 switch statement instead a lot of if statements.
7018 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7019 Allow 64-bit values to be in Altivec registers as well as
7020 traditional floating point registers.
7021 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7023 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7025 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7026 MAX_REGS_PER_ADDRESS == 1.
7028 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7030 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7031 * config/i386/i386.c (ix86_option_override_internal): Simplify
7032 setting of opts->x_flag_entry.
7034 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7035 Jakub Jelinek <jakub@redhat.com>
7036 Richard Biener <rguenther@suse.de>
7038 PR tree-optimization/81503
7039 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7040 folded constant fits in the target type; reorder tests for clarity.
7042 2017-08-29 Martin Liska <mliska@suse.cz>
7044 * passes.def: Include pass_lower_switch.
7045 * stmt.c (dump_case_nodes): Remove and move to
7046 tree-switch-conversion.
7047 (case_values_threshold): Likewise.
7048 (expand_switch_as_decision_tree_p): Likewise.
7049 (emit_case_decision_tree): Likewise.
7050 (expand_case): Likewise.
7051 (balance_case_nodes): Likewise.
7052 (node_has_low_bound): Likewise.
7053 (node_has_high_bound): Likewise.
7054 (node_is_bounded): Likewise.
7055 (emit_case_nodes): Likewise.
7056 (struct simple_case_node): New struct.
7057 (add_case_node): Remove.
7058 (emit_case_dispatch_table): Use vector instead of case_list.
7059 (reset_out_edges_aux): Remove.
7060 (compute_cases_per_edge): Likewise.
7061 (expand_case): Build list of simple_case_node.
7062 (expand_sjlj_dispatch_table): Use it.
7063 * tree-switch-conversion.c (struct case_node): Moved from
7064 stmt.c and adjusted.
7065 (emit_case_nodes): Likewise.
7066 (node_has_low_bound): Likewise.
7067 (node_has_high_bound): Likewise.
7068 (node_is_bounded): Likewise.
7069 (case_values_threshold): Likewise.
7070 (reset_out_edges_aux): Likewise.
7071 (compute_cases_per_edge): Likewise.
7072 (add_case_node): Likewise.
7073 (dump_case_nodes): Likewise.
7074 (balance_case_nodes): Likewise.
7075 (expand_switch_as_decision_tree_p): Likewise.
7076 (emit_jump): Likewise.
7077 (emit_case_decision_tree): Likewise.
7078 (try_switch_expansion): Likewise.
7079 (do_jump_if_equal): Likewise.
7080 (emit_cmp_and_jump_insns): Likewise.
7081 (fix_phi_operands_for_edge): New function.
7082 (record_phi_operand_mapping): Likewise.
7083 (class pass_lower_switch): New pass.
7084 (pass_lower_switch::execute): New function.
7085 (make_pass_lower_switch): Likewise.
7086 (conditional_probability):
7087 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7088 * tree-pass.h: Add make_pass_lower_switch.
7090 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7093 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7096 2017-08-29 Richard Biener <rguenther@suse.de>
7098 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7099 we do not add a DW_AT_inline attribute twice.
7100 (gen_subprogram_die): Remove code setting DW_AT_inline on
7101 DECL_ABSTRACT_P nodes.
7103 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7105 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7106 calls to internal functions.
7107 (gimplify_modify_expr): Likewise.
7108 * tree-call-cdce.c (use_internal_fn): Likewise.
7109 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7110 (convert_to_divmod): Set the nothrow flag.
7111 * tree-if-conv.c (predicate_mem_writes): Likewise.
7112 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7113 (vectorizable_call): Likewise.
7114 (vectorizable_store): Likewise.
7115 (vectorizable_load): Likewise.
7116 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7117 (vect_recog_mask_conversion_pattern): Likewise.
7119 2017-08-29 Martin Liska <mliska@suse.cz>
7122 * gcc.c (driver_handle_option): Add new argument.
7123 * opts-common.c (handle_option): Pass
7124 target_option_override_hook.
7125 * opts-global.c (lang_handle_option): Add new option.
7126 (set_default_handlers): Add new argument.
7127 (decode_options): Likewise.
7128 * opts.c (target_handle_option): Likewise.
7129 (common_handle_option): Call target_option_override_hook.
7130 * opts.h (struct cl_option_handler_func): Add hook for
7131 target option override.
7132 (struct cl_option_handlers): Likewise.
7133 (set_default_handlers): Add new argument.
7134 (decode_options): Likewise.
7135 (common_handle_option): Likewise.
7136 (target_handle_option): Likewise.
7137 * toplev.c (toplev::main): Pass targetm.target_option.override
7140 2017-08-29 Richard Biener <rguenther@suse.de>
7141 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7143 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7144 life to the active subtree.
7146 2017-08-28 Jeff Law <law@redhat.com>
7148 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7149 derive_equivalences.
7150 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7152 (edge_info::derive_equivalences): New private member function
7154 * tree-ssa-dom.c (class edge_info): Changed from a struct
7155 to a class. Add ctor/dtor, methods and data members.
7156 (edge_info::edge_info): Renamed from allocate_edge_info.
7157 Initialize additional members.
7158 (edge_info::~edge_info): New.
7159 (free_dom_edge_info): Delete the edge info.
7160 (record_edge_info): Use new class & associated member functions.
7161 Tighten forms for testing for edge equivalences.
7162 (record_temporary_equivalences): Iterate over the simple
7163 equivalences rather than assuming there's only one per edge.
7164 (cprop_into_successor_phis): Iterate over the simple
7165 equivalences rather than assuming there's only one per edge.
7166 (optimize_stmt): Use operand_equal_p rather than pointer
7167 equality for mini-DSE code.
7169 2017-08-28 Nathan Sidwell <nathan@acm.org>
7171 * gcc.c (execute): Fold SIGPIPE handling into switch
7172 statement. Adjust internal error message.
7174 2017-08-28 Richard Biener <rguenther@suse.de>
7177 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7178 Do nothing for removed DIEs.
7180 2017-08-28 Richard Biener <rguenther@suse.de>
7182 PR tree-optimization/81977
7183 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7186 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7189 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7190 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7191 using targetm.gen_mem_thread_fence.
7193 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7196 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7197 predicate to register_operand. Reorder operands.
7198 (*btr<mode>): Ditto.
7199 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7200 (*btr<mode>_mask): Ditto.
7202 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7204 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7205 * config/rs6000/xmmintrin.h: New file.
7206 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7208 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7211 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7212 parameter and_insn and return it.
7213 (recombine_lvx_pattern): Insert a copy to ensure availability of
7214 the base register of the copied masking operation at the point of
7215 the instruction replacement.
7216 (recombine_stvx_pattern): Likewise.
7218 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7220 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7221 undocumented switches.
7222 (-mpower9-dform-vector): Likewise.
7223 (-mpower9-dform): Likewise.
7224 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7225 comments to delete references to -mpower9-dform* switches.
7226 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7227 Delete reference to -mpower9-dform* switches, test for
7228 -mpower9-vector instead.
7229 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7230 (OTHER_P9_VECTOR_MASKS): Likewise.
7231 (POWERPC_MASKS): Likewise.
7232 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7233 tests against -mpower9-dform* to -mpower9-vector. Delete code
7234 that checked for -mpower9-dform* consistancy with other options.
7235 Add test for -mpower9-misc to enable other power9 switches.
7236 (rs6000_init_hard_regno_mode_ok): Likewise.
7237 (rs6000_option_override_internal): Likewise.
7238 (rs6000_emit_prologue): Likewise.
7239 (rs6000_emit_epilogue): Likewise.
7240 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7241 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7242 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7244 (emit_fusion_p9_store): Likewise.
7245 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7246 resetting these macros if the assembler does not support ISA 3.0
7248 (TARGET_P9_DFORM_VECTOR): Likewise.
7249 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7250 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7252 2017-08-25 Alan Modra <amodra@gmail.com>
7254 PR rtl-optimization/81747
7255 * cse.c (cse_extended_basic_block): Don't attempt to record
7256 equivalences for degenerate conditional branches that branch
7257 to their fall-through.
7259 2017-08-24 Martin Sebor <msebor@redhat.com>
7262 * gimple-fold.c (size_must_be_zero_p): New function.
7263 (gimple_fold_builtin_memory_op): Call it.
7265 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7267 * config/rs6000/mm_malloc.h: New file.
7269 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7271 PR tree-optimization/81913
7272 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7273 analysis when either IVs in condition can wrap.
7275 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7277 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7278 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7280 2017-08-24 Richard Biener <rguenther@suse.de>
7283 * targhooks.c (default_target_can_inline_p): Properly
7284 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7285 is present and always compare.
7286 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7287 infer -mfpmath=sse from TARGET_SSE_P.
7288 (ix86_can_inline_p): Properly use target_option_default_node when
7289 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7291 2017-08-24 Richard Biener <rguenther@suse.de>
7294 * dwarf2out.c (output_die): Handle flag_generate_offload like
7296 (output_comp_unit): Likewise.
7297 (gen_array_type_die): Likewise.
7298 (dwarf2out_early_finish): Likewise.
7299 (note_variable_value_in_expr): Likewise.
7300 (dwarf2out_finish): Likewise. Adjust assert.
7301 * cgraphunit.c (symbol_table::compile): Move setting of
7302 flag_generate_offload earlier ...
7303 (symbol_table::finalize_compilation_unit): ... here, before
7304 early debug finalization.
7306 2017-08-24 Richard Biener <rguenther@suse.de>
7308 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7309 and ipa-fnsummary.h.
7310 (ix86_can_inline_p): When ix86_fpmath flags do not match
7311 check whether the callee uses FP math at all.
7313 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7316 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7317 instead of TYPE_PRECISION.
7319 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7320 Alan Hayward <alan.hayward@arm.com>
7321 David Sherwood <david.sherwood@arm.com>
7323 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7325 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7327 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7329 (ix86_frame::stack_realign_allocate): New field.
7330 (struct machine_frame_state): Modify comments.
7331 (machine_frame_state::sp_realigned_fp_end): New field.
7332 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7334 (sp_valid_at): Add assertion to assure no attempt to access invalid
7335 offset of a realigned stack.
7336 (fp_valid_at): Likewise.
7337 (choose_baseaddr): Modify comments.
7338 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7339 ix86_expand_prologue.
7340 (ix86_expand_prologue): Modify stack realignment and allocation.
7341 (ix86_expand_epilogue): Modify comments.
7342 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7343 avx2_runtime, avx512f, and avx512f_runtime.
7345 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7347 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7348 (mstackrealign): Do not init to -1.
7349 * config/i386/i386.c (ix86_option_override_internal):
7350 Check opts_set, not opts when setting default value of
7351 opts->x_ix86_force_align_arg_pointer.
7353 2017-08-23 Richard Biener <rguenther@suse.de>
7355 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7356 lang_hooks.decl_printable_name.
7357 * print-rtl-function.c (print_rtx_function): Likewise.
7358 * tree-pretty-print.c (dump_function_header): Likewise.
7360 2017-08-23 Richard Biener <rguenther@suse.de>
7363 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7364 -g0 at compile-time.
7366 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7369 * doc/sourcebuild.texi (Other hardware attributes):
7372 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7375 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7376 Fix single-use check.
7378 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7380 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7382 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7384 * doc/install.texi: Modify to add more details on running selected
7387 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7389 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7390 is combined with -mabi=ms.
7391 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7394 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7396 PR tree-optimization/81488
7397 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7398 and cached_basis fields.
7399 (MAX_SPREAD): New constant.
7400 (alloc_cand_and_find_basis): Initialize new fields.
7401 (clear_visited): New function.
7402 (create_phi_basis_1): Rename from create_phi_basis, set visited
7403 and cached_basis fields.
7404 (create_phi_basis): New wrapper function.
7405 (phi_add_costs_1): Rename from phi_add_costs, add spread
7406 parameter, set visited field, short-circuit when limits reached.
7407 (phi_add_costs): New wrapper function.
7408 (record_phi_increments_1): Rename from record_phi_increments, set
7410 (record_phi_increments): New wrapper function.
7411 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7412 (phi_incr_cost): New wrapper function.
7413 (all_phi_incrs_profitable_1): Rename from
7414 all_phi_incrs_profitable, set visited field.
7415 (all_phi_incrs_profitable): New wrapper function.
7417 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7418 Alan Hayward <alan.hayward@arm.com>
7419 David Sherwood <david.sherwood@arm.com>
7421 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7422 that takes the outer and inner modes.
7423 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7424 comparison as the canonical test for a paradoxical subreg.
7425 * combine.c (simplify_set): Use paradoxical_subreg_p.
7426 (make_extraction): Likewise.
7427 (force_to_mode): Likewise.
7428 (rtx_equal_for_field_assignment_p): Likewise.
7429 (gen_lowpart_for_combine): Likewise.
7430 (simplify_comparison): Likewise.
7431 * cse.c (equiv_constant): Likewise.
7432 * expmed.c (store_bit_field_1): Likewise.
7433 * final.c (alter_subreg): Likewise.
7434 * fwprop.c (propagate_rtx): Likewise.
7435 (forward_propagate_subreg): Likewise.
7436 * ira-conflicts.c (ira_build_conflicts): Likewise.
7437 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7438 * lra-constraints.c (curr_insn_transform): Likewise.
7439 (split_reg): Likewise.
7440 * lra-eliminations.c (move_plus_up): Likewise.
7441 (lra_eliminate_regs_1): Likewise.
7442 * recog.c (general_operand): Likewise.
7443 * ree.c (combine_reaching_defs): Likewise.
7444 * reload.c (push_reload): Likewise.
7445 (find_reloads): Likewise.
7446 * reload1.c (elimination_effects): Likewise.
7447 (compute_reload_subreg_offset): Likewise.
7448 (choose_reload_regs): Likewise.
7449 * rtlanal.c (subreg_lsb_1): Likewise.
7450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7451 (simplify_subreg): Likewise.
7452 * var-tracking.c (track_loc_p): Likewise.
7453 * emit-rtl.c (byte_lowpart_offset): Likewise.
7454 (paradoxical_subreg_p): Delete out-of-line definition.
7456 2017-08-22 Jeff Law <law@redhat.com>
7458 PR tree-optimization/81741
7459 PR tree-optimization/71947
7460 * tree-ssa-dom.c: Include tree-inline.h.
7461 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7462 equivalences if one is more expensive to compute than the other.
7463 * tree-ssa-scopedtables.h (class const_or_copies): Make
7464 record_const_or_copy_raw method private.
7465 (class avail_exprs_stack): New method simplify_binary_operation.
7466 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7467 avail_exprs_stack::simplify_binary_operation as needed.
7468 (avail_exprs_stack::simplify_binary_operation): New function.
7470 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7472 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7473 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7474 (DOT_SYMBOLS): Likewise.
7475 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7476 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7477 (RS6000_ABI_NAME): Likewise.
7478 (TARGET_CMODEL): Likewise.
7479 (TOC_SECTION_ASM_OP): Likewise.
7480 (SET_CMODEL): New macro.
7481 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7483 2017-08-22 Richard Biener <rguenther@suse.de>
7485 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7486 to allow for free-lang-data replacements similar to verify_type_variant.
7488 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7489 Alan Hayward <alan.hayward@arm.com>
7490 David Sherwood <david.sherwood@arm.com>
7492 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7493 VOIDmode for the LABEL_REF.
7495 2017-08-22 Richard Biener <rguenther@suse.de>
7497 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7499 2017-08-22 Richard Biener <rguenther@suse.de>
7501 * common.opt (feliminate-dwarf2-dups): Ignore.
7502 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7503 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7504 same_die_p_wrap, compute_section_prefix,
7505 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7506 (comdat_symbol_id, comdat_symbol_number): Likewise.
7507 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7509 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7510 (output_die): Mark unreachable path unreachable.
7511 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7512 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7513 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7514 (dwarf2out_early_finish): Likewise.
7516 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7518 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7520 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7523 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7524 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7525 (avr_attribute_table) <io, io_low, address>: Initialize
7526 .decl_required with true.
7528 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7530 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7531 undocumented debugging options.
7532 (-mvsx-scalar-double): Likewise.
7533 (-mallow-df-permute): Likewise.
7534 (-mvectorize-builtins): Likewise.
7535 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7536 (rs6000_builtin_vectorized_function): Likewise.
7537 (rs6000_builtin_md_vectorized_function): Likewise.
7538 (rs6000_opt_vars): Likewise.
7540 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7543 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7544 (*btrq_imm): Rename from *btrq.
7545 (*btcq_imm): Rename from *btcq.
7546 (btsc): New code attribute.
7547 (*<btsc><mode>): New insn pattern.
7548 (*btr<mode>): Ditto.
7549 (*<btsc><mode>_mask): New insn_and_split pattern.
7550 (*btr<mode>_mask): Ditto.
7552 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7553 Alan Hayward <alan.hayward@arm.com>
7554 David Sherwood <david.sherwood@arm.com>
7556 * function.c (pad_below): Simplify padding calculation.
7558 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7559 Alan Hayward <alan.hayward@arm.com>
7560 David Sherwood <david.sherwood@arm.com>
7562 * target.def (function_prologue): Remove frame size argument.
7563 (function_epilogue): Likewise.
7564 * doc/tm.texi: Regenerate.
7565 * final.c (final_start_function): Update call to function_prologue.
7566 (final_end_function): Update call to function_epilogue.
7567 (default_function_pro_epilogue): Remove frame size argument.
7568 * output.h (default_function_pro_epilogue): Likewise.
7569 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7570 (arm_output_function_prologue): Likewise.
7571 * config/frv/frv.c (frv_function_prologue): Likewise.
7572 (frv_function_epilogue): Likewise.
7573 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7574 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7575 (ia64_output_function_epilogue): Likewise.
7576 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7577 (m32r_output_function_epilogue): Likewise.
7578 * config/microblaze/microblaze.c (microblaze_function_prologue)
7579 (microblaze_function_epilogue): Likewise.
7580 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7581 (mips_output_function_epilogue): Likewise.
7582 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7583 (mmix_target_asm_function_epilogue): Likewise.
7584 * config/msp430/msp430.c (msp430_start_function): Likewise.
7585 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7586 (nds32_asm_function_epilogue): Likewise.
7587 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7588 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7589 (pa_output_function_epilogue): Likewise.
7590 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7591 (rs6000_output_function_epilogue): Likewise.
7592 * config/rl78/rl78.c (rl78_start_function): Likewise.
7593 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7594 (rs6000_output_function_epilogue): Likewise.
7595 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7596 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7597 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7598 (sparc_asm_function_epilogue): Likewise.
7600 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7602 * tree.h (type_has_mode_precision_p): New function.
7603 * convert.c (convert_to_integer_1): Use it.
7604 * expr.c (expand_expr_real_2): Likewise.
7605 (expand_expr_real_1): Likewise.
7606 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7607 * match.pd: Likewise.
7608 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7609 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7610 * tree-tailcall.c (process_assignment): Likewise.
7611 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7612 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7613 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7614 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7615 (vectorizable_assignment): Likewise.
7616 (vectorizable_shift): Likewise.
7617 (vectorizable_operation): Likewise.
7618 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7620 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7622 * match.pd: Add pow (C, x) simplification.
7624 2017-08-21 Richard Biener <rguenther@suse.de>
7626 PR tree-optimization/81900
7627 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7628 for blocks with abnormal predecessors.
7629 (compute_antic): Do not set visited flag prematurely.
7631 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7634 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7636 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7638 * stor-layout.h (vector_type_mode): Move to...
7639 * tree.h (vector_type_mode): ...here.
7640 * stor-layout.c (vector_type_mode): Move to...
7641 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7643 2017-08-21 Richard Biener <rguenther@suse.de>
7645 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7646 register_external_die hooks.
7647 (debug_false_tree_charstarstar_uhwistar): Declare.
7648 (debug_nothing_tree_charstar_uhwi): Likewise.
7649 * debug.c (do_nothing_debug_hooks): Adjust.
7650 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7651 (debug_nothing_tree_charstar_uhwi): Likewise.
7652 * dbxout.c (dbx_debug_hooks): Adjust.
7653 (xcoff_debug_hooks): Likewise.
7654 * sdbout.c (sdb_debug_hooks): Likewise.
7655 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7656 * dwarf2out.c (macinfo_label_base): New global.
7657 (dwarf2out_register_external_die): New function for the
7658 register_external_die hook.
7659 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7660 (dwarf2_debug_hooks): Use them.
7661 (dwarf2_lineno_debug_hooks): Adjust.
7662 (struct die_struct): Add with_offset flag.
7663 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7664 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7665 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7666 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7667 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7668 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7669 defining section names for the early LTO debug variants.
7670 (reset_indirect_string): New helper.
7671 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7672 (print_dw_val): Add support for offsetted symbol references.
7673 (get_ultimate_context): Split out from is_cxx.
7674 (is_cxx): Use get_ultimate_context.
7675 (is_fortran): Add decl overload.
7676 (compute_comp_unit_symbol): Split out worker from
7677 compute_section_prefix.
7678 (compute_section_prefix): Call compute_comp_unit_symbol and
7679 set comdat_type_p here.
7680 (output_die): Skip DIE symbol output for the LTO added one.
7681 Handle DIE symbol references with offset.
7682 (output_comp_unit): Guard section name mangling properly.
7683 For LTO debug sections emit a symbol at the section beginning
7684 which we use to refer to its DIEs.
7685 (add_abstract_origin_attribute): For DIEs registered via
7686 dwarf2out_register_external_die directly refer to the early
7687 DIE rather than indirectly through the shadow one we created.
7688 Remove obsolete call to dwarf2out_abstract_function for
7689 non-function/block origins.
7690 (gen_array_type_die): When generating early LTO debug do
7691 not emit DW_AT_string_length.
7692 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7693 late when in LTO. As suggested place a gcc_unreachable for
7694 the DECL_ABSTRACT_P case.
7695 (gen_subprogram_die): Avoid another specification DIE
7696 for early built declarations/definitions for the late LTO case.
7697 (gen_variable_die): Add type references for late duplicated VLA dies
7699 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7700 we have the abstract instance already.
7701 (process_scope_var): Adjust decl DIE contexts in LTO which
7702 first puts them in limbo.
7703 (gen_decl_die): Do not generate type DIEs late apart from
7704 types for VLAs or for decls we do not yet have a DIE. Do not
7705 call dwarf2out_abstract_function late.
7706 (dwarf2out_early_global_decl): Make sure to create DIEs
7707 for abstract instances of a decl first.
7708 (dwarf2out_late_global_decl): Adjust comment.
7709 (output_macinfo_op): With multiple macro sections use
7710 macinfo_label_base to distinguish labels.
7711 (output_macinfo): Likewise. Update macinfo_label_base.
7712 Pass in the line info label.
7713 (note_variable_value_in_expr): When generating LTO resolve
7714 all variable values here by generating DIEs as needed.
7715 (init_sections_and_labels): Add early LTO debug flag parameter
7716 and generate different sections and names if set. Add generation
7717 counter for the labels so we can have multiple of them.
7718 (reset_dies): Helper to allow DIEs to be output multiple times.
7719 (dwarf2out_finish): When outputting DIEs to the fat part of an
7720 LTO object first reset DIEs.
7721 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7722 (modified_type_die): Check for decl_ultimate_origin being self
7724 (gen_type_die_with_usage): Likewise.
7725 (gen_typedef_die): Allow decl_ultimate_origin being self.
7726 (set_decl_abstract_flags): Remove.
7727 (set_block_abstract_flags): Likewise.
7728 (dwarf2out_abstract_function): Treat the early generated DIEs
7729 as the abstract copy and only add DW_AT_inline and
7730 DW_AT_artificial here and call set_decl_origin_self.
7731 If the DIE has an abstract origin don't do anything.
7732 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7733 if we have none yet (Go fails to build one, PR78628).
7734 (variably_modified_type_p): Prevent endless recursion for Ada
7735 cyclic pointer types.
7736 * lto-streamer-in.c: Include debug.h.
7737 (dref_queue): New global.
7738 (lto_read_tree_1): Stream in DIE references.
7739 (lto_input_tree): Register DIE references.
7740 (input_function): Stream DECL_DEBUG_ARGS.
7741 * lto-streamer-out.c: Include debug.h.
7742 (lto_write_tree_1): Output DIE references.
7743 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7744 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7745 (output_function): Stream DECL_DEBUG_ARGS.
7746 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7747 Stream DECL_ABSTRACT_ORIGIN.
7748 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7749 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7750 DECL_CONTEXT for file-scope decls.
7751 * lto-streamer.h (struct dref_entry): Declare.
7752 (dref_queue): Likewise.
7753 * cfgexpand.c (pass_expand::execute): Do not call the
7754 outlining_inline_function hook here.
7755 * lto-wrapper.c (debug_obj): New global.
7756 (tool_cleanup): Unlink it if required.
7757 (debug_objcopy): New function.
7758 (run_gcc): Handle early debug sections in the IL files by
7759 extracting them to separate files, partially linkin them and
7760 feeding the result back as result to the linker.
7761 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7762 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7763 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7764 sections into a separate segment.
7765 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7767 (darwin_asm_dwarf_section): Likewise.
7768 (darwin_asm_output_dwarf_offset): Likewise.
7769 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7771 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7772 Alan Hayward <alan.hayward@arm.com>
7773 David Sherwood <david.sherwood@arm.com>
7775 * read-md.h (md_reader::record_potential_iterator_use): Replace
7776 pointer argument with an rtx and an index.
7777 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7778 (apply_mode_iterator): Likewise.
7779 (apply_code_iterator): Likewise.
7780 (apply_int_iterator): Likewise.
7781 (apply_subst_iterator): Likewise.
7782 (record_iterator_use): Likewise.
7783 (record_attribute_use): Likewise.
7784 (md_reader::record_potential_iterator_use): Likewise. Update calls
7785 to record_iterator_use and apply_iterator.
7786 (iterator_use): Replace ptr with x and index.
7787 (attribute_use): Likewise.
7788 (apply_attribute_uses): Update calls to apply_iterator.
7789 (apply_iterators): Likewise. Update initialization of iterator_use.
7790 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7791 and record_potential_iterator_use.
7792 (rtx_reader::read_rtx_operand): Likewise.
7794 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7795 Alan Hayward <alan.hayward@arm.com>
7796 David Sherwood <david.sherwood@arm.com>
7798 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7801 2017-08-21 Richard Biener <rguenther@suse.de>
7804 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7805 at struct end conservatively when comparing common bases.
7807 2017-08-21 Richard Biener <rguenther@suse.de>
7809 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7810 (mem_ref_in_stmt): Remove.
7811 (determine_max_movement): Use ref index to get at the reference.
7812 (invariantness_dom_walker::before_dom_children): Deal with
7813 lim data already initialized.
7814 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7816 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7818 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7819 (TARGET_ISA_ROUND): Ditto.
7820 (TARGET_ROUND): Ditto.
7821 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7822 * config/i386/i386.md: Ditto.
7823 * config/i386/sse.md: Ditto.
7824 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7825 with OPTION_MASK_ISA_SSE4_1.
7827 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7830 * doc/extend.texi (x86 Built-in Functions): Correct the name of
7831 __builtin_ia32_lzcnt_u16.
7833 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7836 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
7837 (rs6000_set_current_function): Rewrite function to use it.
7839 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
7842 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
7843 and TYPE_WARN_IF_NOT_ALIGN.
7844 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
7845 (handle_warn_if_not_align): New.
7846 (place_union_field): Call handle_warn_if_not_align.
7847 (place_field): Call handle_warn_if_not_align.
7848 Copy TYPE_WARN_IF_NOT_ALIGN.
7849 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
7850 (layout_type): Likewise.
7851 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
7853 (tree_decl_common): Add warn_if_not_align.
7854 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
7855 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
7856 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
7857 (DECL_WARN_IF_NOT_ALIGN): Likewise.
7858 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
7859 * doc/extend.texi: Document warn_if_not_aligned attribute.
7860 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
7862 2017-08-17 Martin Liska <mliska@suse.cz>
7865 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
7866 (get_data_dependence): Use it as pointer type.
7867 (distribute_loop): Likewise.
7869 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7871 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
7872 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
7873 (p8_vmrgow_<mode>_direct): New define_insn.
7874 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
7875 handle endianness for vmrgew and vmrgow permute patterns.
7877 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7879 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
7880 * config/rs6000/rs6000-cpus.def: Remove comment.
7881 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
7882 (POWERPC_MASKS): Likewise.
7883 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
7884 use of TARGET_VSX_TIMODE.
7885 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
7886 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
7887 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
7888 (rs6000_option_override_internal): Remove dead code.
7889 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
7890 (rs6000_legitimize_reload_address): Likewise.
7891 (rs6000_legitimate_address_p): Likewise.
7892 (rs6000_opt_masks): Delete "vsx-timode".
7893 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
7894 from function comment.
7895 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
7896 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
7897 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
7899 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
7900 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
7901 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
7904 (define_peephole2): Likewise.
7906 2017-08-17 Martin Sebor <msebor@redhat.com>
7909 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
7910 past the end of an array.
7911 (test_pp_format): Add test cases.
7913 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
7915 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
7916 missing ECF_NOTHROW flags.
7918 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
7921 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
7922 operands residing in integer registers.
7923 (*vsx_le_perm_load_<mode>): Likewise.
7924 (*vsx_le_perm_store_<mode>): Likewise.
7925 (define_peephole2): Add peepholes to optimize the above.
7927 2017-08-17 Marek Polacek <polacek@redhat.com>
7930 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
7931 to mimic what shorten_compare did. Change the return type to bool.
7932 (fold_cond_expr_with_comparison): Update call to
7933 operand_equal_for_comparison_p.
7934 (fold_ternary_loc): Likewise.
7936 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
7938 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
7940 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
7942 2017-08-17 Richard Biener <rguenther@suse.de>
7944 * tree-ssa-structalias.c (solve_graph): When propagating
7945 to successors update the graphs succ edges and avoid duplicate work.
7947 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
7950 * config/i386/i386.c (ix86_option_override_internal): Save target
7951 specific options after ix86_stack_protector_guard_reg was changed.
7953 2017-08-17 Richard Biener <rguenther@suse.de>
7955 PR tree-optimization/81827
7956 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
7957 (new_var_info): Initialize it conservatively.
7958 (get_call_vi): Mark register vars.
7959 (new_scalar_tmp_constraint_exp): Likewise.
7960 (handle_rhs_call): Likewise.
7961 (handle_const_call): Likewise.
7962 (create_function_info_for): Likewise.
7963 (solve_constraints): Sort varinfos to separate register from
7964 non-register vars to pack points-to solution bitmaps during
7967 2017-08-17 Marek Polacek <polacek@redhat.com>
7969 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
7971 2017-08-17 Richard Biener <rguenther@suse.de>
7973 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
7974 to true when overflow is undefined and we saturated the result.
7976 2017-08-17 Alan Modra <amodra@gmail.com>
7979 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
7980 Don't use store multiple if only one reg needs saving.
7981 (interesting_frame_related_regno): New function.
7982 (rs6000_frame_related): Don't emit frame info for regs that
7984 (rs6000_emit_epilogue): Likewise.
7986 2017-08-16 Nathan Sidwell <nathan@acm.org>
7988 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
7989 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
7990 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
7991 * tree.c (free_lang_data_in_type): Use else-if chain. Always
7992 clear TYPE_LANG_1. Remove obsolete member-function stripping.
7993 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
7994 (verify_type): Adjust for TYPE_BINFO move.
7995 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
7996 process TYPE_BINFO directly.
7997 (hash_tree): Likewise.
7998 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8000 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8003 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8005 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8007 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8010 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8011 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8012 Add ix86_binary_operator_ok to insn constraint.
8013 (*iordi_1_bts): Ditto.
8014 (*xordi_1_btc): Ditto.
8015 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8016 Update corresponding peephole2 pattern.
8020 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8022 PR tree-optimization/81832
8023 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8024 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8026 2017-08-16 Marek Polacek <polacek@redhat.com>
8029 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8030 regarding size_zero_node.
8032 2017-08-16 Martin Liska <mliska@suse.cz>
8035 * config.gcc: Respect previously set extra_objs in case
8038 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8040 PR tree-optimization/81835
8041 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8042 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8043 not depend on the phi.
8045 2017-08-16 Alan Modra <amodra@gmail.com>
8047 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8050 2017-08-16 Alan Modra <amodra@gmail.com>
8052 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8053 (save_reg_p): ..into this. Update all callers.
8054 (first_reg_to_save): Simplify.
8056 2017-08-16 Alan Modra <amodra@gmail.com>
8058 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8061 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8065 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8066 constant count if that count is less than 32.
8068 2017-08-15 Nathan Sidwell <nathan@acm.org>
8070 * gcc.c (execute): Emit friendlier message if inferior is killed
8071 by an external cause.
8073 2017-08-15 Richard Biener <rguenther@suse.de>
8075 PR tree-optimization/81790
8076 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8077 CONSTRUCTORs from simplifying and VN.
8079 2017-08-14 Martin Sebor <msebor@redhat.com>
8081 * builtin-attrs.def: Add comments.
8083 2017-08-14 Martin Sebor <msebor@redhat.com>
8086 * doc/extend.texi (attribute nonstring): Document new attribute.
8088 2017-08-14 Martin Sebor <msebor@redhat.com>
8091 * tree-diagnostic.c (default_tree_printer): Handle %G.
8092 * gimple-pretty-print.h (percent_G_format): Declare new function.
8093 * gimple-pretty-print.c (percent_G_format): Define.
8094 * tree-pretty-print.c (percent_K_format): Add argument.
8096 2017-08-14 Martin Sebor <msebor@redhat.com>
8098 PR translation/79998
8099 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8100 Remove a stray space.
8102 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8105 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8106 (*iordi_1_bts): Ditto.
8107 (*xordi_1_btc): Ditto.
8109 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8112 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8113 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8115 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8116 quoted strings, and make more translator-friendly.
8117 (darwin_rs6000_override_options): Likewise.
8118 (rs6000_option_override_internal): Likewise.
8119 (rs6000_return_in_memory): Fix overlong line.
8120 (init_cmulative_args): Use quoted strings, and make more
8121 translator-friendly.
8122 (rs6000_pass_by_reference): Fix overlong line.
8123 (def_builtin): Use quoted strings.
8124 (altivec_expand_predicate_builtin): Use quoted strings, and make
8125 more translator-friendly.
8126 (htm_expand_builtin): Use quoted strings.
8127 (cpu_expand_builtin): Use quoted strings, and make more
8128 translator-friendly.
8129 (altivec_expand_builtin): Likewise.
8130 (paired_expand_predicate_builtin): Likewise.
8131 (rs6000_invalid_builtin): Likewise.
8132 (builtin_function_type): Use quoted strings.
8133 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8134 more translator-friendly.
8135 (rs6000_trampoline_init): Likewise.
8136 (rs6000_handle_altivec_attribute): Likewise.
8137 (rs6000_inner_target_options): Use quoted strings.
8138 (rs6000_disable_incompatible_switches): Likewise.
8139 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8140 strings, and make more translator-friendly.
8141 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8143 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8145 PR tree-optimization/81799
8146 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8147 cond_expr to simple gimple operand.
8149 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8152 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8154 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8158 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8159 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8160 TARGET_GASISR_PROLOGUES.
8161 * config/avr/avr.c (avr_option_override): Same.
8162 (avr_pass_pre_proep::execute): Same.
8164 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8167 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8168 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8170 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8172 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8173 define_insn_and_split. Split to a memory load from 0 in
8174 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8175 using PTR mode iterator.
8176 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8177 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8178 (*add_tp_<mode>): Redefine as define_insn_and_split.
8179 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8180 address space. Merge with *add_tp_x32 using PTR mode iterator.
8181 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8182 Split to an add with a memory load from 0 in
8183 DEFAULT_TLS_SEG_REG address space.
8185 2017-08-12 Andrew Pinski <apinski@cavium.com>
8187 * config/aarch64/aarch64-option-extensions.def (rdma):
8188 Fix feature string to what Linux prints out in /proc/cpuinfo.
8190 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8193 * dwarf2out.c (modified_type_die): For C typedef types that have
8194 an ultimate origin, process the ultimate origin instead of the
8196 (gen_typedef_die): Assert that input DECLs have no ultimate
8198 (gen_type_die_with_usage): For typedef variants that have an
8199 ultimate origin, just call gen_decl_die on the original DECL.
8200 (process_scope_var): Avoid creating DIEs for local typedefs and
8201 concrete static variables.
8203 2017-08-12 Alan Modra <amodra@gmail.com>
8207 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8208 match gnu-user.h startfile.
8209 (ENDFILE_LINUX_SPEC): Similarly.
8211 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8214 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8216 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8218 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8219 * config/aarch64/aarch64.md (mov<mode>): Change.
8220 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8221 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8222 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8224 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8226 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8227 for storage order barriers.
8229 2017-08-11 Martin Liska <mliska@suse.cz>
8232 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8233 variables of void type.
8235 2017-08-11 Martin Liska <mliska@suse.cz>
8237 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8238 TARGET_SUPPORTS_ALIASES.
8239 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8240 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8241 (optimize_weakref): Likewise.
8242 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8243 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8244 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8246 2017-08-11 Martin Liska <mliska@suse.cz>
8249 * config/i386/i386.c (make_resolver_func): Do complete
8250 refactoring of the function.
8252 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8255 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8256 * config/i386/i386.c (ix86_stack_protect_guard): Use
8257 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8258 * doc/invoke.texi (x86 Options): Document
8259 -mstack-protector-guard-symbol= option.
8261 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8263 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8264 * config/i386/i386.c (ix86_split_stack_guard): New function.
8265 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8266 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8267 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8268 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8269 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8270 (split_stack_space_check): Call ix86_split_stack_guard.
8272 2017-08-10 Martin Sebor <msebor@redhat.com>
8274 * print-tree.c (print_node): Print location using the established
8276 Replace spaces with colons.
8277 (debug_raw, debug): Ditto.
8279 2017-08-10 Martin Sebor <msebor@redhat.com>
8282 * pretty-print.c (pp_format): Correct the handling of %s precision.
8284 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8287 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8289 (ix86_finalize_stack_frame_flags): This. Also clear
8290 frame_pointer_needed if -fno-omit-frame-pointer is used without
8292 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8293 with ix86_finalize_stack_frame_flags.
8294 (ix86_expand_epilogue): Likewise.
8295 (ix86_expand_split_stack_prologue): Likewise.
8296 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8298 2017-08-10 Martin Liska <mliska@suse.cz>
8301 * c-attribs.c (handle_target_attribute):
8302 Report warning for an empty string argument of target attribute.
8304 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8307 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8309 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8310 or DECL_NONLOCAL labels.
8311 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8312 or DECL_NONLOCAL labels here.
8314 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8316 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8317 to indicate when early gimple folding has been disabled.
8318 (rs6000_gimple_fold_builtin): Add debug content.
8319 (rs6000_invalid_builtin): Fix whitespace.
8320 (rs6000_expand_builtin): Fix whitespace.
8321 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8323 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8326 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8327 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8329 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8331 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8332 (qdf24xx): Likewise.
8333 * config/aarch64/aarch64.md: Include falkor.md.
8334 * config/aarch64/falkor.md: New.
8336 2017-08-09 Marek Polacek <polacek@redhat.com>
8339 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8340 * diagnostic.c (emit_diagnostic): Add a comment.
8341 (emit_diagnostic_valist): New function.
8343 2017-08-09 Marek Polacek <polacek@redhat.com>
8346 * input.c (make_location): New overload.
8347 * input.h (make_location): Declare.
8349 2017-08-08 Alan Modra <amodra@gmail.com>
8350 H.J. Lu <hongjiu.lu@intel.com>
8353 * gcc.c (NO_PIE_SPEC): Delete.
8354 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8356 (LINK_PIE_SPEC): ..to here.
8357 (LINK_COMMAND_SPEC): Support -no-pie.
8358 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8359 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8361 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8362 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8363 (ENDFILE_CRTEND_SPEC): Similarly.
8365 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8368 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8369 (mstack-protector-guard-offset=): Ditto.
8370 * config/i386/i386.c (ix86_option_override): Handle
8371 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8373 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8374 ix86_stack_protect_guard_offset variables.
8375 (TARGET_STACK_PROTECT_GUARD): Always define.
8376 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8377 and -mstack-protector-guard-offset= options.
8379 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8381 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8382 boundary case for the last candidate.
8384 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8386 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8387 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8389 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8392 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8393 * config/aarch64/aarch64-builtins.c
8394 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8395 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8396 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8398 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8399 Andrew Pinski <pinskia@gmail.com>
8402 * internal-fn.def (XORSIGN): New.
8403 * optabs.def (xorsign_optab): New.
8404 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8405 (convert_expand_mult_copysign): New.
8406 (pass_optimize_widening_mul::execute): Call
8407 convert_expand_mult_copysign.
8409 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8411 PR tree-optimization/81354
8412 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8413 Insert on edges rather than explicitly creating landing pads.
8414 (analyze_candidates_and_replace): Commit edge inserts.
8416 2017-08-08 Richard Biener <rguenther@suse.de>
8419 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8420 (expand_simple_operations): Also look through ADDR_EXPRs with
8421 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8423 2017-08-08 Richard Biener <rguenther@suse.de>
8425 PR tree-optimization/81723
8426 * tree-vect-slp.c (struct bst_traits): New hash traits.
8427 (bst_fail): New global.
8428 (vect_build_slp_tree_2): New worker, split out from ...
8429 (vect_build_slp_tree): ... this now wrapping it with using
8430 bst_fail set to cache SLP tree build fails. Properly handle
8432 (vect_analyze_slp_instance): Allocate and free bst_fail.
8434 2017-08-08 Martin Liska <mliska@suse.cz>
8437 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8438 LABEL_DECLs that can be from a different function.
8440 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8442 PR tree-optimization/81744
8443 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8444 loop's number of iterations.
8446 2017-08-08 Martin Liska <mliska@suse.cz>
8448 * asan.c: Include header files.
8449 * attribs.c (build_decl_attribute_variant): New function moved
8451 (build_type_attribute_qual_variant): Likewise.
8452 (cmp_attrib_identifiers): Likewise.
8453 (simple_cst_list_equal): Likewise.
8454 (omp_declare_simd_clauses_equal): Likewise.
8455 (attribute_value_equal): Likewise.
8456 (comp_type_attributes): Likewise.
8457 (build_type_attribute_variant): Likewise.
8458 (lookup_ident_attribute): Likewise.
8459 (remove_attribute): Likewise.
8460 (merge_attributes): Likewise.
8461 (merge_type_attributes): Likewise.
8462 (merge_decl_attributes): Likewise.
8463 (merge_dllimport_decl_attributes): Likewise.
8464 (handle_dll_attribute): Likewise.
8465 (attribute_list_equal): Likewise.
8466 (attribute_list_contained): Likewise.
8467 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8468 (lookup_attribute_by_prefix): Likewise.
8469 * bb-reorder.c: Include header files.
8470 * builtins.c: Likewise.
8471 * calls.c: Likewise.
8472 * cfgexpand.c: Likewise.
8473 * cgraph.c: Likewise.
8474 * cgraphunit.c: Likewise.
8475 * convert.c: Likewise.
8476 * dwarf2out.c: Likewise.
8477 * final.c: Likewise.
8478 * fold-const.c: Likewise.
8479 * function.c: Likewise.
8480 * gimple-expr.c: Likewise.
8481 * gimple-fold.c: Likewise.
8482 * gimple-pretty-print.c: Likewise.
8483 * gimple.c: Likewise.
8484 * gimplify.c: Likewise.
8485 * hsa-common.c: Likewise.
8486 * hsa-gen.c: Likewise.
8487 * internal-fn.c: Likewise.
8488 * ipa-chkp.c: Likewise.
8489 * ipa-cp.c: Likewise.
8490 * ipa-devirt.c: Likewise.
8491 * ipa-fnsummary.c: Likewise.
8492 * ipa-inline.c: Likewise.
8493 * ipa-visibility.c: Likewise.
8495 * lto-cgraph.c: Likewise.
8496 * omp-expand.c: Likewise.
8497 * omp-general.c: Likewise.
8498 * omp-low.c: Likewise.
8499 * omp-offload.c: Likewise.
8500 * omp-simd-clone.c: Likewise.
8501 * opts-global.c: Likewise.
8502 * passes.c: Likewise.
8503 * predict.c: Likewise.
8504 * sancov.c: Likewise.
8505 * sanopt.c: Likewise.
8506 * symtab.c: Likewise.
8507 * toplev.c: Likewise.
8508 * trans-mem.c: Likewise.
8509 * tree-chkp.c: Likewise.
8510 * tree-eh.c: Likewise.
8511 * tree-into-ssa.c: Likewise.
8512 * tree-object-size.c: Likewise.
8513 * tree-parloops.c: Likewise.
8514 * tree-profile.c: Likewise.
8515 * tree-ssa-ccp.c: Likewise.
8516 * tree-ssa-live.c: Likewise.
8517 * tree-ssa-loop.c: Likewise.
8518 * tree-ssa-sccvn.c: Likewise.
8519 * tree-ssa-structalias.c: Likewise.
8520 * tree-ssa.c: Likewise.
8521 * tree-streamer-in.c: Likewise.
8522 * tree-vectorizer.c: Likewise.
8523 * tree-vrp.c: Likewise.
8525 * ubsan.c: Likewise.
8526 * varasm.c: Likewise.
8527 * varpool.c: Likewise.
8528 * tree.c: Remove functions moved to attribs.[ch].
8530 * config/aarch64/aarch64.c: Add attrs.h header file.
8531 * config/alpha/alpha.c: Likewise.
8532 * config/arc/arc.c: Likewise.
8533 * config/arm/arm.c: Likewise.
8534 * config/avr/avr.c: Likewise.
8535 * config/bfin/bfin.c: Likewise.
8536 * config/c6x/c6x.c: Likewise.
8537 * config/cr16/cr16.c: Likewise.
8538 * config/cris/cris.c: Likewise.
8539 * config/darwin.c: Likewise.
8540 * config/epiphany/epiphany.c: Likewise.
8541 * config/fr30/fr30.c: Likewise.
8542 * config/frv/frv.c: Likewise.
8543 * config/ft32/ft32.c: Likewise.
8544 * config/h8300/h8300.c: Likewise.
8545 * config/i386/winnt.c: Likewise.
8546 * config/ia64/ia64.c: Likewise.
8547 * config/iq2000/iq2000.c: Likewise.
8548 * config/lm32/lm32.c: Likewise.
8549 * config/m32c/m32c.c: Likewise.
8550 * config/m32r/m32r.c: Likewise.
8551 * config/m68k/m68k.c: Likewise.
8552 * config/mcore/mcore.c: Likewise.
8553 * config/microblaze/microblaze.c: Likewise.
8554 * config/mips/mips.c: Likewise.
8555 * config/mmix/mmix.c: Likewise.
8556 * config/mn10300/mn10300.c: Likewise.
8557 * config/moxie/moxie.c: Likewise.
8558 * config/msp430/msp430.c: Likewise.
8559 * config/nds32/nds32-isr.c: Likewise.
8560 * config/nds32/nds32.c: Likewise.
8561 * config/nios2/nios2.c: Likewise.
8562 * config/nvptx/nvptx.c: Likewise.
8563 * config/pa/pa.c: Likewise.
8564 * config/pdp11/pdp11.c: Likewise.
8565 * config/powerpcspe/powerpcspe.c: Likewise.
8566 * config/riscv/riscv.c: Likewise.
8567 * config/rl78/rl78.c: Likewise.
8568 * config/rx/rx.c: Likewise.
8569 * config/s390/s390.c: Likewise.
8570 * config/sh/sh.c: Likewise.
8571 * config/sol2.c: Likewise.
8572 * config/sparc/sparc.c: Likewise.
8573 * config/spu/spu.c: Likewise.
8574 * config/stormy16/stormy16.c: Likewise.
8575 * config/tilegx/tilegx.c: Likewise.
8576 * config/tilepro/tilepro.c: Likewise.
8577 * config/v850/v850.c: Likewise.
8578 * config/vax/vax.c: Likewise.
8579 * config/visium/visium.c: Likewise.
8580 * config/xtensa/xtensa.c: Likewise.
8582 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8585 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8586 constraints since the -mupper-regs-* switches have been
8588 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8589 into a vector from a double word element that was extracted from
8590 another vector, and eliminate extra XXPERMDI instructions.
8591 (vsx_concat_<mode>_2): Likewise.
8592 (vsx_concat_<mode>_3): Likewise.
8593 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8594 concat to allow optimizing inserts from previous extracts.
8596 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8598 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8599 memory reference to a SSP offset in TLS address space.
8600 (ix86_print_operand) <case '@'>: Remove.
8601 (ix86_print_operand_punct_valid_p): Remove '@' code.
8602 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8604 (stack_tls_protect_set_<mode>): Remove.
8605 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8606 (stack_tls_protect_test_<mode>): Remove.
8607 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8609 2017-08-07 Olivier Hainque <hainque@adacore.com>
8612 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8614 2017-08-07 Douglas Rupp <rupp@adacore.com>
8616 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8617 variable was referenced as multidir in command.
8619 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8622 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8625 2017-08-07 Martin Liska <mliska@suse.cz>
8627 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8628 * config/rl78/rl78.c: Add include of attribs.h.
8629 * config/sh/sh.c: Likewise.
8630 * config/v850/v850.c: Likewise.
8632 2017-08-07 Tom de Vries <tom@codesourcery.com>
8635 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8637 2017-08-07 Martin Liska <mliska@suse.cz>
8639 * config/mips/mips.c: Include attribs.h.
8641 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8644 * doc/invoke.texi: Document change in behvaior for -Ofast for
8647 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8649 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8651 (aarch64_pop_regs): Likewise.
8652 (aarch64_gen_load_pair): Likewise.
8653 (aarch64_save_callee_saves): Likewise.
8654 (aarch64_restore_callee_saves): Likewise.
8656 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8658 * config/i386/i386.c: Revert the last change.
8660 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8663 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8665 (ix86_finalize_stack_frame_flags): This. Also clear
8666 frame_pointer_needed if -fno-omit-frame-pointer is used without
8668 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8669 with ix86_finalize_stack_frame_flags.
8670 (ix86_expand_epilogue): Likewise.
8671 (ix86_expand_split_stack_prologue): Likewise.
8673 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8676 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8677 to P_AES for Westmere.
8679 2017-08-07 Jonathan Yong <10walls@gmail.com>
8681 * config/i386/mingw.opt (fset-stack-executable): Removed.
8682 * config/i386/cygming.opt (fset-stack-executable): Moved
8684 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8686 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8688 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8690 2017-08-07 Marek Polacek <polacek@redhat.com>
8693 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8695 2017-08-07 Martin Liska <mliska@suse.cz>
8697 * attribs.h (canonicalize_attr_name): New function.
8698 (cmp_attribs): Move from c-format.c and adjusted.
8699 (is_attribute_p): Moved from tree.h.
8700 * tree-inline.c: Add new includes.
8701 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8702 (private_is_attribute_p): Remove.
8703 (private_lookup_attribute): Likewise.
8704 (private_lookup_attribute_by_prefix): Simplify.
8705 (remove_attribute): Use is_attribute_p.
8706 * tree.h: Remove removed declarations.
8708 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8711 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8712 instead of computing it in the function. Formatting fix.
8713 (expand_case): Don't rely on default_edge being the first edge,
8714 clear it if removing it, pass default_edge to
8715 emit_case_dispatch_table.
8716 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8719 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8721 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8722 insn in the function, emit NOP after the insn.
8724 2017-08-06 Tom de Vries <tom@codesourcery.com>
8726 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8729 2017-08-06 Tom de Vries <tom@codesourcery.com>
8731 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8734 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8736 PR tree-optimization/57371
8737 * match.pd: New pattern.
8739 2017-08-04 Marek Polacek <polacek@redhat.com>
8742 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8743 perform the computation in offset_int.
8745 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8747 PR tree-optimization/81136
8748 * tree-vectorizer.h: Include tree-hash-traits.h.
8749 (vec_base_alignments): New typedef.
8750 (vec_info): Add a base_alignments field.
8751 (vect_record_base_alignments): Declare.
8752 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8754 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8755 (create_data_ref): Add an is_conditional_in_stmt argument.
8756 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8757 the is_conditional_in_stmt field.
8758 (data_ref_loc): Add an is_conditional_in_stmt field.
8759 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8760 (find_data_references_in_stmt): Update call to create_data_ref.
8761 (graphite_find_data_references_in_stmt): Likewise.
8762 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8763 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8764 (vect_record_base_alignment): New function.
8765 (vect_record_base_alignments): Likewise.
8766 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8767 for nested statements even if we fail to compute a misalignment.
8768 Use pooled base alignments for unconditional references.
8769 (vect_find_same_alignment_drs): Compare base addresses instead
8771 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8772 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8774 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8776 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8777 Add an explicit name for the enum. Use auto_vec for slp_instances
8779 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8781 (_bb_vec_info): Add a constructor and destructor.
8782 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8783 (destroy_loop_vec_info): Delete.
8784 (vect_destroy_datarefs): Likewise.
8785 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8786 (vec_info::vec_info): New function.
8787 (vec_info::~vec_info): Likewise.
8788 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8789 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8790 destroy_loop_vec_info.
8791 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8792 (_loop_vec_info::_loop_vec_info): ...this.
8793 (destroy_loop_vec_info): Replace with...
8794 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8795 the stmt_vec_infos. Leave handling of vec_info information to its
8796 destructor. Remove explicit vector releases.
8797 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8798 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8799 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8800 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8801 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8802 (destroy_bb_vec_info): Replace with...
8803 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8804 information to its destructor.
8805 (vect_slp_analyze_bb_1): Use new and delete instead of
8806 new_bb_vec_info and destroy_bb_vec_info.
8807 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8810 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8812 * tree-data-ref.h (subscript): Add access_fn field.
8813 (data_dependence_relation): Add could_be_independent_p.
8814 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8815 (same_access_functions): Move to tree-data-ref.c.
8816 * tree-data-ref.c (ref_contains_union_access_p): New function.
8817 (access_fn_component_p): Likewise.
8818 (access_fn_components_comparable_p): Likewise.
8819 (dr_analyze_indices): Add a reference to access_fn_component_p.
8820 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8822 (constant_access_functions): Likewise.
8823 (add_other_self_distances): Likewise.
8824 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8825 (initialize_data_dependence_relation): Use XCNEW and remove
8826 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8827 of access functions that have the same type. Allow the
8828 subsequence to end with different bases in some circumstances.
8829 Record the chosen access functions in SUB_ACCESS_FN.
8830 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
8831 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
8832 (subscript_dependence_tester_1): Likewise dra and drb.
8833 (build_classic_dist_vector): Update calls accordingly.
8834 (subscript_dependence_tester): Likewise.
8835 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
8836 DDR_COULD_BE_INDEPENDENT_P.
8837 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
8838 comp_alias_ddrs instead of may_alias_ddrs.
8839 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8841 (vect_analyze_data_ref_dependence): Use it if
8842 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
8843 distance vectors if that fails.
8844 (dependence_distance_ge_vf): New function.
8845 (vect_prune_runtime_alias_test_list): Use it. Don't clear
8846 LOOP_VINFO_MAY_ALIAS_DDRS.
8848 2017-08-04 Richard Biener <rguenther@suse.de>
8851 * fold-const.c (fold_binary_loc): Properly restrict
8852 minus_var0 && minus_var1 case when associating undefined overflow
8855 2017-08-04 Tom de Vries <tom@codesourcery.com>
8857 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
8859 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8862 Don't start diagnostic messages with a capital letter.
8863 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8865 (rs6000_invalid_builtin): Likewise.
8866 (rs6000_trampoline_init): Likewise.
8868 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8871 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
8872 after setting changeable df flags.
8874 2017-08-03 Richard Biener <rguenther@suse.de>
8876 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
8877 up if the use is in USE - X.
8879 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
8881 * toplev.c (dumpfile.h): New include.
8882 (internal_error_reentered): New static function. Use it...
8883 (internal_error_function): ...here to handle reentered internal_error.
8885 2017-08-03 Richard Biener <rguenther@suse.de>
8888 * fold-const.c (split_tree): Add minus_var and minus_con
8889 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
8890 here but always use minus_*.
8891 (associate_trees): Assert we never associate with MINUS_EXPR
8892 and NULL first operand. Do not recurse for PLUS_EXPR operands
8893 when associating as MINUS_EXPR either.
8894 (fold_binary_loc): Track minus_var and minus_con.
8896 2017-08-03 Tom de Vries <tom@codesourcery.com>
8899 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
8900 ACCEL_COMPILER, apply finish_options on
8901 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
8903 2017-08-03 Tom de Vries <tom@codesourcery.com>
8906 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
8907 function_entry_patch_area_size > 0.
8909 2017-08-03 Jakub Jelinek <jakub@redhat.com>
8912 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
8913 instead of 10??LU, perform unit multiplication in wide_int,
8914 don't change alloc_object_size_limit if the limit is larger
8917 PR tree-optimization/81655
8918 PR tree-optimization/81588
8919 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
8920 the case when ranges[i].low and high are 1 for unsigned type with
8924 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
8925 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
8927 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8929 * tree-vrp.h: Add include guard.
8931 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
8934 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
8935 (ud2): New insn pattern.
8936 * config/i386/i386.c (ix86_expand_epilogue):
8937 For naked functions, generate ud2 instead of trap insn.
8939 2017-08-02 Marek Polacek <polacek@redhat.com>
8942 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
8944 2017-08-02 Tom de Vries <tom@codesourcery.com>
8945 Cesar Philippidis <cesar@codesourcery.com>
8947 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
8948 Add missing edge probabilities.
8950 2017-08-02 Tamar Christina <tamar.christina@arm.com>
8952 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
8955 2017-08-02 Jakub Jelinek <jakub@redhat.com>
8958 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
8959 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
8960 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
8961 prologue_seq sequences - if any.
8963 2017-08-02 Richard Biener <rguenther@suse.de>
8965 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
8966 via vectors if supported, integer extracts via punning if supported
8967 or otherwise vector extracts.
8969 2017-08-02 Richard Biener <rguenther@suse.de>
8971 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
8973 (bitmap_insert_into_set): ... this.
8975 2017-08-02 Richard Biener <rguenther@suse.de>
8977 PR tree-optimization/81633
8979 2015-08-17 Alan Hayward <alan.hayward@arm.com>
8981 PR tree-optimization/71752
8982 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
8984 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
8986 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
8987 (machine_function::call_ms2sysv_pad_out): Remove field.
8988 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
8989 (ix86_compute_frame_layout): Likewise.
8991 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
8994 * config/i386/i386.c (ix86_set_func_type): Disallow naked
8995 attribute with interrupt attribute.
8997 2017-08-01 Andrew Pinski <apinski@cavium.com>
8999 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9000 BIT_INSERT_EXPR's operand 1
9001 to see if the types precision matches.
9003 2017-08-01 Martin Liska <mliska@suse.cz>
9006 * builtins.c (expand_builtin_memcpy_args): Remove.
9007 (expand_builtin_memcpy): Call newly added function
9008 expand_builtin_memory_copy_args.
9009 (expand_builtin_memcpy_with_bounds): Likewise.
9010 (expand_builtin_mempcpy): Remove last argument.
9011 (expand_builtin_mempcpy_with_bounds): Likewise.
9012 (expand_builtin_memory_copy_args): New function created from
9013 expand_builtin_mempcpy_args with small modifications.
9014 (expand_builtin_mempcpy_args): Remove.
9015 (expand_builtin_stpcpy): Remove unused argument.
9016 (expand_builtin): Likewise.
9017 (expand_builtin_with_bounds): Likewise.
9019 2017-08-01 Martin Liska <mliska@suse.cz>
9022 Make mempcpy more optimal (PR middle-end/70140).
9024 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9027 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9028 __builtin_vec_cmpne verify both arguments are compatible vectors
9029 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9030 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9031 move computation of aligned to after checking the argument types.
9035 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9036 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9039 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9040 Doug Rupp <rupp@adacore.com>
9041 Olivier Hainque <hainque@adacore.com>
9043 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9044 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9046 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9047 for TARGET_OS_CPP_BUILTIN.
9048 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9049 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9051 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9052 passing required abi options to the assembler for EABI configurations.
9053 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9054 of .text.hot and .text.unlikely sections for kernel modules when
9055 using ARM style exceptions.
9056 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9057 options. Add EXTRA_CC1_SPEC.
9058 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9060 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9062 (ARM_TARGET2_DWARF_FORMAT): Define.
9063 * config/arm/t-vxworks: Adjust multilib control to removal of the
9064 Diab command line options.
9066 2017-08-01 Martin Liska <mliska@suse.cz>
9068 PR gcov-profile/81561
9069 * gcov.c (unblock): Make unblocking safe as we need to preserve
9070 index correspondence of blocks and block_lists.
9072 2017-08-01 Richard Biener <rguenther@suse.de>
9074 PR tree-optimization/81181
9075 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9076 (compute_antic): ... end of iteration here.
9078 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9080 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9081 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9082 (ftree-slp-vectorize): Likewise.
9083 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9084 can no longer be set independent of flag_tree_loop_vectorize.
9085 * omp-general.c (emp_max_vf): Likewise.
9086 * opts.c (enable_fdo_optimizations): Remove references to
9087 flag_tree_vectorize, these are now implicit.
9088 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9089 and leave it for the options machinery.
9091 2017-08-01 Martin Liska <mliska@suse.cz>
9094 * builtins.c (expand_builtin_memcpy_args): Remove.
9095 (expand_builtin_memcpy): Call newly added function
9096 expand_builtin_memory_copy_args.
9097 (expand_builtin_memcpy_with_bounds): Likewise.
9098 (expand_builtin_mempcpy): Remove last argument.
9099 (expand_builtin_mempcpy_with_bounds): Likewise.
9100 (expand_builtin_memory_copy_args): New function created from
9101 expand_builtin_mempcpy_args with small modifications.
9102 (expand_builtin_mempcpy_args): Remove.
9103 (expand_builtin_stpcpy): Remove unused argument.
9104 (expand_builtin): Likewise.
9105 (expand_builtin_with_bounds): Likewise.
9107 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9110 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9111 print "ds:" only for immediates in generic address space.
9113 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9116 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9117 (ix86_function_ok_for_sibcall): Return false for naked functions.
9119 2017-08-01 Richard Biener <rguenther@suse.de>
9121 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9122 (compute_antic): Seed worklist with exit block predecessors.
9123 * cfganal.c (dfs_find_deadend): For a cycle return the source
9124 of the edge closing it.
9126 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9128 * config/aarch64/aarch64.c
9129 (aarch64_can_const_movi_rtx_p): Move 0 check.
9131 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9133 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9134 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9136 * match.pd: Ditto in address comparison pattern.
9138 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9140 PR tree-optimization/81627
9141 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9142 closed ssa form for store-store chain.
9144 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9146 PR tree-optimization/81620
9147 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9148 for store-store chain.
9150 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9152 PR tree-optimization/81588
9153 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9154 ranges[i].in_p, invert comparison code ccode. For >/>=,
9155 swap rhs1 and rhs2 and comparison code unconditionally,
9156 for </<= don't do that. Don't swap rhs1/rhs2 again if
9157 ranges[i].in_p, instead invert comparison code ccode if
9158 opcode or oe->rank is BIT_IOR_EXPR.
9161 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9162 a direct optab to conversion optab.
9163 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9164 with GET_MODE_INNER as last argument instead of optab_handler.
9165 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9166 vector extraction if possible and optab is available.
9167 * expr.c (store_constructor): Use convert_optab_handler instead
9168 of optab_handler. Use vector initialization from smaller
9169 vectors if possible and optab is available.
9170 * tree-vect-stmts.c (vectorizable_load): Likewise.
9171 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9173 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9174 of vec_init from half-sized vectors with the same element mode.
9175 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9176 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9177 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9178 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9179 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9180 after mode in gen_vec_extract* calls.
9181 (vec_extract<mode>): Renamed to ...
9182 (vec_extract<mode><ssescalarmodelower>): ... this.
9183 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9184 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9185 element mode after mode in gen_vec_init* calls.
9186 (VEC_INIT_HALF_MODE): New mode iterator.
9187 (vec_init<mode>): Renamed to ...
9188 (vec_init<mode><ssescalarmodelower>): ... this.
9189 (vec_init<mode><ssehalfvecmodelower>): New expander.
9190 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9191 (vec_extractv2sfsf): ... this.
9192 (vec_initv2sf): Renamed to ...
9193 (vec_initv2sfsf): ... this.
9194 (vec_extractv2si): Renamed to ...
9195 (vec_extractv2sisi): ... this.
9196 (vec_initv2si): Renamed to ...
9197 (vec_initv2sisi): ... this.
9198 (vec_extractv4hi): Renamed to ...
9199 (vec_extractv4hihi): ... this.
9200 (vec_initv4hi): Renamed to ...
9201 (vec_initv4hihi): ... this.
9202 (vec_extractv8qi): Renamed to ...
9203 (vec_extractv8qiqi): ... this.
9204 (vec_initv8qi): Renamed to ...
9205 (vec_initv8qiqi): ... this.
9206 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9207 (vec_init<mode>): Renamed to ...
9208 (vec_init<mode><VEC_base_l>): ... this.
9209 (vec_extract<mode>): Renamed to ...
9210 (vec_extract<mode><VEC_base_l>): ... this.
9211 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9212 (vec_initv2sfsf): ... this.
9213 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9214 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9215 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9216 element mode after mode in gen_vec_init* calls.
9217 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9218 (vec_init<mode><Vel>): ... this.
9219 (vec_extract<mode>): Renamed to ...
9220 (vec_extract<mode><Vel>): ... this.
9221 * config/aarch64/iterators.md (Vel): New mode attribute.
9222 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9223 Add element mode after mode in gen_vec_extract* calls.
9224 * config/s390/vector.md (non_vec_l): New mode attribute.
9225 (vec_extract<mode>): Renamed to ...
9226 (vec_extract<mode><non_vec_l>): ... this.
9227 (vec_init<mode>): Renamed to ...
9228 (vec_init<mode><non_vec_l>): ... this.
9229 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9230 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9232 * config/arm/iterators.md (V_elem_l): New mode attribute.
9233 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9234 (vec_extract<mode><V_elem_l>): ... this.
9235 (vec_extractv2di): Renamed to ...
9236 (vec_extractv2didi): ... this.
9237 (vec_init<mode>): Renamed to ...
9238 (vec_init<mode><V_elem_l>): ... this.
9239 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9240 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9241 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9242 Add element mode after gen_vec_extract* calls.
9243 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9244 (vec_init<mode><unitmode>): ... this.
9245 (vec_extract<mode>): Renamed to ...
9246 (vec_extract<mode><unitmode>): ... this.
9247 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9248 (vec_init<mode><unitmode>): ... this.
9249 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9250 (vec_initv2sfsf): ... this.
9251 (vec_extractv2sf): Renamed to ...
9252 (vec_extractv2sfsf): ... this.
9253 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9254 Add element mode after gen_vec_extract* calls.
9255 * config/mips/mips.md (unitmode): New mode iterator.
9256 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9257 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9258 * config/spu/spu.md (inner_l): New mode attribute.
9259 (vec_init<mode>): Renamed to ...
9260 (vec_init<mode><inner_l>): ... this.
9261 (vec_extract<mode>): Renamed to ...
9262 (vec_extract<mode><inner_l>): ... this.
9263 * config/sparc/sparc.md (veltmode): New mode iterator.
9264 (vec_init<VMALL:mode>): Renamed to ...
9265 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9266 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9267 (vec_initv2sisi): ... this.
9268 (vec_initv2sf): Renamed to ...
9269 (vec_initv2sfsf): ... this.
9270 (vec_extractv2sf): Renamed to ...
9271 (vec_extractv2sfsf): ... this.
9272 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9273 (vec_init<mode>): Renamed to ...
9274 (vec_init<mode><VEC_base_l>): ... this.
9275 (vec_extract<mode>): Renamed to ...
9276 (vec_extract<mode><VEC_base_l>): ... this.
9277 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9278 (vec_initv2sfsf): ... this.
9279 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9280 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9281 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9282 gen_vec_init* calls.
9284 2017-08-01 Richard Biener <rguenther@suse.de>
9286 PR tree-optimization/81297
9287 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9288 TREE_OVERFLOW from INTEGER_CSTs.
9290 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9292 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9294 2017-07-31 Carl Love <cel@us.ibm.com>
9296 * config/rs6000/rs6000-c: Add support for built-in functions
9297 vector signed char vec_xl_be (signed long long, signed char *);
9298 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9299 vector signed int vec_xl_be (signed long long, signed int *);
9300 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9301 vector signed long long vec_xl_be (signed long long, signed long long *);
9302 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9303 vector signed short vec_xl_be (signed long long, signed short *);
9304 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9305 vector double vec_xl_be (signed long long, double *);
9306 vector float vec_xl_be (signed long long, float *);
9307 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9308 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9309 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9311 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9312 (altivec_expand_builtin): Add switch statement to call
9313 altivec_expand_xl_be for each builtin.
9314 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9315 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9316 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9317 __builtin_vsx_le_be_v16qi.
9318 * doc/extend.texi: Update the built-in documentation file for the
9319 new built-in functions.
9321 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9324 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9326 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9328 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9330 * config.gcc: Add z14.
9331 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9332 CPU model numbers for z13s and z14.
9333 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9335 * config/s390/s390-opts.h (enum processor_type): Rename
9336 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9337 * config/s390/s390.c (processor_table): Add field for CPU name to
9338 be passed to Binutils.
9339 (s390_asm_output_machine_for_arch): Use the new field in
9340 processor_table for Binutils.
9341 (s390_expand_builtin): Replace arch12 with z14.
9342 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9343 (s390_get_sched_attrmask): Likewise.
9344 (s390_get_unit_mask): Likewise.
9345 * config/s390/s390.opt: Add z14 to processor_type enum.
9347 2017-07-31 Martin Jambor <mjambor@suse.cz>
9350 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9351 regardless of optimization level.
9353 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9354 Martin Liska <mliska@suse.cz>
9356 * predict.def: Remove old comment and adjust probability.
9357 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9360 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9363 * config/i386/i386.c (ix86_function_naked): New function.
9364 (ix86_can_use_return_insn_p): Return false for naked functions.
9365 (ix86_expand_prologue): Skip prologue for naked functions.
9366 (ix86_expand_epilogue): Skip epilogue for naked functions
9367 and emit trap instruction.
9368 (ix86_warn_func_return): New function.
9369 (ix86_attribute_table): Add "naked" attribute specification.
9370 (TARGET_WARN_FUNC_RETURN): Define.
9371 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9373 2017-07-31 Martin Liska <mliska@suse.cz>
9375 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9376 (dump_gimple_bb_header): Always dump BB info.
9377 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9379 2017-07-31 Martin Liska <mliska@suse.cz>
9382 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9383 also with current_function_decl non-null equality.
9385 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9388 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9389 change type to the element type, instead add eltype variable and
9390 use it where we are interested in the element type.
9392 PR tree-optimization/81603
9393 * ipa-polymorphic-call.c
9394 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9395 offset arithmetic in offset_int, bail out if the resulting bit offset
9396 doesn't fit into shwi.
9398 2017-07-31 Martin Liska <mliska@suse.cz>
9400 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9401 (gimplify_save_expr): Fix comment.
9403 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9406 * config/i386/i386.c (ix86_function_arg): Update arguments for
9408 (ix86_compute_frame_layout): Set the initial stack offset to
9409 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9410 INCOMING_FRAME_SP_OFFSET.
9411 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9412 stack before exception handler returns.
9413 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9414 the 'ERROR_CODE' for exception handler.
9416 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9418 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9419 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9420 (ASM_OUTPUT_REG_POP): Ditto.
9421 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9422 instead of asm_fprintf to output pure string.
9424 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9426 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9427 to imported_module_or_decl hook.
9428 (debug_nothing_tree_tree_tree_bool): Remove.
9429 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9430 * debug.c (do_nothing_debug_hooks): Use
9431 debug_nothing_tree_tree_tree_bool_bool instead of
9432 debug_nothing_tree_tree_tree_bool.
9433 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9434 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9435 * sdbout.c (sdb_debug_hooks): Likewise.
9436 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9437 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9439 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9440 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9441 attribute, don't add anything.
9443 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9445 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9446 (fold_build2_stat_loc): Likewise.
9447 (fold_build3_stat_loc): Likewise.
9448 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9449 (fold_build1_loc): Remove macro.
9450 (fold_build2_loc): Likewise.
9451 (fold_build3_loc): Likewise.
9453 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9455 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9456 (gimple_build_debug_bind_source_stat): Likewise.
9457 * gimple.h (gimple_build_debug_bind): Remove macro.
9458 (gimple_build_debug_bind_source): Likewise.
9460 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9462 * bitmap.c (bitmap_alloc): Adjust.
9463 (bitmap_gc_alloc): Likewise.
9464 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9466 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9468 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9469 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9470 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9471 (bitmap_gc_alloc_stat): Likewise.
9472 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9474 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9476 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9477 * rtl.h (shallow_copy_rtx): Remove macro.
9479 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9481 * emit-rtl.c (gen_raw_REG): Adjust.
9482 * gengenrtl.c (gendef): Likewise.
9483 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9484 * rtl.h (rtx_alloc): Remove macro.
9486 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9488 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9489 (build_tree_list_stat): Likewise.
9490 * tree.h (build_tree_list): Remove macro.
9491 (build_tree_list_vec): Likewise.
9493 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9495 * tree.c (make_vector_stat): Remove _stat from name.
9496 (build_vector_stat): Likewise.
9497 * tree.h (make_vector_stat): Remove macro.
9498 (build_vector_stat): Likewise.
9500 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9502 * tree.h (build_var_debug_value): Remove prototype.
9504 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9506 * tree.c (tree_cons_stat): Remove _stat from name.
9507 * tree.h (tree_cons): Remove macro.
9509 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9511 * tree.c (build_vl_exp_stat): Remove _stat from name.
9512 * tree.h (build_vl_exp): Remove macro.
9514 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9516 * tree.c (build_decl_stat): Remove _stat from name.
9517 * tree.h (build_decl): Remove macro.
9519 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9521 * gimple.c (gimple_build_with_ops_stat): Adjust.
9522 (gimple_alloc_stat): Remove _stat from name.
9523 * gimple.h (gimple_alloc): Remove macro.
9525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9527 * tree.c (make_tree_vec_stat): Remove _stat from name.
9528 (grow_tree_vec_stat): Likewise.
9529 * tree.h (make_tree_vec_stat): Adjust prototype.
9530 (grow_tree_vec_stat): Likewise.
9531 (make_tree_vec): Remove macro.
9532 (grow_tree_vec): Likewise.
9534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9536 * fold-const.c (fold_build1_stat_loc): Adjust.
9537 (fold_build2_stat_loc): Likewise.
9538 (fold_build3_stat_loc): Likewise.
9539 * tree.c (build0_stat): Remove _stat from name.
9540 (build1_stat): Likewise.
9541 (build2_stat): Likewise.
9542 (build3_stat): Likewise.
9543 (build4_stat): Likewise.
9544 (build5_stat): Likewise.
9545 * tree.h (build1_loc): Remove macro, and rename _stat function
9547 (build2_loc): Likewise.
9548 (build3_loc): Likewise.
9549 (build4_loc): Likewise.
9550 (build5_loc): Likewise.
9552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9554 * tree.c (make_int_cst_stat): Remove _stat from name.
9555 * tree.h (make_int_cst_stat): Adjust prototype.
9556 (make_int_cst): Remove macro.
9558 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9560 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9561 * tree.h (make_tree_binfo_stat): Adjust prototype.
9562 (make_tree_binfo): Remove.
9564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9566 * tree.c (copy_node_stat): Rename to copy_node.
9567 (build_distinct_type_copy): Adjust.
9568 * tree.h (copy_node_stat): Adjust prototype.
9569 (copy_node): Remove macro.
9571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9573 * tree.c (make_node_stat): rename to make_node.
9574 (build_tree_list_stat): Adjust.
9575 (build0_stat): Likewise.
9576 (build2_stat): Likewise.
9577 (build3_stat): Likewise.
9578 (build4_stat): Likewise.
9579 (build5_stat): Likewise.
9580 (build_decl_stat): Likewise.
9581 * tree.h (make_node_stat): Adjust prototype.
9582 (make_node): remove macro.
9584 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9586 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9587 (PPC_FEATURE2_SCV): Likewise.
9588 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9590 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9592 * config/aarch64/aarch64.c
9593 (aarch64_internal_mov_immediate): Add new special pattern.
9594 * config/aarch64/aarch64.md (*movdi_aarch64):
9595 Add reg/32bit const mov case.
9597 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9598 Richard Sandiford <richard.sandiford@linaro.org>
9600 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9601 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9602 Add integer and movi cases.
9603 (movi-split-hf-df-sf split, fp16): New.
9604 (enabled): Added TARGET_FP_F16INST.
9605 * config/aarch64/iterators.md (GPF_HF): New.
9606 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9608 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9610 * config/aarch64/aarch64.c
9611 (aarch64_simd_container_mode): Add prototype.
9612 (aarch64_expand_mov_immediate): Add HI support.
9613 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9614 (aarch64_can_const_movi_rtx_p): New.
9615 (aarch64_preferred_reload_class):
9616 Remove restrictions of using FP registers for certain SIMD operations.
9617 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9618 (aarch64_valid_floating_const): Add integer move validation.
9619 (aarch64_simd_imm_scalar_p): Remove.
9620 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9621 (aarch64_legitimate_constant_p): Expand list of supported cases.
9622 * config/aarch64/aarch64-protos.h
9623 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9624 (aarch64_reinterpret_float_as_int): New.
9625 (aarch64_simd_imm_scalar_p): Remove.
9626 * config/aarch64/constraints.md (Uvi): New.
9627 (Dd): Split into Ds and new Dd.
9628 * config/aarch64/aarch64.md (*movsi_aarch64):
9630 (*movdi_aarch64): Add SIMD mov case.
9632 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9634 * tree-predcom.c: (struct chain): Handle store-store chain in which
9635 stores for elimination only store loop invariant values.
9636 (execute_pred_commoning_chain): Ditto.
9637 (prepare_initializers_chain_store_elim): Ditto.
9638 (prepare_finalizers): Ditto.
9639 (is_inv_store_elimination_chain): New function.
9640 (initialize_root_vars_store_elim_1): New function.
9642 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9644 * tree-predcom.c: Revise general description of the pass.
9645 (enum chain_type): New enum type for store elimination.
9646 (struct chain): New field supporting store elimination.
9647 (struct component): Ditto.
9648 (dump_chain): Dump store-stores chain.
9649 (release_chain): Release resources.
9650 (split_data_refs_to_components): Compute and create component
9651 contains only stores for elimination.
9652 (get_chain_last_ref_at): New function.
9653 (make_invariant_chain): Initialization.
9654 (make_rooted_chain): Specify chain type in parameter and record it.
9655 (add_looparound_copies): Skip for store-stores chain.
9656 (determine_roots_comp): Compute type of chain and pass it to
9658 (initialize_root_vars_store_elim_2): New function.
9659 (finalize_eliminated_stores): New function.
9660 (remove_stmt): Handle store for elimination.
9661 (execute_pred_commoning_chain): Execute predictive commoning on
9663 (determine_unroll_factor): Skip unroll for store-stores chain.
9664 (prepare_initializers_chain_store_elim): New function.
9665 (prepare_initializers_chain): Hanlde store-store chain.
9666 (prepare_finalizers_chain, prepare_finalizers): New function.
9667 (tree_predictive_commoning_loop): Return integer value indicating
9668 if loop is unrolled or lcssa form is corrupted.
9669 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9671 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9673 * tree-predcom.c (initialize_root): Delete.
9674 (execute_pred_commoning_chain): Initialize root vars and replace
9675 reference of non-combined chain directly, rather than call above
9678 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9680 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9681 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9683 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9685 * tree-predcom.c (struct chain): New field init_seq.
9686 (release_chain): Release init_seq.
9687 (prepare_initializers_chain): Record intialization stmts in above
9689 (insert_init_seqs): New function.
9690 (tree_predictive_commoning_loop): Call insert_init_seqs.
9692 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9694 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9696 2017-07-28 Richard Biener <rguenther@suse.de>
9698 * match.pd: Remove superfluous :c.
9699 * genmatch.c (simplify::id): Add member.
9700 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9702 (current_id): New global.
9703 (dt_node::parent): Move from ...
9704 (dt_operand::parent): ... here. Add for_id member.
9705 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9706 (decision_tree::find_node): Relax order requirement when
9707 merging DT_TRUE nodes to ones inbetween the current simplify
9708 and the one we try to merge with. Add diagnostic whenever
9709 we need to enforce pattern order by not merging.
9710 (decision_tree::insert): Set current_id.
9711 (decision_tree::print_node): Dump parent node and for_id.
9712 (parser::last_id): Add member.
9713 (parser::push_simplify): Assign unique id.
9714 (parser::parser): Initialize last_id.
9716 2017-07-28 Martin Liska <mliska@suse.cz>
9719 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9720 gimple_build_debug_bind.
9722 2017-07-28 Richard Biener <rguenther@suse.de>
9724 PR tree-optimization/81502
9725 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9727 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9729 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9730 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9731 for BIT_FIELD_REF args.
9732 * fold-const.c (make_bit_field_ref): Likewise.
9733 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9735 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9738 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9739 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9740 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9741 Or it into SANITIZER_UNDEFINED.
9742 * ubsan.c: Include gimple-fold.h and varasm.h.
9743 (ubsan_expand_ptr_ifn): New function.
9744 (instrument_pointer_overflow): New function.
9745 (maybe_instrument_pointer_overflow): New function.
9746 (instrument_object_size): Formatting fix.
9747 (pass_ubsan::execute): Call instrument_pointer_overflow
9748 and maybe_instrument_pointer_overflow.
9749 * internal-fn.c (expand_UBSAN_PTR): New function.
9750 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9751 * sanitizer.def (__ubsan_handle_pointer_overflow,
9752 __ubsan_handle_pointer_overflow_abort): New builtins.
9753 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9754 * internal-fn.def (UBSAN_PTR): New internal function.
9755 * opts.c (sanitizer_opts): Add pointer-overflow.
9756 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9757 * fold-const.c (build_range_check): Compute pointer range check in
9758 integral type if pointer arithmetics would be needed. Formatting
9761 2017-07-28 Martin Liska <mliska@suse.cz>
9764 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9765 parameters that are of a variable-length.
9767 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9769 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9771 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9772 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9773 (CRT_CALL_STATIC_FUNCTION): Likewise.
9774 (ASM_DEFAULT_SPEC): New define.
9775 (ASM_SPEC32): Likewise.
9776 (ASM_SPEC64): Likewise.
9777 (ASM_SPEC_COMMON): Likewise.
9778 (ASM_SPEC): Likewise.
9779 (INVALID_64BIT): Likewise.
9780 (LINK_OS_DEFAULT_SPEC): Likewise.
9781 (LINK_OS_SPEC32): Likewise.
9782 (LINK_OS_SPEC64): Likewise.
9783 (POWERPC_LINUX): Likewise.
9784 (PTRDIFF_TYPE): Likewise.
9785 (RESTORE_FP_PREFIX): Likewise.
9786 (RESTORE_FP_SUFFIX): Likewise.
9787 (SAVE_FP_PREFIX): Likewise.
9788 (SAVE_FP_SUFFIX): Likewise.
9789 (SIZE_TYPE): Likewise.
9790 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9791 (TARGET_64BIT): Likewise.
9792 (TARGET_64BIT): Likewise.
9793 (TARGET_AIX): Likewise.
9794 (WCHAR_TYPE_SIZE): Likewise.
9795 (WCHAR_TYPE): Undefine.
9796 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9797 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9798 (CPP_OS_RTEMS_SPEC): Delete.
9799 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9800 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9802 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9804 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9806 PR tree-optimization/81578
9807 * tree-parloops.c (build_new_reduction): Bail out if
9808 reduction_code isn't one of the standard OpenMP reductions.
9809 Move the details printing after that decision.
9811 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9813 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9814 related to reload_in_progress.
9815 (splat_input_operand): Likewise.
9816 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9818 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9820 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9821 (TARGET_INSTANTIATE_DECLS): Likewise.
9822 (legitimate_indexed_address_p): Delete reload_in_progress code.
9823 (rs6000_debug_legitimate_address_p): Likewise.
9824 (rs6000_eliminate_indexed_memrefs): Likewise.
9825 (rs6000_emit_le_vsx_store): Likewise.
9826 (rs6000_emit_move_si_sf_subreg): Likewise.
9827 (rs6000_emit_move): Likewise.
9828 (register_to_reg_type): Likewise.
9829 (rs6000_pre_atomic_barrier): Likewise.
9830 (rs6000_machopic_legitimize_pic_address): Likewise.
9831 (rs6000_allocate_stack_temp): Likewise.
9832 (rs6000_address_for_fpconvert): Likewise.
9833 (rs6000_address_for_altivec): Likewise.
9834 (rs6000_secondary_memory_needed_rtx): Delete function.
9835 (rs6000_check_sdmode): Likewise.
9836 (rs6000_alloc_sdmode_stack_slot): Likewise.
9837 (rs6000_instantiate_decls): Likewise.
9838 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
9839 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
9840 Delete reload_in_progress.
9841 (*vec_reload_and_plus_<mptrsize>): Likewise.
9842 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
9843 (vsx_div_v2di): Likewise.
9844 (vsx_udiv_v2di): Likewise.
9846 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9848 * config/rs6000/rs6000.opt (mlra): Replace with stub.
9849 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
9850 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
9851 (rs6000_debug_reg_global): Delete print of LRA status.
9852 (rs6000_option_override_internal): Delete dead LRA related code.
9853 (rs6000_lra_p): Delete function.
9854 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
9856 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
9858 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
9859 * config/riscv/rtems.h: New file.
9861 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9862 Sudakshina Das <sudi.das@arm.com>
9864 * config/aarch64/aarch64.md
9865 (define_split for and<mode>3nr_compare): Move
9866 non aarch64_logical_operand to a register.
9867 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
9868 register immediate operand to a register.
9869 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
9871 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9874 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
9876 2017-07-27 Richard Biener <rguenther@suse.de>
9878 PR tree-optimization/81573
9879 PR tree-optimization/81494
9880 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
9881 multi defuse cycle case.
9883 2017-07-27 Richard Biener <rguenther@suse.de>
9885 PR tree-optimization/81571
9886 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
9889 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
9891 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
9892 earlier and only if MASK_FPU is set. Adjust formatting.
9894 2017-07-27 Martin Liska <mliska@suse.cz>
9896 * opt-functions.awk: Add validation of value of Init.
9897 * optc-gen.awk: Pass new argument.
9899 2017-07-27 Martin Liska <mliska@suse.cz>
9901 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
9902 Fix wrong condition.
9904 2017-07-27 Martin Liska <mliska@suse.cz>
9906 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
9907 BBs and edges seen by autoFDO.
9909 2017-07-27 Richard Biener <rguenther@suse.de>
9911 PR tree-optimization/81502
9912 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
9913 with incompatible but same sized type.
9914 (execute_update_addresses_taken): Likewise.
9916 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
9918 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
9919 flag_tree_loop_vectorize rather than flag_tree_vectorize.
9921 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9924 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
9925 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
9926 Change s_operand to memory_operand.
9928 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
9930 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
9931 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
9932 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
9933 Emit instructions rather than returning an expression. Handle TFmode
9934 and KFmode by casting to TImode.
9935 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
9936 (rs6000_emit_le_vsx_store): Likewise.
9937 * config/rs6000/vsx.md (VSX_TI): New iterator.
9938 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
9939 (*vsx_le_undo_permute_<mode>): Likewise.
9940 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
9941 emit the split sequence.
9942 (*vsx_le_perm_store_<mode>): Likewise.
9944 2017-07-27 Jakub Jelinek <jakub@redhat.com>
9946 PR tree-optimization/81555
9947 PR tree-optimization/81556
9948 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
9949 if true, force CHANGED for the recursive invocation.
9950 (reassociate_bb): Remember original length of ops array, pass
9951 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
9953 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
9954 attributes for noipa attribute. For naked attribute use
9955 lookup_attribute first before lookup_attribute_spec.
9956 * final.c (rest_of_handle_final): Disable IPA RA for functions with
9958 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
9959 for functions with noipa attribute.
9960 (cgraph_externally_visible_p): Return true for functions with noipa
9962 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
9963 for functions with noipa attribute.
9964 * doc/extend.texi: Document noipa function attribute.
9965 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
9966 also for functions with noipa attribute.
9967 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
9969 2017-07-26 Andrew Pinski <apinski@cavium.com>
9971 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
9972 vec_unalign_load_cost and vec_unalign_store_cost.
9974 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9976 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
9977 -mvsx-small-integer option.
9978 (ISA_3_0_MASKS_IEEE): Likewise.
9979 (OTHER_VSX_VECTOR_MASKS): Likewise.
9980 (POWERPC_MASKS): Likewise.
9981 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
9982 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
9983 code, only testing for DImode being allowed in non-VSX floating
9985 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
9986 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
9988 (rs6000_option_override_internal): Delete -mvsx-small-integer.
9989 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
9990 TARGET_P8_VECTOR test.
9991 (rs6000_secondary_reload_simple_move): Likewise.
9992 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
9993 since TARGET_P9_VECTOR was already tested.
9994 (rs6000_opt_masks): Remove -mvsx-small-integer.
9995 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
9996 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
9998 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
9999 test for TARGET_VEXTRACTUB was used, and that uses
10001 (p9 extract splitter): Likewise.
10002 (vsx_extract_<mode>_di_p9): Likewise.
10003 (vsx_extract_<mode>_store_p9): Likewise.
10004 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10005 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10006 the elimination of TARGET_VSX_SMALL_INTEGER.
10007 (vsx_extract_<mode>_p8): Likewise.
10008 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10009 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10010 (vsx_set_<mode>_p9): Likewise.
10011 (vsx_set_v4sf_p9): Likewise.
10012 (vsx_set_v4sf_p9_zero): Likewise.
10013 (vsx_insert_extract_v4sf_p9): Likewise.
10014 (vsx_insert_extract_v4sf_p9_2): Likewise.
10015 * config/rs6000/rs6000.md (sign extend splitter): Change
10016 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10017 (floatsi<mode>2_lfiwax_mem): Likewise.
10018 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10019 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10020 since a test for TARGET_P9_VECTOR was used.
10021 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10022 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10023 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10024 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10025 TARGET_P8_VECTOR test.
10026 (fix_trunc<mode>si2_stfiwx): Likewise.
10027 (fix_trunc<mode>si2_internal): Likewise.
10028 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10029 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10031 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10032 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10033 TARGET_P8_VECTOR test.
10034 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10035 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10036 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10038 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10039 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10040 since a test for TARGET_P9_VECTOR was used.
10041 (splitter for loading small constants): Likewise.
10043 2017-07-26 Andrew Pinski <apinski@cavium.com>
10045 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10048 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10051 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10052 (fp_valid_at): Likewise.
10054 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10056 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10057 (qdf24xx_addrcost_table): Likewise.
10058 (cortexa57_tunings): Update to use generic_branch_cost.
10059 (cortexa72_tunings): Likewise.
10060 (cortexa73_tunings): Likewise.
10061 (qdf24xx_tunings): Likewise.
10063 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10065 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10066 (thunderx2t99_branch_cost): Likewise.
10067 (cortexa35_tunings): Update to use generic_branch_cost.
10068 (cortexa53_tunings): Likewise.
10069 (cortexa57_tunings): Likewise.
10070 (cortexa72_tunings): Likewise.
10071 (cortexa73_tunings): Likewise.
10072 (thunderx2t99_tunings): Likewise.
10074 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10076 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10077 (sparc_option_override): Honour MASK_FSMULD.
10078 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10079 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10080 * config/sparc/sparc.opt (mfsmuld): New option.
10081 * doc/invoke.texi (mfsmuld): Document option.
10083 2017-07-26 Marek Polacek <polacek@redhat.com>
10085 PR middle-end/70992
10086 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10088 2017-07-26 Richard Biener <rguenther@suse.de>
10090 * gimple-match-head.c (do_valueize): Return OP if valueize
10092 (get_def): New helper to get at the def stmt of a SSA name
10093 if valueize allows.
10094 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10095 do_valueize to get at the def stmt.
10096 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10098 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10100 PR middle-end/46932
10101 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10103 2017-07-26 Martin Liska <mliska@suse.cz>
10106 * function.c (expand_function_start): Make expansion of
10107 nonlocal_goto_save_area after parm_birth_insn.
10109 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10111 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10112 from all CPU target flags enable members.
10114 2017-07-26 Richard Biener <rguenther@suse.de>
10116 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10117 (decision_tree::gen): Make 'type' const.
10118 (write_predicate): Likewise.
10120 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10122 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10123 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10124 (rs6000_option_override_internal): Likewise.
10125 (rs6000_expand_vector_set): Likewise.
10126 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10127 (TARGET_UPPER_REGS_SF): Likewise.
10128 (TARGET_UPPER_REGS_DI): Likewise.
10129 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10130 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10131 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10132 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10133 (Splitters for DI constants in Altivec registers): Likewise.
10134 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10135 (vsx_set_v4sf_p9): Likewise.
10136 (vsx_set_v4sf_p9_zero): Likewise.
10137 (vsx_insert_extract_v4sf_p9): Likewise.
10138 (vsx_insert_extract_v4sf_p9_2): Likewise.
10140 2017-07-25 Carl Love <cel@us.ibm.com>
10142 * doc/extend.texi: Update the built-in documentation file for the
10143 existing built-in functions
10144 vector signed char vec_cnttz (vector signed char);
10145 vector unsigned char vec_cnttz (vector unsigned char);
10146 vector signed short vec_cnttz (vector signed short);
10147 vector unsigned short vec_cnttz (vector unsigned short);
10148 vector signed int vec_cnttz (vector signed int);
10149 vector unsigned int vec_cnttz (vector unsigned int);
10150 vector signed long long vec_cnttz (vector signed long long);
10151 vector unsigned long long vec_cnttz (vector unsigned long long);
10153 2017-07-25 Andrew Pinski <apinski@cavium.com>
10155 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10156 accesses where the use is for the first operand of a BIT_INSERT.
10158 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10161 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10162 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10164 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10166 * config/i386/gstabs.h: Delete.
10167 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10169 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10171 * config/i386/i386.c (ix86_decompose_address): Do not check for
10172 register RTX when looking at index_reg or base_reg.
10173 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10175 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10177 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10178 to update EH info here.
10180 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10182 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10184 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10186 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10188 2017-07-25 Torsten Duwe <duwe@suse.de>
10190 * common.opt: Introduce -fpatchable-function-entry
10191 command line option, and its variables function_entry_patch_area_size
10192 and function_entry_patch_area_start.
10193 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10194 including a two-value parser.
10195 * target.def (print_patchable_function_entry): New target hook.
10196 * targhooks.h (default_print_patchable_function_entry): New function.
10197 * targhooks.c (default_print_patchable_function_entry): Likewise.
10198 * toplev.c (process_options): Switch off IPA-RA if
10199 patchable function entries are being generated.
10200 * varasm.c (assemble_start_function): Look at the
10201 patchable-function-entry command line switch and current
10202 function attributes and maybe generate NOP instructions by
10203 calling the print_patchable_function_entry hook.
10204 * doc/extend.texi: Document patchable_function_entry attribute.
10205 * doc/invoke.texi: Document -fpatchable_function_entry
10206 command line option.
10207 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10209 * doc/tm.texi: Re-generate.
10211 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10214 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10215 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10217 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10219 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10221 2017-07-25 Richard Biener <rguenther@suse.de>
10223 PR tree-optimization/81455
10224 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10225 not walk in cycles when looking for guards.
10227 2017-07-25 Richard Biener <rguenther@suse.de>
10229 PR tree-optimization/81529
10230 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10231 when optimizing backedge uses.
10233 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10235 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10237 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10238 to dl_section_ref. On AIX, append an expression to subtract
10239 the size of the section length to dl_section_ref.
10241 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10243 * configure.ac: If any of the config.* scripts fail, exit 1.
10244 * configure: Regenerate.
10246 2017-07-25 Richard Biener <rguenther@suse.de>
10248 PR middle-end/81546
10249 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10250 of immediate uses, be more verbose on errors.
10252 2017-07-25 Richard Biener <rguenther@suse.de>
10254 PR tree-optimization/81510
10255 * tree-vect-loop.c (vect_is_simple_reduction): When the
10256 reduction stmt is not inside the loop bail out.
10258 2017-07-25 Richard Biener <rguenther@suse.de>
10260 PR tree-optimization/81303
10261 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10262 profitability check against LOOP_VINFO_NITERSM1.
10264 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10266 * domwalk.c (cmp_bb_postorder): Simplify.
10267 (sort_bbs_postorder): New function. Use it...
10268 (dom_walker::walk): ...here to optimize common cases.
10270 2017-07-25 Martin Liska <mliska@suse.cz>
10273 * ipa-visibility.c (function_and_variable_visibility): Make the
10274 redirection just on target that supports aliasing.
10275 Fix GNU coding style.
10277 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10280 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10282 * config/arm/rtems.h: Likewise.
10283 * config/bfin/rtems.h: Likewise.
10284 * config/i386/rtemself.h: Likewise.
10285 * config/lm32/rtems.h: Likewise.
10286 * config/m32c/rtems.h: Likewise.
10287 * config/m68k/rtemself.h: Likewise.
10288 * config/microblaze/rtems.h: Likewise.
10289 * config/mips/rtems.h: Likewise.
10290 * config/moxie/rtems.h: Likewise.
10291 * config/nios2/rtems.h: Likewise.
10292 * config/powerpcspe/rtems.h: Likewise.
10293 * config/rs6000/rtems.h: Likewise.
10294 * config/rtems.h: Likewise.
10295 * config/sh/rtems.h: Likewise.
10296 * config/sh/rtemself.h: Likewise.
10297 * config/sparc/rtemself.h: Likewise.
10299 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10302 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10303 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10304 * tree-ssa-structalias.c (alias_get_name): Same.
10306 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10309 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10310 instructions if no du chain is found.
10312 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10314 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10316 2017-07-25 Richard Biener <rguenther@suse.de>
10318 PR middle-end/81505
10319 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10322 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10324 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10325 upper-regs options.
10326 (ISA_2_7_MASKS_SERVER): Likewise.
10327 (ISA_3_0_MASKS_IEEE): Likewise.
10328 (OTHER_P8_VECTOR_MASKS): Likewise.
10329 (OTHER_VSX_VECTOR_MASKS): Likewise.
10330 (POWERPC_MASKS): Likewise.
10331 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10332 duplicate list of options.
10333 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10334 explicit -mupper-regs options.
10335 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10336 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10337 alias for -mupper-regs-df.
10338 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10339 (rs6000_init_hard_regno_mode_ok): Likewise.
10340 (rs6000_option_override_internal): Likewise.
10341 (rs6000_opt_masks): Likewise.
10342 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10343 options in terms of whether -mvsx or -mpower8-vector was used.
10344 (TARGET_UPPER_REGS_DI): Likewise.
10345 (TARGET_UPPER_REGS_SF): Likewise.
10346 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10347 -mupper-regs-* options.
10349 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10351 * passes.c (emergency_dump_function): Print some empty lines and a
10352 header before the RTL dump.
10354 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10356 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10358 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10361 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10362 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10364 2017-07-24 Carl Love <cel@us.ibm.com>
10366 * config/rs6000/rs6000-c.c: Add support for built-in functions
10367 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10368 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10369 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10370 vec_extract_fp_from_shortl): Add defines for the two builtins.
10371 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10372 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10374 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10375 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10376 * doc/extend.texi: Update the built-in documentation file for the
10377 new built-in function.
10379 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10382 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10384 * doc/generic.texi: Likewise.
10385 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10386 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10388 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10390 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10391 (aarch64_mls_elt_merge<mode>): Likewise.
10393 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10395 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10396 having __cxa_atexit.
10398 2017-07-23 Michael Collison <michael.collison@arm.com>
10400 * config/arm/arm.c (arm_option_override): Deprecate
10401 use of -mstructure-size-boundary.
10402 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10403 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10405 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10408 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10409 Reduce cost estimate for direct moves.
10411 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10414 * config/i386/i386.c (ix86_option_override_internal): Disable
10415 BMI, BMI2 and TBM instructions for -m16.
10417 2017-07-21 Carl Love <cel@us.ibm.com>
10419 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10420 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10421 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10422 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10423 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10424 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10425 VMULOSW): New enum "unspec" values.
10426 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10427 altivec_vmulosw): New patterns.
10428 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10429 VMULOSW): Add definitions.
10431 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10433 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10434 (qdf24xx): Likewise.
10435 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10436 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10437 (AARCH64_FL_V8_1): Renumber.
10438 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10439 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10440 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10441 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10442 rdma to feature modifiers list.
10444 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10446 PR middle-end/56727
10447 * ipa-visibility (function_and_variable_visibility): Convert
10448 recursive PLT call to direct call if appropriate.
10450 2017-07-21 Andrew Pinski <apinski@cavium.com>
10452 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10453 operand 1 to see if the types precision matches.
10454 * fold-const.c (operand_equal_p): Likewise.
10456 2017-07-21 Richard Biener <rguenther@suse.de>
10458 PR tree-optimization/81303
10459 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10460 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10461 (vect_peeling_hash_get_lowest_cost): Adjust.
10462 (vect_enhance_data_refs_alignment): Likewise. Use
10463 vect_get_peeling_costs_all_drs to compute the penalty for no
10464 peeling to match up costs.
10466 2017-07-21 Richard Biener <rguenther@suse.de>
10468 PR tree-optimization/81500
10469 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10470 we didn't identify a reduction path.
10472 2017-07-21 Tom de Vries <tom@codesourcery.com>
10473 Cesar Philippidis <cesar@codesourcery.com>
10475 PR gcov-profile/81442
10476 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10479 2017-07-21 Tom de Vries <tom@codesourcery.com>
10482 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10484 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10485 nvptx_override_options_after_change.
10487 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10489 * dwarf2out.c (output_file_names): Avoid double testing for
10490 dwarf_version >= 5.
10492 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10494 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10496 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10498 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10500 (try_crossjump_to_edge): Do not punt on partitioned functions.
10502 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10504 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10505 Put all BBs reachable only via paths crossing cold region to cold
10507 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10509 2016-07-21 Richard Biener <rguenther@suse.de>
10511 PR tree-optimization/81303
10512 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10513 into account prologue and epilogue iterations when raising
10514 min_profitable_iters to sth at least covering one vector iteration.
10516 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10518 * config/arm/arm.c (arm_test_cpu_arch_dat):
10521 2017-07-20 Nathan Sidwell <nathan@acm.org>
10523 Remove TYPE_METHODS.
10524 * tree.h (TYPE_METHODS): Delete.
10525 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10526 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10527 (dbxout_type_methods): Scan TYPE_FIELDS.
10528 (dbxout_type): Don't check TYPE_METHODS here.
10529 * function.c (use_register_for_decl): Always ignore register for
10530 class types when not optimizing.
10531 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10532 * tree.c (free_lang_data_in_type): Stitch out member functions and
10533 templates from TYPE_FIELDS.
10534 (build_distinct_type_copy, verify_type_variant,
10535 verify_type): Member fns are on TYPE_FIELDS.
10536 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10537 * tree-pretty-print.c (dump_generic_node): Likewise.
10539 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10542 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10543 V2TImode and V4TImode.
10544 (ix86_expand_vector_extract): Likewise.
10545 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10546 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10547 (ssescalarmode): Handle V4TImode and V2TImode.
10548 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10549 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10550 (VEXTRACTI128_MODE): New mode iterator.
10551 (splitter for *vec_extractv?ti first element): New.
10552 (VEC_INIT_MODE): New mode iterator.
10553 (vec_init<mode>): Consolidate 3 expanders into one using
10554 VEC_INIT_MODE mode iterator.
10556 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10558 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10559 non_spilled_static_chain_regno_p.
10561 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10563 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10565 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10567 * bb-reorder.c (connect_traces): Allow copying of blocks within
10570 2017-07-20 Richard Biener <rguenther@suse.de>
10572 * gimple.h (gimple_phi_result): Add gphi * overload.
10573 (gimple_phi_result_ptr): Likewise.
10574 (gimple_phi_arg): Likewise. Adjust index assert to only
10575 allow actual argument accesses rather than all slots available
10577 (gimple_phi_arg_def): Add gphi * overload.
10578 * tree-phinodes.c (make_phi_node): Initialize only actual
10580 (resize_phi_node): Clear memory not covered by old node,
10581 do not initialize excess argument slots.
10582 (reserve_phi_args_for_new_edge): Initialize new argument slot
10585 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10587 PR tree-optimization/81388
10589 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10591 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10592 by removing computation of may_be_zero.
10594 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10595 Tom de Vries <tom@codesourcery.com>
10597 PR middle-end/81030
10598 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10599 when gimple level profile disagrees with what RTL expander did.
10601 2017-07-20 Richard Biener <rguenther@suse.de>
10603 PR tree-optimization/61171
10604 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10605 (vect_analyze_stmt): Add slp instance parameter.
10606 (vectorizable_reduction): Likewise.
10607 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10608 (vect_is_simple_reduction): Deal with chains not detected
10609 as SLP reduction chain, specifically not properly associated
10610 chains containing a mix of plus/minus.
10611 (get_reduction_op): Remove.
10612 (get_initial_defs_for_reduction): Simplify, pass in whether
10613 this is a reduction chain, pass in the SLP node for the PHIs.
10614 (vect_create_epilog_for_reduction): Get the SLP instance as
10616 (vectorizable_reduction): Get the SLP instance as arg.
10617 During analysis remember the SLP node with the PHIs in the
10618 instance. Simplify getting at the vectorized reduction PHIs.
10619 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10620 through SLP instance.
10621 (vect_slp_analyze_operations): Likewise.
10622 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10623 (vect_transform_stmt): Likewise.
10625 2017-07-20 Tom de Vries <tom@codesourcery.com>
10627 PR tree-optimization/81489
10628 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10629 read of phi arg location to before loop that modifies phi.
10631 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10633 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10636 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10638 PR middle-end/81331
10639 * except.c (execute): Fix ordering issue.
10641 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10643 PR rtl-optimization/81423
10644 * combine.c (make_compound_operation_int): Don't try to optimize
10645 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10647 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10649 PR rtl-optimization/81423
10650 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10651 with a constant that is -1 in the truncated to mode.
10653 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10655 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10656 (determine_unlikely_bbs): ... here.
10657 * predict.h (propagate_unlikely_bbs_forward): Declare.
10658 * cfgexpand.c (pass_expand::execute): Use it.
10659 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10661 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10662 propagate_unlikely_bbs_forward.
10664 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10666 PR middle-end/81331
10667 * except.c (maybe_add_nop_after_section_switch): New function.
10670 2017-07-19 Tom de Vries <tom@codesourcery.com>
10672 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10674 2017-07-19 Tom de Vries <tom@codesourcery.com>
10676 * gimple.h (gimple_phi_arg): Make assert more strict.
10678 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10680 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10681 * config/rs6000/mmintrin.h: New file.
10682 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10684 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10686 PR tree-optimization/81346
10687 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10689 2017-07-19 Tom de Vries <tom@codesourcery.com>
10691 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10693 2017-07-19 Tom de Vries <tom@codesourcery.com>
10695 * config/nvptx/nvptx-modes.def: Add V2DImode.
10696 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10697 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10698 (nvptx_output_mov_insn): Handle lack of mov.b128.
10699 (nvptx_print_operand): Handle 'H' and 'L' codes.
10700 (nvptx_vector_mode_supported): Allow V2DImode.
10701 (nvptx_preferred_simd_mode): New function.
10702 (nvptx_data_alignment): New function.
10703 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10704 nvptx_preferred_simd_mode.
10705 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10707 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10709 2017-07-19 Tom de Vries <tom@codesourcery.com>
10711 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10712 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10713 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10714 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10715 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10716 (mov<VECIM>_insn): New define_insn.
10717 (define_expand "mov<VECIM>): New define_expand.
10719 2017-07-19 Tom de Vries <tom@codesourcery.com>
10721 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10723 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10725 PR tree-optimization/81346
10726 * fold-const.h (fold_div_compare, range_check_type): Declare.
10727 * fold-const.c (range_check_type): New function.
10728 (build_range_check): Use range_check_type.
10729 (fold_div_compare): No longer static, rewritten into
10730 a match.pd helper function.
10731 (fold_comparison): Don't call fold_div_compare here.
10732 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10733 as helper function.
10735 2017-07-19 Nathan Sidwell <nathan@acm.org>
10737 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10738 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10739 * tree.c (find_decls_types_r, verify_type): Use
10740 TYPE_{MIN,MAX}_VALUE_RAW.
10741 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10742 (hash_tree): Likewise.
10743 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10745 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10748 2017-07-18 Tom de Vries <tom@codesourcery.com>
10750 PR middle-end/81464
10751 * omp-expand.c (expand_omp_for_static_chunk): Handle
10752 equal-argument loop exit phi.
10754 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10757 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10758 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10759 operand 2 predicate.
10760 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10761 operand 2 predicate.
10762 (ror,rol -> rorx splitters): Use const_int_operand as
10763 operand 2 predicate.
10765 2017-06-18 Richard Biener <rguenther@suse.de>
10767 PR tree-optimization/81410
10768 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10769 the gap in the ! slp_perm SLP case after each group.
10771 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10773 PR middle-end/81463
10774 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10777 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10779 PR middle-end/81462
10780 * predict.c (set_even_probabilities): Cleanup; do not affect
10781 probabilities that are already known.
10782 (combine_predictions_for_bb): Call even when count is set.
10784 2017-07-18 Nathan Sidwell <nathan@acm.org>
10786 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10789 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10792 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10793 optimization for loop niter analysis.
10795 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10798 * config/avr/avr.c (avr_optimize_casesi): Don't use
10799 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10801 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10803 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10804 body_cost_vec from _vect_peel_extended_info.
10805 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10806 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10809 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10811 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10813 2017-07-18 Richard Biener <rguenther@suse.de>
10815 PR tree-optimization/80620
10816 PR tree-optimization/81403
10817 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10818 info when re-using a VN table entry.
10820 2017-07-18 Richard Biener <rguenther@suse.de>
10822 PR tree-optimization/81418
10823 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10824 vectype_in. Verify that with lane-reducing reduction operations
10825 we have a single def-use cycle.
10827 2017-07-17 Carl Love <cel@us.ibm.com>
10829 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
10831 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10832 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10833 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10834 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10835 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10836 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10837 VMULOSW): New enum "unspec" values.
10838 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10839 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10840 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10841 altivec_vmulosw): New patterns.
10842 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10843 VMULOSW): Add definitions.
10844 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
10846 * config/alpha/alpha.c: Include predict.h.
10848 2017-07-17 Yury Gribov <tetra2005@gmail.com>
10850 * tree-vrp.c (compare_assert_loc): Fix comparison function
10851 to return predictable results.
10853 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10855 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
10857 (subdi3): Likewise.
10858 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
10859 * doc/invoke.texi (mexpand-adddi): Update text.
10861 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10863 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
10864 that also clobbers the CC register. The old expand code is moved
10866 (*arc_clzsi2): ... here.
10867 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
10868 the CC register. The old expand code is moved to ...
10869 (arc_ctzsi2): ... here.
10871 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10873 * config/arc/arc.opt (mindexed-loads): Use initial value
10874 TARGET_INDEXED_LOADS_DEFAULT.
10875 (mauto-modify-reg): Use initial value
10876 TARGET_AUTO_MODIFY_REG_DEFAULT.
10877 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10878 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10879 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
10880 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
10882 2017-07-17 Martin Liska <mliska@suse.cz>
10885 * opts.c (finish_options): Do not allow -fgnu-tm
10886 w/ -fsanitize={kernel-,}address. Say sorry.
10888 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10891 * tree-loop-distribution.c (classify_partition): Only assert on
10892 numer of iterations.
10893 (merge_dep_scc_partitions): Delete prameter. Update function call.
10894 (distribute_loop): Remove code handling loop with unknown niters.
10895 (pass_loop_distribution::execute): Skip loop with unknown niters.
10897 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10900 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
10901 function sort_partitions_by_post_order.
10903 2017-07-17 Bin Cheng <bin.cheng@arm.com>
10905 PR tree-optimization/81374
10906 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
10907 the max index of basic blocks, rather than number of basic blocks.
10909 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10911 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
10913 (arc_legitimate_pic_operand_p): Likewise.
10914 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
10916 (arc_needs_pcl_p): Likewise.
10917 (arc_legitimate_pc_offset_p): Likewise.
10918 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
10919 function is also used in constrains.md.
10920 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
10921 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
10922 PLUS. Only return true/false in known cases, otherwise assert.
10923 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
10924 is already called in arc_legitimate_constant_p.
10925 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
10927 (LEGITIMATE_PIC_OPERAND_P): Use
10928 arc_raw_symbolic_reference_mentioned_p function.
10929 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
10934 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10935 Andrew Burgess <andrew.burgess@embecosm.com>
10937 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
10938 (arc_return_address_register): New function.
10939 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
10940 (arc_handle_fndecl_attribute): Add naked attribute.
10941 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10942 (TARGET_WARN_FUNC_RETURN): Likewise.
10943 (arc_allocate_stack_slots_for_args): New function.
10944 (arc_warn_func_return): Likewise.
10945 (machine_function): Change type fn_type.
10946 (arc_compute_function_type): Consider new naked function type,
10947 change function return type.
10948 (arc_must_save_register): Adapt to handle new
10949 arc_compute_function_type's return type.
10950 (arc_expand_prologue): Likewise.
10951 (arc_expand_epilogue): Likewise.
10952 (arc_return_address_regs): Delete.
10953 (arc_return_address_register): New function.
10954 (arc_epilogue_uses): Use above function.
10955 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
10956 (arc_function_type): Change encoding, add naked type.
10957 (ARC_INTERRUPT_P): Change to handle the new encoding.
10958 (ARC_FAST_INTERRUPT_P): Likewise.
10959 (ARC_NORMAL_P): Define.
10960 (ARC_NAKED_P): Likewise.
10961 (arc_compute_function_type): Delete prototype.
10962 * config/arc/arc.md (in_ret_delay_slot): Use
10963 arc_return_address_register function.
10964 (simple_return): Likewise.
10965 (p_return_i): Likewise.
10967 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10969 PR tree-optimization/81428
10970 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
10971 can't be built for those types.
10973 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10975 Remove stuff dead since r239246.
10977 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
10978 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
10979 (avr_inform_devices): Remove dead stuff.
10981 2017-07-17 Tamar Christina <tamar.christina@arm.com>
10983 * config/arm/arm_neon.h: Fix softp typo.
10985 2017-07-17 Jakub Jelinek <jakub@redhat.com>
10987 PR tree-optimization/81365
10988 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
10989 aggregate moves onto bb predecessor edges, make sure there are no
10990 loads that could alias the lhs in between the start of bb and the
10993 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
10996 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
10997 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
10998 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11000 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11002 PR tree-optimization/81396
11003 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11004 (init_symbolic_number): Initialize it to 1.
11005 (perform_symbolic_merge): Add n_ops from both operands into the new
11007 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11008 without base_addr as useless if they need more than one operation.
11009 (bswap_replace): Handle !bswap case for NULL base_addr.
11011 2017-07-17 Tom de Vries <tom@codesourcery.com>
11014 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11017 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11019 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11020 conditional builtin define __FIX_LEON3FT_B2BST.
11022 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11024 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11025 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11028 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11030 PR rtl-optimization/81424
11031 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11032 to remove potential trapping from operands if -fnon-call-exceptions.
11034 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11036 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11037 profile_proability for scalling.
11038 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11040 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11042 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11044 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11046 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11047 fixpoint arithmetics.
11049 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11051 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11052 fixpoint arithmetics.
11054 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11056 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11057 fixpoint arithmetics.
11059 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11061 * profile-count.h (profile_probability::from_reg_br_prob_note,
11062 profile_probability::to_reg_br_prob_note): New functions.
11063 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11064 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11065 * predict.c (probability_reliable_p): Update.
11066 (edge_probability_reliable_p): Update.
11067 (br_prob_note_reliable_p): Update.
11068 (invert_br_probabilities): Update.
11069 (add_reg_br_prob_note): New function.
11070 (combine_predictions_for_insn): Update.
11071 * asan.c (asan_clear_shadow): Update.
11072 * cfgbuild.c (compute_outgoing_frequencies): Update.
11073 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11074 (update_br_prob_note): Update.
11075 (rtl_verify_edges): Update.
11076 (purge_dead_edges): Update.
11077 (fixup_reorder_chain): Update.
11078 * emit-rtl.c (try_split): Update.
11079 * ifcvt.c (cond_exec_process_insns): Update.
11080 (cond_exec_process_if_block): Update.
11081 (dead_or_predicable): Update.
11082 * internal-fn.c (expand_addsub_overflow): Update.
11083 (expand_neg_overflow): Update.
11084 (expand_mul_overflow): Update.
11085 * loop-doloop.c (doloop_modify): Update.
11086 * loop-unroll.c (compare_and_jump_seq): Update.
11087 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11088 * predict.h: Update.
11089 * reorg.c (mostly_true_jump): Update.
11091 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11092 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11093 * config/arc/arc.c: (emit_unlikely_jump): Update.
11094 * config/arm/arm.c: (emit_unlikely_jump): Update.
11095 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11096 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11097 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11098 (ix86_print_operand): Update.
11099 (ix86_split_fp_branch): Update.
11100 (predict_jump): Update.
11101 * config/ia64/ia64.c (ia64_print_operand): Update.
11102 * config/mmix/mmix.c (mmix_print_operand): Update.
11103 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11104 (rs6000_expand_split_stack_prologue): Update.
11105 * config/rs6000/rs6000.c: Update.
11106 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11107 (s390_expand_vec_movstr): Update.
11108 (s390_expand_cs_tdsi): Update.
11109 (s390_expand_split_stack_prologue): Update.
11110 * config/sh/sh.c (sh_print_operand): Update.
11111 (expand_cbranchsi4): Update.
11112 (expand_cbranchdi4): Update.
11113 * config/sparc/sparc.c (output_v9branch): Update.
11114 * config/spu/spu.c (get_branch_target): Update.
11115 (ea_load_store_inline): Update.
11116 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11117 * config/tilepro/tilepro.c: Update.
11119 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11121 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11122 (gimplify_save_expr): Likewise.
11124 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11126 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11128 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11130 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11132 * timevar.def (TV_IPA_FNSUMMARY): Define.
11134 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11136 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11137 to back store errata sensitive sequence from being generated.
11138 (sqrtdf2_fix): Likewise.
11140 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11142 * tree-ssa-threadupdate.c (compute_path_counts,
11143 update_joiner_offpath_counts): Use profile_probability.
11145 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11148 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11150 * config/arm/arm-c.c (arm_cpu_builtins): Define
11151 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11153 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11155 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11156 array entries to represent __ieee128 versions of the
11157 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11158 scalar_extract_sig, and scalar_insert_exp built-in functions.
11159 (altivec_resolve_overloaded_builtin): Add special case handling
11160 for the __builtin_scalar_insert_exp function, as represented by
11161 the P9V_BUILTIN_VEC_VSIEDP constant.
11162 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11163 exponent support for __ieee128 argument.
11164 (VSESQP): Add scalar extract signature support for __ieee128
11166 (VSTDCNQP): Add scalar test negative support for __ieee128
11168 (VSIEQP): Add scalar insert exponent support for __int128 argument
11169 with __ieee128 result.
11170 (VSIEQPF): Add scalar insert exponent support for __ieee128
11171 argument with __ieee128 result.
11172 (VSTDCQP): Add scalar test data class support for __ieee128
11174 (VSTDCNQP): Add overload support for scalar test negative with
11175 __ieee128 argument.
11176 (VSTDCQP): Add overload support for scalar test data class
11177 __ieee128 argument.
11178 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11179 UNSPEC_VSX_SXSIGDP.
11180 (UNSPEC_VSX_SIEXPQP): New constant.
11181 (xsxexpqp): New insn for VSX scalar extract exponent quad
11183 (xsxsigqp): New insn for VSX scalar extract significand quad
11185 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11186 precision with floating point argument.
11187 (xststdcqp): New expand for VSX scalar test data class quad
11189 (xststdcnegqp): New expand for VSX scalar test negative quad
11191 (xststdcqp): New insn to match expansions for VSX scalar test data
11192 class quad precision and VSX scalar test negative quad precision.
11193 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11194 special case operand checking to enforce that second operand of
11195 VSX scalar test data class with quad precision argument is a 7-bit
11197 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11198 prototypes and descriptions of __ieee128 versions of
11199 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11200 scalar_test_data_class, and scalar_test_neg built-in functions.
11202 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11204 PR tree-optimization/81162
11205 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11206 replace a negate with an add.
11208 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11210 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11212 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11214 * config/arm/arm-c.c (arm_cpu_builtins): Define
11215 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11217 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11219 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11220 (armv8-r): Set ARM Cortex-R52 as default CPU.
11221 * config/arm/arm-tables.opt: Regenerate.
11222 * config/arm/arm-tune.md: Regenerate.
11223 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11225 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11226 extension for -mcpu=cortex-r52.
11228 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11230 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11231 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11232 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11233 (fp-armv8): Define it as FP_ARMv8 only.
11234 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11235 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11237 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11238 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11239 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11240 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11241 than TARGET_FPU_ARMV8.
11242 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11243 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11244 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11246 * config/arm/neon.md (neon_vrint): Likewise.
11247 (neon_vcvt): Likewise.
11248 (neon_<fmaxmin_op><mode>): Likewise.
11249 (<fmaxmin><mode>3): Likewise.
11250 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11251 * config/arm/predicates.md (arm_cond_move_operator): Check against
11252 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11254 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11256 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11257 to top of function.
11259 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11261 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11262 loop in comment with memset.
11264 2017-07-14 Martin Liska <mliska@suse.cz>
11266 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11267 * dwarf2out.c (is_java): Remove the function.
11268 (output_pubname): Remove usage of the function.
11269 (lower_bound_default): Remove usage of DW_LANG_Java.
11270 (gen_compile_unit_die): Likewise.
11271 * gcc.c: Remove compiler defaults for .java and .zip files.
11272 * gimple-expr.c (remove_suffix): Change as there's no longer
11273 extension than 4-letter one.
11274 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11275 (gimplify_save_expr): Likewise.
11276 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11277 as it's possible even for other languages than Java.
11278 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11279 * lto-opts.c (lto_write_options): Remove reference to Java.
11280 * opts.c (strip_off_ending): Update file extension handling.
11281 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11282 * tree-eh.c (lower_resx): Likewise.
11283 * tree.c (free_lang_data_in_type): Remove dead code.
11284 (find_decls_types_r): Likewise.
11285 (build_common_builtin_nodes): Remove Java from a comment.
11286 (verify_type): Remove dead code.
11287 * varasm.c (assemble_external): Remove Java from a comment.
11289 2017-07-14 Martin Liska <mliska@suse.cz>
11291 * opts.c (finish_options): Add quotes.
11292 (common_handle_option): Likewise.
11294 2017-07-14 Martin Liska <mliska@suse.cz>
11296 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11297 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11298 Remove N_SO_PASCAL.
11299 * dwarf2out.c (lower_bound_default): Do not handle
11301 (gen_compile_unit_die): Likewise.
11302 * gcc.c: Remove default extension binding for GNU Pascal.
11303 * stmt.c: Remove Pascal language from a comment.
11304 * xcoffout.c: Likewise.
11306 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11309 * diagnostic-show-locus.c (fixit_cmp): New function.
11310 (layout::layout): Sort m_fixit_hints.
11311 (column_range::column_range): Assert that the values are valid.
11312 (struct char_span): New struct.
11313 (correction::overwrite): New method.
11314 (struct source_line): New struct.
11315 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11316 calls in terms of classes source_line and char_span, and
11317 correction::overwrite.
11318 (selftest::test_overlapped_fixit_printing_2): New function.
11319 (selftest::diagnostic_show_locus_c_tests): Call it.
11321 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11323 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11324 early if there is no lhs.
11326 2017-07-13 Martin Liska <mliska@suse.cz>
11328 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11329 (gen_reference_type_die): Likewise.
11330 * stor-layout.c: Remove Pascal-related comment.
11332 2017-07-13 Martin Liska <mliska@suse.cz>
11334 * opts.c (finish_options): Add quotes to error messages.
11335 (parse_sanitizer_options): Likewise.
11337 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11339 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11341 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11343 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11345 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11347 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11349 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11351 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11354 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11355 provides the hardware capability bits, define the macro
11356 __BUILTIN_CPU_SUPPORTS__.
11357 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11358 if GLIBC does not provide the hardware capability bits. Add a
11359 gcc_unreachable call if the built-in cpu function is neither
11360 __builtin_cpu_is nor __builtin_cpu_supports.
11361 (rs6000_get_function_versions_dispatcher): Change the warning
11362 that an old GLIBC is used which does not export the capability
11363 bits to be an error.
11364 * doc/extend.texi (target_clones attribute): Document the
11365 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11366 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11367 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11368 the macros defined by GCC if the newer GLIBC is available.
11370 2017-07-12 Jeff Law <law@redhat.com>
11372 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11373 remaining includes slightly.
11374 * config/riscv/riscv-builtins.c: Include profile-count.h.
11376 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11379 * config/avr/avr.c (avr_set_current_function): In diagnostic
11380 messages: Quote keywords and (parts of) identifiers.
11381 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11384 2017-07-12 Carl Love <cel@us.ibm.com>
11386 * config/rs6000/rs6000-c.c: Add support for built-in functions
11387 vector bool char vec_revb (vector bool char);
11388 vector bool short vec_revb (vector short char);
11389 vector bool int vec_revb (vector bool int);
11390 vector bool long long vec_revb (vector bool long long);
11391 * doc/extend.texi: Update the built-in documentation file for the
11392 new built-in functions.
11394 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11396 * config/s390/s390.md: Remove movcc splitter.
11398 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11400 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11401 load/store on condition.
11403 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11406 * config/avr/avr.c (avr_encode_section_info)
11407 [progmem && !TREE_READONLY]: Error if progmem object needs
11410 2017-07-11 Michael Collison <michael.collison@arm.com>
11412 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11415 2017-07-11 Carl Love <cel@us.ibm.com>
11417 * config/rs6000/rs6000-c.c: Add support for builtins
11418 vector unsigned int vec_parity_lsbb (vector signed int);
11419 vector unsigned int vec_parity_lsbb (vector unsigned int);
11420 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11421 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11422 vector unsigned long long vec_parity_lsbb (vector signed long long);
11423 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11424 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11425 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11426 * doc/extend.texi: Update the built-in documentation file for the
11427 new built-in functions.
11429 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11431 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11432 (layout::m_primary_loc): New field.
11433 (layout::layout): Initialize new field. Move location filtering
11434 logic from here to...
11435 (layout::maybe_add_location_range): ...this new method. Add
11436 support for filtering to just the lines already specified by other
11438 (layout::will_show_line_p): New method.
11439 (gcc_rich_location::add_location_if_nearby): New method.
11440 (selftest::test_add_location_if_nearby): New test function.
11441 (selftest::diagnostic_show_locus_c_tests): Call it.
11442 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11445 2017-07-11 Tom de Vries <tom@codesourcery.com>
11447 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11448 (bb_first_real_insn): New function.
11449 (nvptx_single): Add extra initialization of broadcasted condition
11452 2017-07-11 Nathan Sidwell <nathan@acm.org>
11454 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11456 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11458 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11459 Binutils doc as TEXI will mess them up.
11460 * doc/invoke.texi (AVR Options): Same here.
11462 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11464 * config/sparc/sparc.opt (mfix-ut700): New option.
11465 (mfix-gr712rc): Likewise.
11466 (sparc_fix_b2bst): New variable.
11467 * doc/invoke.texi (SPARC options): Document them.
11468 (ARM options): Fix warnings.
11469 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11470 instructions to prevent sequences that can trigger the store-store
11471 errata for certain LEON3FT processors.
11472 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11473 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11474 * config/sparc/sparc.md (fix_b2bst): New attribute.
11475 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11477 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11480 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11482 (*rsqrtsf2_sse): Ditto.
11484 (div<mode>3): Macroize insn from divdf3 and divsf3
11485 using MODEF mode iterator.
11487 2017-07-10 Martin Sebor <msebor@redhat.com>
11489 PR tree-optimization/80397
11490 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11491 instead of testing for equality to INTEGER_TYPE.
11493 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11495 * config.gcc: Remove uclibc from arc target spec.
11497 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11499 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11501 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11504 * lto-wrapper.c (remove_option): New function.
11505 (merge_and_complain): Merge PIC/PIE options more realistically.
11507 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11509 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11513 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11514 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11515 * config.in: Regenerate.
11516 * configure: Regenerate.
11517 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11518 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11519 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11520 (TARGET_GASISR_PROLOGUES): ...target mask.
11521 * common/config/avr/avr-common.c
11522 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11523 Set -mgas-isr-prologues.
11524 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11525 INSERT_PASS_BEFORE for it.
11526 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11527 * config/avr/avr.c (avr_option_override)
11528 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11529 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11530 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11531 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11532 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11534 (make_avr_pass_pre_proep): New function.
11535 (emit_push_sfr) <treg>: Add argument to function and use it
11536 instead of TMP_REG.
11537 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11538 and set machine->gasisr.yes.
11539 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11540 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11541 __gcc_isr.n_pushed to .L__stack_usage.
11542 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11543 (avr_asm_final_postscan_insn): ...this new static function.
11544 * config/avr/avr.h (machine_function)
11545 <is_no_gccisr, use_L__stack_usage>: New fields.
11546 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11547 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11548 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11549 (gasisr, *gasisr): New expander and insn.
11550 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11551 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11552 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11554 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11556 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11559 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11561 Move jump-tables out of .text again.
11564 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11565 (ASM_OUTPUT_ADDR_VEC): New function.
11566 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11567 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11568 INSN_ADDRESSes as asm comment.
11569 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11570 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11571 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11572 * config/avr/avr.md (*tablejump): Adjust comment.
11573 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11574 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11576 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11577 (avr_output_addr_vec): New proto.
11578 (avr_log_t) <insn_addresses>: New field.
11580 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11583 * config/i386/i386.c (ix86_function_arg_advance): Set
11584 outgoing_args_on_stack to true if there are outgoing arguments
11586 (ix86_function_arg): Likewise.
11587 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11588 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11589 * config/i386/i386.h (machine_function): Add
11590 outgoing_args_on_stack.
11592 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11594 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11595 supporting pthreds.
11596 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11598 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11600 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11601 (REAL_H): Remove $(MACHMODE_H).
11602 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11604 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11605 $(MACHMODE_H) and double-int.h.
11606 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11608 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11611 2017-07-07 Andrew Pinski <apinski@cavium.com>
11613 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11614 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11616 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11618 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11619 Add warning if GCC was not configured to link against a GLIBC that
11620 exports the hardware capability bits.
11621 (make_resolver_func): Make resolver function private and not a
11622 COMDAT function. Create the name with clone_function_name instead
11623 of make_unique_name.
11626 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11627 correct operand in doing the split.
11629 2017-07-07 Carl Love <cel@us.ibm.com>
11631 * config/rs6000/rs6000-c: Add support for built-in function
11632 vector unsigned short vec_pack_to_short_fp32 (vector float,
11634 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11635 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11636 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11637 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11638 (convert_4f32_8i16): Add define_expand.
11639 * doc/extend.texi: Update the built-in documentation file for the
11640 new built-in function.
11642 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11644 * config/sparc/m8.md: New file.
11645 * config/sparc/sparc.md: Include m8.md.
11647 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11649 * config/sparc/sparc.opt: New option -mvis4b.
11650 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11651 (sparc_option_override): Handle VIS4B.
11652 (enum sparc_builtins): Define
11653 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11654 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11655 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11656 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11657 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11658 (check_constant_argument): New function.
11659 (sparc_vis_init_builtins): Define builtins
11660 __builtin_vis_dictunpack{8,16,32},
11661 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11662 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11663 __builtin_vis_fpcmpde{8,16,32}shl and
11664 __builtin_vis_fpcmpur{8,16,32}shl.
11665 (sparc_expand_builtin): Check that the constant operands to
11666 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11667 constant and in range.
11668 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11670 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11671 (SPARC_IMM5_P): Likewise.
11672 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11673 (enabled): Handle vis4b.
11674 (UNSPEC_DICTUNPACK): New unspec.
11675 (UNSPEC_FPCMPSHL): Likewise.
11676 (UNSPEC_FPUCMPSHL): Likewise.
11677 (UNSPEC_FPCMPDESHL): Likewise.
11678 (UNSPEC_FPCMPURSHL): Likewise.
11679 (cpu_feature): New CPU feature `vis4b'.
11680 (dictunpack{8,16,32}): New insns.
11681 (FPCSMODE): New mode iterator.
11682 (fpcscond): New code iterator.
11683 (fpcsucond): Likewise.
11684 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11685 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11686 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11687 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11688 * config/sparc/constraints.md: Define constraints `q' for unsigned
11689 2-bit integer constants and `t' for unsigned 5-bit integer
11691 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11693 (imm5_operand_dictunpack16): Likewise.
11694 (imm5_operand_dictunpack32): Likewise.
11695 (imm2_operand): Likewise.
11696 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11697 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11698 ditunpack* and fpcmp*shl builtins.
11700 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11702 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11703 * config.in: Add HAVE_AS_SPARC6 define.
11704 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11706 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11708 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11709 (CPP_CPU_SPEC): Handle m8.
11710 (ASM_CPU_SPEC): Likewise.
11711 * config/sparc/sparc-opts.h (enum processor_type): Add
11713 * config/sparc/sparc.c (m8_costs): New struct.
11714 (sparc_option_override): Handle TARGET_CPU_m8.
11715 (sparc32_initialize_trampoline): Likewise.
11716 (sparc64_initialize_trampoline): Likewise.
11717 (sparc_issue_rate): Likewise.
11718 (sparc_register_move_cost): Likewise.
11719 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11720 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11721 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11722 (CPP_CPU_SPEC): Handle M8.
11723 (ASM_CPU_SPEC): Likewise.
11724 (AS_M8_FLAG): Define.
11725 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11726 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11727 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11729 * configure: Regenerate.
11730 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11733 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11735 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11737 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11738 ("*movdi_insn_sp32"): Do not set v3pipe.
11739 ("*movsi_insn"): Likewise.
11740 ("*movdi_insn_sp64"): Likewise.
11741 ("*movsf_insn"): Likewise.
11742 ("*movdf_insn_sp32"): Likewise.
11743 ("*movdf_insn_sp64"): Likewise.
11744 ("*zero_extendsidi2_insn_sp64"): Likewise.
11745 ("*sign_extendsidi2_insn"): Likewise.
11746 ("*mov<VM32:mode>_insn"): Likewise.
11747 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11748 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11749 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11750 ("<vlop:code><VL:mode>3"): Likewise.
11751 ("*not_<vlop:code><VL:mode>3"): Likewise.
11752 ("*nand<VL:mode>_vis"): Likewise.
11753 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11754 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11755 ("one_cmpl<VL:mode>2"): Likewise.
11756 ("faligndata<VM64:mode>_vis"): Likewise.
11757 ("alignaddrsi_vis"): Likewise.
11758 ("alignaddrdi_vis"): Likweise.
11759 ("alignaddrlsi_vis"): Likewise.
11760 ("alignaddrldi_vis"): Likewise.
11761 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11762 ("bmaskdi_vis"): Likewise.
11763 ("bmasksi_vis"): Likewise.
11764 ("bshuffle<VM64:mode>_vis"): Likewise.
11765 ("cmask8<P:mode>_vis"): Likewise.
11766 ("cmask16<P:mode>_vis"): Likewise.
11767 ("cmask32<P:mode>_vis"): Likewise.
11768 ("pdistn<P:mode>_vis"): Likewise.
11769 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11771 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11773 * config/sparc/sparc.md ("subtype"): New insn attribute.
11774 ("*wrgsr_sp64"): Set insn subtype.
11775 ("*rdgsr_sp64"): Likewise.
11776 ("alignaddrsi_vis"): Likewise.
11777 ("alignaddrdi_vis"): Likewise.
11778 ("alignaddrlsi_vis"): Likewise.
11779 ("alignaddrldi_vis"): Likewise.
11780 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11781 ("fexpand_vis"): Likewise.
11782 ("fpmerge_vis"): Likewise.
11783 ("faligndata<VM64:mode>_vis"): Likewise.
11784 ("bshuffle<VM64:mode>_vis"): Likewise.
11785 ("cmask8<P:mode>_vis"): Likewise.
11786 ("cmask16<P:mode>_vis"): Likewise.
11787 ("cmask32<P:mode>_vis"): Likewise.
11788 ("fchksm16_vis"): Likewise.
11789 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11790 ("fmean16_vis"): Likewise.
11791 ("fp<plusminus_insn>64_vis"): Likewise.
11792 ("<plusminus_insn>v8qi3"): Likewise.
11793 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11794 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11795 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11796 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11797 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11798 ("*movqi_insn"): Likewise.
11799 ("*movhi_insn"): Likewise.
11800 ("*movsi_insn"): Likewise.
11801 ("movsi_pic_gotdata_op"): Likewise.
11802 ("*movdi_insn_sp32"): Likewise.
11803 ("*movdi_insn_sp64"): Likewise.
11804 ("movdi_pic_gotdata_op"): Likewise.
11805 ("*movsf_insn"): Likewise.
11806 ("*movdf_insn_sp32"): Likewise.
11807 ("*movdf_insn_sp64"): Likewise.
11808 ("*zero_extendhisi2_insn"): Likewise.
11809 ("*zero_extendqihi2_insn"): Likewise.
11810 ("*zero_extendqisi2_insn"): Likewise.
11811 ("*zero_extendqidi2_insn"): Likewise.
11812 ("*zero_extendhidi2_insn"): Likewise.
11813 ("*zero_extendsidi2_insn_sp64"): Likewise.
11814 ("ldfsr"): Likewise.
11815 ("prefetch_64"): Likewise.
11816 ("prefetch_32"): Likewise.
11817 ("tie_ld32"): Likewise.
11818 ("tie_ld64"): Likewise.
11819 ("*tldo_ldub_sp32"): Likewise.
11820 ("*tldo_ldub1_sp32"): Likewise.
11821 ("*tldo_ldub2_sp32"): Likewise.
11822 ("*tldo_ldub_sp64"): Likewise.
11823 ("*tldo_ldub1_sp64"): Likewise.
11824 ("*tldo_ldub2_sp64"): Likewise.
11825 ("*tldo_ldub3_sp64"): Likewise.
11826 ("*tldo_lduh_sp32"): Likewise.
11827 ("*tldo_lduh1_sp32"): Likewise.
11828 ("*tldo_lduh_sp64"): Likewise.
11829 ("*tldo_lduh1_sp64"): Likewise.
11830 ("*tldo_lduh2_sp64"): Likewise.
11831 ("*tldo_lduw_sp32"): Likewise.
11832 ("*tldo_lduw_sp64"): Likewise.
11833 ("*tldo_lduw1_sp64"): Likewise.
11834 ("*tldo_ldx_sp64"): Likewise.
11835 ("*mov<VM32:mode>_insn"): Likewise.
11836 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11837 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11839 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11841 * config/sparc/sparc.md ("type"): New insn type viscmp.
11842 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
11844 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
11845 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
11846 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11847 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
11849 ("n7_vis_logical_11cycle"): Likewise.
11850 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
11851 * config/sparc/niagara2.md ("niag3_vis": Likewise.
11852 * config/sparc/niagara.md ("niag_vis"): Likewise.
11853 * config/sparc/ultra3.md ("us3_fga"): Likewise.
11854 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
11856 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11858 * config/sparc/sparc.md: New instruction type `bmask'.
11859 (bmaskdi_vis): Use the `bmask' type.
11860 (bmasksi_vis): Likewise.
11861 * config/sparc/ultra3.md (us3_array): Likewise.
11862 * config/sparc/niagara7.md (n7_array): Likewise.
11863 * config/sparc/niagara4.md (n4_array): Likewise.
11864 * config/sparc/niagara2.md (niag2_vis): Likewise.
11865 (niag3_vis): Likewise.
11866 * config/sparc/niagara.md (niag_vis): Likewise.
11868 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11870 * ipa-comdats.c: Remove optimize check from gate.
11871 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
11872 for functions not optimized.
11873 (ipa_fn_summary_read): Skip optimize check.
11874 (ipa_fn_summary_write): Likewise.
11875 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
11877 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
11879 (can_inline_edge_p): Check flag_pcc_struct_return for match.
11880 (check_callers): Give up on caller which is not optimized.
11881 (inline_small_functions): Likewise.
11882 (ipa_inline): Do not give up when not optimizing.
11883 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
11884 away unoptimizes cdtors.
11885 (whole_program_function_and_variable_visibility): Do
11886 ipa_discover_readonly_nonaddressable_vars in LTO mode.
11887 * ipa.c (process_references): Do not check optimize.
11888 (symbol_table::remove_unreachable_nodes): Update optimize check.
11889 (set_writeonly_bit): Update optimize check.
11890 (pass_ipa_cdtor_merge::gate): Do not check optimize.
11891 (pass_ipa_single_use::gate): Remove.
11893 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11895 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
11896 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
11897 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
11898 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
11899 permute_load, permute_store, adjust_extract, adjust_splat,
11900 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
11901 replace_swap_with_copy, dump_swap_insn_table,
11902 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
11903 recombine_lvx_pattern, recombine_stvx_pattern,
11904 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
11905 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
11906 to file rs6000-p8swap.c.
11907 * config/rs6000/rs6000-p8swap.c: New file.
11908 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
11909 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
11910 and rs6000*-*-* targets.
11912 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11914 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
11916 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11918 * lto-wrapper.c (merge_and_complain): Do not merge
11919 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
11920 fsigned_zeros, ftrapping_math, fwrapv.
11921 (append_compiler_options): Do not track these options.
11922 (append_linker_options): Likewie
11924 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11926 * cgraphunit.c (cgraph_node::finalize_function): When
11927 !flag_toplevel_reorde set no_reorder flag.
11928 (varpool_node::finalize_decl): Likewise.
11929 (symbol_table::compile): Drop no toplevel reorder path.
11931 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
11933 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
11934 edges; zero probability is not better than uninitialized.
11936 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11938 * asan.h (asan_sanitize_allocas_p): Declare.
11939 * asan.c (asan_sanitize_allocas_p): New function.
11940 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
11941 (handle_builtin_alloca): Likewise.
11942 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
11943 if !asan_sanitize_allocas_p.
11944 * params.def (asan-instrument-allocas): Add new option.
11945 * params.h (ASAN_PROTECT_ALLOCAS): Define.
11946 * opts.c (common_handle_option): Disable allocas sanitization for
11949 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
11951 * asan.c: Include gimple-fold.h.
11952 (get_last_alloca_addr): New function.
11953 (handle_builtin_stackrestore): Likewise.
11954 (handle_builtin_alloca): Likewise.
11955 (asan_emit_allocas_unpoison): Likewise.
11956 (get_mem_refs_of_builtin_call): Add new parameter, remove const
11957 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
11958 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
11959 (instrument_builtin_call): Pass gimple iterator to
11960 get_mem_refs_of_builtin_call.
11961 (last_alloca_addr): New global.
11962 * asan.h (asan_emit_allocas_unpoison): Declare.
11963 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
11964 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
11965 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
11966 if function calls alloca.
11967 * gimple-fold.c (replace_call_with_value): Remove static keyword.
11968 * gimple-fold.h (replace_call_with_value): Declare.
11969 * internal-fn.c: Include asan.h.
11970 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
11971 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
11973 2017-07-06 David Malcolm <dmalcolm@redhat.com>
11975 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
11976 (C_SELFTEST_FLAGS): New.
11977 (CPP_SELFTEST_FLAGS): New.
11978 (SELFTEST_DEPS): New, from deps of s-selftest.
11979 (C_SELFTEST_DEPS): New, from deps of s-selftest.
11980 (CPP_SELFTEST_DEPS): New.
11981 (selftest): Add dependency on s-selftest-c++.
11982 (s-selftest): Rename to...
11983 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
11984 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
11985 than SELFTEST_FLAGS.
11986 (selftest-gdb): Rename to...
11987 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
11989 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
11990 (selftest-valgrind): Rename to...
11991 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
11993 (selftest-valgrind): Reintroduce as an alias for
11994 selftest-c-valgrind.
11995 (s-selftest-c++): New.
11996 (selftest-c++-gdb): New.
11997 (selftest-c++-valgrind): New.
11999 2017-07-06 Olivier Hainque <hainque@adacore.com>
12001 * gcc.c (process_command): When deciding if undefined variables
12002 should be ignored when processing specs, accept "gcc -v" as well.
12004 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12006 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12007 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12009 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12011 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12012 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12013 * config/arm/arm-tables.opt: Regenerate.
12014 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12016 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12018 2017-07-06 Carl Love <cel@us.ibm.com>
12020 * ChangeLog: Clean up from mid air collision
12022 2017-07-06 Carl Love <cel@us.ibm.com>
12024 * config/rs6000/rs6000-c.c: Add support for built-in functions
12025 vector signed int vec_subc (vector signed int, vector signed int);
12026 vector signed __int128 vec_subc (vector signed __int128,
12027 vector signed __int128);
12028 vector unsigned __int128 vec_subc (vector unsigned __int128,
12029 vector unsigned __int128);
12030 vector signed int vec_sube (vector signed int, vector signed int,
12031 vector signed int);
12032 vector unsigned int vec_sube (vector unsigned int,
12033 vector unsigned int,
12034 vector unsigned int);
12035 vector signed __int128 vec_sube (vector signed __int128,
12036 vector signed __int128,
12037 vector signed__int128);
12038 vector unsigned __int128 vec_sube (vector unsigned __int128,
12039 vector unsigned __int128,
12040 vector unsigned __int128);
12041 vector signed int vec_subec (vector signed int, vector signed int,
12042 vector signed int);
12043 vector unsigned int vec_subec (vector unsigned int,
12044 vector unsigned int,
12045 vector unsigned int);
12046 vector signed __int128 vec_subec (vector signed __int128,
12047 vector signed __int128,
12048 vector signed__int128);
12049 vector unsigned __int128 vec_subec (vector unsigned __int128,
12050 vector unsigned __int128,
12051 vector unsigned __int128);
12052 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12053 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12054 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12055 BU_ALTIVEC_OVERLOAD_X definitions.
12056 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12057 * doc/extend.texi: Update the built-in documentation file for the new
12058 built-in functions.
12060 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12063 * diagnostic-show-locus.c (layout::layout): Use start and finish
12064 spelling location for the start and finish of each range.
12065 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12066 Add unused aspect param.
12067 * input.c (expand_location_1): Add "aspect" param, and use it
12068 to access the correct part of the location.
12069 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12071 (expand_location_to_spelling_point): Likewise.
12072 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12073 param, and pass it to expand_location_1.
12075 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12077 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12078 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12079 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12080 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12081 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12082 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12083 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12084 _mm_maskz_getmant_ss): New intrinsics.
12085 (__builtin_ia32_getexpss128_mask): Changed to ...
12086 __builtin_ia32_getexpss128_round ... this.
12087 (__builtin_ia32_getexpsd128_mask): Changed to ...
12088 __builtin_ia32_getexpsd128_round ... this.
12089 * config/i386/i386-builtin-types.def
12090 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12091 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12092 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12093 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12094 __builtin_ia32_getmantss_mask_round): New builtins.
12095 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12096 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12097 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12098 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12099 * config/i386/sse.md
12100 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12101 avx512f_sgetexp<mode><mask_scalar_name>
12102 <round_saeonly_scalar_name> ... this.
12103 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12104 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12105 vgetexp<ssescalarmodesuffix>
12106 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12107 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12108 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12109 avx512f_vgetmant<mode><mask_scalar_name>
12110 <round_saeonly_scalar_name> ... this.
12111 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12112 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12113 vgetmant<ssescalarmodesuffix>
12114 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12115 %0<mask_scalar_operand4>, %1, %2
12116 <round_saeonly_scalar_mask_op4>, %3} ... this.
12117 * config/i386/subst.md (mask_scalar_operand4,
12118 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12119 round_saeonly_scalar_nimm_predicate): New subst attributes.
12121 2017-07-06 Julia Koval <julia.koval@intel.com>
12123 * config/i386/i386.c (ix86_erase_embedded_rounding):
12124 Remove code for old rounding pattern.
12126 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12128 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12130 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12132 * doc/sourcebuild.texi (Test Directives, Variants of
12133 dg-require-support): Add documentation for dg-require-stack-check.
12135 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12137 * config/i386/subst.md (mask_scalar, round_scalar,
12138 round_saeonly_scalar): New meta-templates.
12139 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12140 round_scalar_mask_operand3, round_scalar_mask_op3,
12141 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12142 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12143 round_saeonly_scalar_constraint,
12144 round_saeonly_scalar_prefix): New subst attribute.
12145 * config/i386/sse.md
12146 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12147 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12148 <round_scalar_name> ... this.
12149 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12150 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12151 <round_scalar_name> ... this.
12152 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12153 <sse>_vm<code><mode>3<mask_scalar_name>
12154 <round_saeonly_scalar_name> ... this.
12155 (v<plusminus_mnemonic><ssescalarmodesuffix>
12156 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12157 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12158 v<plusminus_mnemonic><ssescalarmodesuffix>
12159 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12160 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12161 (v<multdiv_mnemonic><ssescalarmodesuffix>
12162 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12163 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12164 v<multdiv_mnemonic><ssescalarmodesuffix>
12165 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12166 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12167 (v<maxmin_float><ssescalarmodesuffix>
12168 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12169 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12170 v<maxmin_float><ssescalarmodesuffix>
12171 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12172 %0<mask_scalar_operand3>, %1, %<iptr>2
12173 <round_saeonly_scalar_mask_op3>} ... this.
12175 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12177 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12178 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12180 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12181 Alan Hayward <alan.hayward@arm.com>
12182 David Sherwood <david.sherwood@arm.com>
12184 * combine.c (simplify_if_then_else): Remove "enum" before
12186 * compare-elim.c (can_eliminate_compare): Likewise.
12187 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12189 (aarch64_lookup_simd_builtin_type): Likewise.
12190 (aarch64_simd_builtin_type): Likewise.
12191 (aarch64_init_simd_builtin_types): Likewise.
12192 (aarch64_simd_expand_args): Likewise.
12193 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12195 (aarch64_reverse_mask): Likewise.
12196 (aarch64_simd_emit_reg_reg_move): Likewise.
12197 (aarch64_gen_adjusted_ldpstp): Likewise.
12198 (aarch64_ccmp_mode_to_code): Likewise.
12199 (aarch64_operands_ok_for_ldpstp): Likewise.
12200 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12201 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12203 (aarch64_min_divisions_for_recip_mul): Likewise.
12204 (aarch64_reassociation_width): Likewise.
12205 (aarch64_get_condition_code_1): Likewise.
12206 (aarch64_simd_emit_reg_reg_move): Likewise.
12207 (aarch64_simd_attr_length_rglist): Likewise.
12208 (aarch64_reverse_mask): Likewise.
12209 (aarch64_operands_ok_for_ldpstp): Likewise.
12210 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12211 (aarch64_gen_adjusted_ldpstp): Likewise.
12212 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12214 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12215 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12216 (arm_lookup_simd_builtin_type): Likewise.
12217 (arm_simd_builtin_type): Likewise.
12218 (arm_init_simd_builtin_types): Likewise.
12219 (arm_expand_builtin_args): Likewise.
12220 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12221 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12222 (ft32_setup_incoming_varargs): Likewise.
12223 (ft32_function_arg): Likewise.
12224 (ft32_function_arg_advance): Likewise.
12225 (ft32_pass_by_reference): Likewise.
12226 (ft32_arg_partial_bytes): Likewise.
12227 (ft32_valid_pointer_mode): Likewise.
12228 (ft32_addr_space_pointer_mode): Likewise.
12229 (ft32_addr_space_legitimate_address_p): Likewise.
12230 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12232 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12233 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12234 (iamcu_alignment): Likewise.
12235 (canonicalize_vector_int_perm): Likewise.
12236 (ix86_noce_conversion_profitable_p): Likewise.
12237 (ix86_mpx_bound_mode): Likewise.
12238 (ix86_operands_ok_for_move_multiple): Likewise.
12239 * config/microblaze/microblaze-protos.h
12240 (microblaze_expand_conditional_branch_reg): Likewise.
12241 * config/microblaze/microblaze.c
12242 (microblaze_expand_conditional_branch_reg): Likewise.
12243 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12245 (rs6000_reassociation_width): Likewise.
12246 (rs6000_invalid_binary_op): Likewise.
12247 (fusion_p9_p): Likewise.
12248 (emit_fusion_p9_load): Likewise.
12249 (emit_fusion_p9_store): Likewise.
12250 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12252 (riscv_hard_regno_mode_ok_p): Likewise.
12253 (riscv_address_insns): Likewise.
12254 (riscv_split_symbol): Likewise.
12255 (riscv_legitimize_move): Likewise.
12256 (riscv_function_value): Likewise.
12257 (riscv_hard_regno_nregs): Likewise.
12258 (riscv_expand_builtin): Likewise.
12259 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12260 (riscv_build_integer): Likewise.
12261 (riscv_split_integer): Likewise.
12262 (riscv_legitimate_constant_p): Likewise.
12263 (riscv_cannot_force_const_mem): Likewise.
12264 (riscv_regno_mode_ok_for_base_p): Likewise.
12265 (riscv_valid_base_register_p): Likewise.
12266 (riscv_valid_offset_p): Likewise.
12267 (riscv_valid_lo_sum_p): Likewise.
12268 (riscv_classify_address): Likewise.
12269 (riscv_legitimate_address_p): Likewise.
12270 (riscv_address_insns): Likewise.
12271 (riscv_load_store_insns): Likewise.
12272 (riscv_force_binary): Likewise.
12273 (riscv_split_symbol): Likewise.
12274 (riscv_force_address): Likewise.
12275 (riscv_legitimize_address): Likewise.
12276 (riscv_move_integer): Likewise.
12277 (riscv_legitimize_const_move): Likewise.
12278 (riscv_legitimize_move): Likewise.
12279 (riscv_address_cost): Likewise.
12280 (riscv_subword): Likewise.
12281 (riscv_output_move): Likewise.
12282 (riscv_canonicalize_int_order_test): Likewise.
12283 (riscv_emit_int_order_test): Likewise.
12284 (riscv_function_arg_boundary): Likewise.
12285 (riscv_pass_mode_in_fpr_p): Likewise.
12286 (riscv_pass_fpr_single): Likewise.
12287 (riscv_pass_fpr_pair): Likewise.
12288 (riscv_get_arg_info): Likewise.
12289 (riscv_function_arg): Likewise.
12290 (riscv_function_arg_advance): Likewise.
12291 (riscv_arg_partial_bytes): Likewise.
12292 (riscv_function_value): Likewise.
12293 (riscv_pass_by_reference): Likewise.
12294 (riscv_setup_incoming_varargs): Likewise.
12295 (riscv_print_operand): Likewise.
12296 (riscv_elf_select_rtx_section): Likewise.
12297 (riscv_save_restore_reg): Likewise.
12298 (riscv_for_each_saved_reg): Likewise.
12299 (riscv_register_move_cost): Likewise.
12300 (riscv_hard_regno_mode_ok_p): Likewise.
12301 (riscv_hard_regno_nregs): Likewise.
12302 (riscv_class_max_nregs): Likewise.
12303 (riscv_memory_move_cost): Likewise.
12304 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12305 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12306 (rl78_addr_space_address_mode): Likewise.
12307 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12309 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12310 (rs6000_reassociation_width): Likewise.
12311 (rs6000_invalid_binary_op): Likewise.
12312 (fusion_p9_p): Likewise.
12313 (emit_fusion_p9_load): Likewise.
12314 (emit_fusion_p9_store): Likewise.
12315 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12316 (ok_for_simple_move_operands): Likewise.
12317 (ok_for_simple_move_strict_operands): Likewise.
12318 (ok_for_simple_arith_logic_operands): Likewise.
12319 (visium_legitimize_reload_address): Likewise.
12320 (visium_select_cc_mode): Likewise.
12321 (output_cbranch): Likewise.
12322 (visium_split_double_move): Likewise.
12323 (visium_expand_copysign): Likewise.
12324 (visium_expand_int_cstore): Likewise.
12325 (visium_expand_fp_cstore): Likewise.
12326 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12327 (visium_function_arg): Likewise.
12328 (visium_function_arg_advance): Likewise.
12329 (visium_libcall_value): Likewise.
12330 (visium_setup_incoming_varargs): Likewise.
12331 (visium_legitimate_constant_p): Likewise.
12332 (visium_legitimate_address_p): Likewise.
12333 (visium_legitimize_address): Likewise.
12334 (visium_secondary_reload): Likewise.
12335 (visium_register_move_cost): Likewise.
12336 (visium_memory_move_cost): Likewise.
12337 (prepare_move_operands): Likewise.
12338 (ok_for_simple_move_operands): Likewise.
12339 (ok_for_simple_move_strict_operands): Likewise.
12340 (ok_for_simple_arith_logic_operands): Likewise.
12341 (visium_function_value_1): Likewise.
12342 (rtx_ok_for_offset_p): Likewise.
12343 (visium_legitimize_reload_address): Likewise.
12344 (visium_split_double_move): Likewise.
12345 (visium_expand_copysign): Likewise.
12346 (visium_expand_int_cstore): Likewise.
12347 (visium_expand_fp_cstore): Likewise.
12348 (visium_split_cstore): Likewise.
12349 (visium_select_cc_mode): Likewise.
12350 (visium_split_cbranch): Likewise.
12351 (output_cbranch): Likewise.
12352 (visium_print_operand_address): Likewise.
12353 * expmed.c (flip_storage_order): Likewise.
12354 * expmed.h (emit_cstore): Likewise.
12355 (flip_storage_order): Likewise.
12356 * genrecog.c (validate_pattern): Likewise.
12357 * hsa-gen.c (gen_hsa_addr): Likewise.
12358 * internal-fn.c (expand_arith_overflow): Likewise.
12359 * ira-color.c (allocno_copy_cost_saving): Likewise.
12360 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12361 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12362 (process_invariant_for_inheritance): Likewise.
12363 * lra-eliminations.c (move_plus_up): Likewise.
12364 * omp-low.c (lower_oacc_reductions): Likewise.
12365 * simplify-rtx.c (simplify_subreg): Likewise.
12366 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12367 (TARGET_CHKP_BOUND_MODE): Likewise..
12368 * targhooks.c (default_chkp_bound_mode): Likewise.
12369 (default_setup_incoming_vararg_bounds): Likewise.
12370 * targhooks.h (default_chkp_bound_mode): Likewise.
12371 (default_setup_incoming_vararg_bounds): Likewise.
12372 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12373 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12374 (have_whole_vector_shift): Likewise.
12375 * tree-vect-stmts.c (vectorizable_load): Likewise.
12376 * doc/tm.texi: Regenerate.
12378 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12380 Graceful degrade if Binutils PR21472 is not available.
12383 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12384 .rodata in flash test fails.
12385 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12386 * confgure: Regenerate.
12387 * config.in: Regenerate.
12388 * config/avr/avr.c (avr_asm_named_section)
12389 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12390 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12391 (avr_asm_init_sections): Same.
12393 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12395 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12396 (fma<VH:mode>4_intrinsic): Likewise.
12397 (*fmsub<VCVTF:mode>4): Likewise.
12398 (*fmsub<VH:mode>4_intrinsic): Likewise.
12400 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12403 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12404 Don't depend on "optimize > 0".
12405 (out_movhi_r_mr, out_movqi_mr_r): Same.
12406 (out_movhi_mr_r, out_movqi_r_mr): Same.
12407 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12408 io_address_operand on "optimize > 0".
12410 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12412 * tree-loop-distribution.c: Add general explanantion on the pass.
12413 (generate_loops_for_partition): Mark distributed loop.
12414 (pg_add_dependence_edges): New parameter. Handle alias data
12415 dependence specially and record it in the parameter if asked.
12416 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12417 (init_partition_graph_vertices, add_partition_graph_edge): New.
12418 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12419 (free_partition_graph_vdata, build_partition_graph): New.
12420 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12421 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12422 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12423 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12424 (version_for_distribution_p, finalize_partitions): New.
12425 (distribute_loop): Handle alias data dependence specially. Factor
12426 out loop fusion code as functions and call these functions.
12428 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12430 * tree-loop-distribution.c (classify_partition): New parameter and
12431 better handle reduction statement.
12432 (rdg_build_partitions): Revise comment.
12433 (distribute_loop): Compute statements in all partitions and pass it
12434 to classify_partition.
12436 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12438 * tree-loop-distribution.c (enum partition_type): New.
12439 (struct partition): New field type.
12440 (partition_merge_into): Add parameter. Update partition type.
12441 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12442 (build_rdg_partition_for_vertex): Compute partition type.
12443 (rdg_build_partitions): Dump partition type.
12444 (distribute_loop): Update calls to partition_merge_into.
12446 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12448 * tree-loop-distribution.c (struct ddr_hasher): New.
12449 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12451 (classify_partition): Call get_data_dependence.
12452 (pg_add_dependence_edges): Ditto.
12453 (distribute_loop): Release data dependence hash table.
12455 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12457 * tree-loop-distribution.c (ref_base_address): Delete.
12458 (similar_memory_accesses): Rename ...
12459 (share_memory_accesses): ... to this. Check if partitions access
12460 the same memory reference.
12461 (distribute_loop): Call share_memory_accesses.
12463 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12465 * tree-loop-distribution.c (struct partition): New field recording
12466 its data reference.
12467 (partition_alloc, partition_free): Init and release data refs.
12468 (partition_merge_into): Merge data refs.
12469 (build_rdg_partition_for_vertex): Collect data refs for partition.
12470 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12472 (distribute_loop): Remve data refs from vertice data of partition
12475 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12477 * tree-loop-distribution.c (params.h): Include header file.
12478 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12479 (datarefs_vec): New global var.
12480 (create_rdg_vertices): Use datarefs_vec directly.
12481 (free_rdg): Don't free data references.
12482 (build_rdg): Update use. Don't free data references.
12483 (distribute_loop): Compute global variable for data references.
12484 Bail out if there are too many data references.
12486 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12488 * tree-loop-distribution.c (loop_nest): New global var.
12489 (build_rdg): Use loop directly, rather than loop nest.
12490 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12492 (distribute_loop): Compute global variable loop nest. Update use.
12494 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12496 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12497 (partition_merge_into): New parameter. Dump reason for fusion.
12498 (distribute_loop): Update use of partition_merge_into.
12500 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12502 * tree-loop-distribution.c (bb_top_order_index): New.
12503 (bb_top_order_index_size, bb_top_order_cmp): New.
12504 (stmts_from_loop): Use topological order.
12505 (pass_loop_distribution::execute): Compute and release topological
12506 order for basic blocks.
12508 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12510 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12515 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12516 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12517 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12518 * internal-fn.def (LOOP_DIST_ALIAS): New.
12519 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12520 (fold_loop_internal_call): ... this.
12521 (vect_loop_dist_alias_call): New function.
12522 (set_uid_loop_bbs): Call fold_loop_internal_call.
12523 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12526 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12529 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12530 Require dead FLAGS_REG at the beginning of a peephole.
12532 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12535 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12536 arguments in the call to __builtin_ia32_sbb_u32.
12537 (_subborrow_u64): Swap _X and _Y arguments in the call to
12538 __builtin_ia32_sbb_u64.
12540 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12543 * tree-vrp.c (compare_assert_loc): Turn into a function template
12544 with stable template parameter. Only test if a->e is NULL,
12545 !a->e == !b->e has been verified already. Use e == NULL or
12546 e != NULL instead of e or ! e tests. If stable is true, don't use
12547 iterative_hash_expr, on the other side allow a or b or both NULL
12548 and sort the NULLs last.
12549 (process_assert_insertions): Sort using compare_assert_loc<false>
12550 instead of compare_assert_loc, later sort using
12551 compare_assert_loc<true> before calling process_assert_insertions_for
12552 in a loop. Use break instead of continue once seen NULL pointer.
12554 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12556 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12557 Cortex-R7 and Cortex-R8 processors.
12559 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12561 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12562 uninitialized while src is not.
12564 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12566 * common/config/arm/arm-common.c: Adjust include path for
12568 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12569 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12570 (arm-cpu-data.h): Likewise.
12571 (arm-cpu-cdata.h): Likewise.
12572 * config/arm/arm-cpu.h: Delete.
12573 * config/arm/arm-cpu-cdata.h: Delete.
12574 * config/arm/arm-cpu-data.h: Delete.
12576 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12578 * config/arm/arm-cpus.in (cortex-a55): New.
12579 (cortex-a75): Likewise.
12580 (cortex-a75.cortex-a55): Likewise.
12581 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12583 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12584 * config/arm/arm-cpu-cdata.h: Regenerate.
12585 * config/arm/arm-cpu-data.h: Regenerate.
12586 * config/arm/arm-cpu.h: Regenerate.
12587 * config/arm/arm-tables.opt: Regenerate.
12588 * config/arm/arm-tune.md: Regenerate.
12590 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12592 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12594 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12596 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12599 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12601 PR tree-optimization/81292
12602 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12603 full_string_p, also call adjust_related_strinfos if the adjustment
12604 is simple, otherwise invalidate related strinfos.
12606 2017-07-04 Martin Liska <mliska@suse.cz>
12609 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12610 newly created variable as DECL_IGNORED_P.
12612 2017-07-04 Martin Liska <mliska@suse.cz>
12615 * ipa-inline.c (inline_small_functions):
12616 Use xstrdup_for_dump.
12618 2017-07-04 Tom de Vries <tom@codesourcery.com>
12620 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12622 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12625 * config/darwin.c (darwin_function_switched_text_sections):
12628 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12630 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12632 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12634 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12636 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12638 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12639 min_profitable_iters, and th as inclusive lower bounds.
12640 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12641 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12642 for min_profitable_iters and min_profitable_estimate.
12643 (vect_transform_loop): Treat th as an inclusive lower bound.
12644 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12646 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12649 * config/darwin.c (darwin_function_switched_text_sections):
12650 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12651 in two pieces, and suppress the use of buf.
12653 2017-07-03 Nathan Sidwell <nathan@acm.org>
12655 * hash-table.h (hash_table_mod1): Fix indentation.
12657 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12659 PR middle-end/81290
12660 * predict.c (force_edge_cold): Be more careful about propagation
12662 * profile-count.h (profile_probability::guessed,
12663 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12665 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12667 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12669 * doc/invoke.texi (rcpc architecture extension): Document it.
12671 2017-07-03 Richard Biener <rguenther@suse.de>
12673 PR tree-optimization/60510
12674 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12675 the scalar reduction PHI and use it.
12676 (vectorizable_reduction): Properly guard the single_defuse_cycle
12677 path for non-SLP reduction chains where we cannot use it.
12678 Rework reduc_def/index and vector type deduction. Rework
12679 vector operand gathering during reduction op code-gen.
12680 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12681 chains dissolve the chain and leave it to non-SLP reduction
12684 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12686 * tree-data-ref.h (dr_alignment): Declare.
12687 * tree-data-ref.c (dr_alignment): New function.
12688 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12689 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12691 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12693 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12695 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12696 and base_misalignment fields.
12697 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12698 * tree-data-ref.c: Include builtins.h.
12699 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12700 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12701 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12702 * tree-vect-data-refs.c: Include tree-cfg.h.
12703 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12704 fields instead of calculating an alignment here.
12705 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12706 innermost_loop_behavior fields.
12708 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12710 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12712 (DR_STEP_ALIGNMENT): New macro.
12713 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12714 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12715 (create_data_ref): Print it.
12716 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12717 to tell whether the step preserves vector (mis)alignment.
12718 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12719 Move the check for an integer step and generalise to all INTEGER_CST.
12720 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12721 Print the outer step alignment.
12723 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12725 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12726 with offset_alignment.
12727 (DR_ALIGNED_TO): Delete.
12728 (DR_OFFSET_ALIGNMENT): New macro.
12729 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12730 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12731 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12732 (create_data_ref): Likewise.
12733 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12734 (vect_analyze_data_refs): Likewise.
12735 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12736 creating dummy innermost behavior.
12738 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12740 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12741 with a "innermost_loop_behavior *" and refeence tree.
12742 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12743 (create_data_ref): Update call accordingly.
12744 * tree-predcom.c (find_looparound_phi): Likewise.
12746 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12748 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12749 fields with dr_wrt_vec_loop.
12750 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12751 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12752 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12753 (vect_dr_behavior): New function.
12754 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12755 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12756 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12757 track whether the step preserves the misalignment.
12758 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12759 Use vect_dr_behavior.
12760 (vect_setup_realignment): Update call accordingly.
12761 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12762 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12763 call to vect_create_addr_base_for_vector_ref.
12764 (vect_create_cond_for_align_checks): Likewise.
12765 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12766 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12767 (vect_recog_mask_conversion_pattern): Likewise.
12768 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12769 (new_stmt_vec_info): Remove redundant zeroing.
12771 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12773 * common/config/arm/arm-common.c (arm_be8_option): New function.
12774 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12775 (ISA_ARMv6): Add isa_bit_be8.
12776 * config/arm/arm.h (arm_be8_option): Add prototype.
12777 (BE8_SPEC_FUNCTION): New define.
12778 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12779 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12780 (mlittle-endian): Similarly.
12781 (mbe8, mbe32): New options.
12782 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12783 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12785 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12787 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12789 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12791 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12792 (cleanup_tree_cfg_bb): Use it.
12793 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12795 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12797 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12800 * loop-doloop.c (add_test): Update profile.
12802 2017-07-03 Martin Liska <mliska@suse.cz>
12805 * sanopt.c (rewrite_usage_of_param): New function.
12806 (sanitize_rewrite_addressable_params): Likewise.
12807 (pass_sanopt::execute): Call rewrite_usage_of_param.
12809 2017-07-03 Richard Biener <rguenther@suse.de>
12811 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12812 back to using VIEW_CONVERT_EXPR.
12814 2017-07-03 Martin Liska <mliska@suse.cz>
12817 * doc/extend.texi: Document when a resolver function is
12818 generated for target_clones.
12820 2017-07-03 Martin Liska <mliska@suse.cz>
12822 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12823 and shadow memory of auto variables which are subject of
12824 use-after-scope sanitization.
12825 (asan_expand_mark_ifn): Add do set only when is_poison.
12827 2016-07-03 Richard Biener <rguenther@suse.de>
12829 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
12831 (vect_force_simple_reduction): Record reduction def -> phi mapping.
12832 (vectorizable_reduction): Perform reduction PHI creation when
12833 visiting a reduction PHI and adjust and simplify code generation
12834 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
12835 (vect_transform_loop): Visit reduction PHIs.
12836 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
12837 defs into the SLP tree.
12838 (vect_build_slp_tree): Reduction defs terminate the recursion.
12839 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
12841 (vect_get_vec_defs_for_stmt_copy): Export.
12842 (vect_get_vec_defs): Likewise.
12843 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
12845 (vect_get_vec_defs_for_stmt_copy): Declare.
12846 (vect_get_vec_defs): Likewise.
12848 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12850 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
12851 parameter with a "loop" parameter and use it instead of the
12852 loop containing DR_STMT. Don't check simple_iv when doing
12853 BB analysis. Describe the two analysis modes in the comment.
12855 2017-07-03 Tom de Vries <tom@codesourcery.com>
12857 PR tree-optimization/69468
12858 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
12859 (find_same_succ_bb): Handle ignore_edge_flags.
12861 2017-07-03 Tom de Vries <tom@codesourcery.com>
12863 PR tree-optimization/81192
12864 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
12866 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
12868 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
12870 2017-07-03 Tom de Vries <tom@codesourcery.com>
12872 PR tree-optimization/81192
12873 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
12874 BB_SAME_SUCC (bb) == NULL.
12876 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12878 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
12881 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12883 * dumpfile.c: Include profile-count.h
12884 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
12886 (insert_cond_bb): Update profile.
12887 * tree-cfg.h (insert_cond_bb): Update prototype.
12888 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
12889 * tree-dump.c: Do not include tree-cfg.
12891 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12893 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
12895 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12897 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
12900 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12902 * tree-complex.c (expand_complex_div_wide): update profile.
12904 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12905 Alan Hayward <alan.hayward@arm.com>
12906 David Sherwood <david.sherwood@arm.com>
12908 * Makefile.in (MACHMODE_H): Remove insn-modes.h
12909 (CORETYPES_H): New define.
12910 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
12911 (insn-modes-inline.h, s-modes-inline-h): New rules.
12912 (generated_files): Add insn-modes-inline.h.
12913 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
12914 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
12915 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
12916 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
12917 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
12918 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
12919 (build/gencodes.o, build/genconditions.o): Likewise.
12920 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
12921 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
12922 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
12923 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
12924 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
12925 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
12926 * coretypes.h: Include everything up to real.h for generators.
12927 Include insn-modes.h first. Include wide-int-print.h after
12928 wide-int.h. Include insn-modes-inline.h and then machmode.h.
12929 * machmode.h: Don't include insn-modes.h here.
12930 * function-tests.c: Remove includes of signop.h, machmode.h,
12931 double-int.h and wide-int.h.
12933 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
12935 * optc-save-gen.awk: Likewise.
12936 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
12937 * godump.c: Remove include of wide-int-print.h.
12938 * pretty-print.h: Likewise.
12939 * wide-int-print.cc: Likewise.
12940 * wide-int.cc: Likewise.
12941 * hash-map-tests.c: Remove include of signop.h.
12942 * hash-set-tests.c: Likewise.
12943 * rtl-tests.c: Likewise.
12944 * mkconfig.sh: Remove include of machmode.h.
12945 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
12947 (emit_insn_modes_inline_h): ...this new function. Emit the code
12948 into an insn-modes-inline.h header file, adding appropriate
12949 include guards and end comments.
12950 (emit_insn_modes_c_header): Remove include of machmode.h.
12951 (emit_min_insn_modes_c_header): Include coretypes.h rather than
12953 (main): Handle -i flag and call emit_insn_modes_inline_h when
12956 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12958 * tree-ssa-strlen.c (strinfo): Rename the length field to
12959 nonzero_chars. Add a full_string_p field.
12960 (compare_nonzero_chars, zero_length_string_p): New functions.
12961 (get_addr_stridx): Add an offset_out parameter.
12962 Use compare_nonzero_chars.
12963 (get_stridx): Update accordingly. Use compare_nonzero_chars.
12964 (new_strinfo): Update after above changes to strinfo.
12965 (set_endptr_and_length): Set full_string_p.
12966 (get_string_length): Update after above changes to strinfo.
12967 (unshare_strinfo): Update call to new_strinfo.
12968 (maybe_invalidate): Likewise.
12969 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
12970 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
12971 as a uhwi instead of an shwi. Update after above changes to
12972 strinfo and new_strinfo.
12973 (zero_length_string): Assert that chainsi contains full strings.
12974 Use zero_length_string_p. Update call to new_strinfo.
12975 (adjust_related_strinfos): Update after above changes to strinfo.
12976 Copy full_string_p from origsi.
12977 (adjust_last_stmt): Use zero_length_string_p.
12978 (handle_builtin_strlen): Update after above changes to strinfo and
12979 new_strinfo. Install the lhs as the string length if the previous
12980 entry didn't describe a full string.
12981 (handle_builtin_strchr): Update after above changes to strinfo
12983 (handle_builtin_strcpy): Likewise.
12984 (handle_builtin_strcat): Likewise.
12985 (handle_builtin_malloc): Likewise.
12986 (handle_pointer_plus): Likewise.
12987 (handle_builtin_memcpy): Likewise. Track nonzero characters
12988 that aren't necessarily followed by a nul terminator.
12989 (handle_char_store): Likewise.
12991 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
12993 PR tree-optimization/80769
12994 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
12995 for malloc and calloc. Document the new invariant that all related
12996 strinfos have delayed lengths or none do.
12997 (verify_related_strinfos): Move earlier in file.
12998 (set_endptr_and_length): New function, split out from...
12999 (get_string_length): ...here. Also set the lengths of related
13001 (zero_length_string): Assert that chainsi has known (rather than
13003 (adjust_related_strinfos): Likewise.
13005 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13007 PR tree-optimization/81136
13008 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13009 assert that two references with the same misalignment have the same
13010 compile-time misalignment if those compile-time misalignments
13013 2017-07-01 Andi Kleen <ak@linux.intel.com>
13015 * print-tree.c (print_node): Print all attributes.
13017 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13019 * cfg.c (scale_bbs_frequencies): New function.
13020 * cfg.h (scale_bbs_frequencies): Declare it.
13021 * cfgloopanal.c (single_likely_exit): Cleanup.
13022 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13024 (scale_loop_profile): Likewise.
13025 (loop_version): Likewise.
13026 (create_empty_loop_on_edge): Update.
13027 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13028 scale_loop_frequencies, scale_loop_profile, loopify,
13029 loop_version): Update prototypes.
13030 * modulo-sched.c (sms_schedule): Update.
13031 * predict.c (unlikely_executed_edge_p): Also check probability.
13032 (probably_never_executed_edge_p): Fix typo.
13033 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13034 * tree-parloops.c (gen_parallel_loop): Update.
13035 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13036 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13037 * tree-ssa-loop-split.c (split_loop): Update.
13038 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13039 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13040 (vect_loop_versioning): Update.
13041 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13043 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13045 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13047 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13049 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13050 to keep profile consistent.
13052 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13054 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13055 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13056 * profile-count.h (max_safe_multiplier): Make unsigned.
13057 (profile_count::guessed_zero): New.
13059 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13061 * bb-reorder.c (fix_up_crossing_landing_pad,
13062 fix_crossing_conditional_branches): Use make_single_succ_edge
13063 to keep profile consistent.
13065 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13067 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13070 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13073 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13074 the right scopes, make sure cond_jump isn't preserved between multiple
13075 iterations. Search for fallthru edge whenever there are 3+ edges and
13076 use find_fallthru_edge for it.
13078 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13080 Patch by Alexander Monakov <amonakov@ispras.ru>
13081 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13082 probabilities consistently.
13084 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13086 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13087 emit_cmp_and_jump_insns.
13089 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13092 * tree-inline.c (expand_call_inline): Combine profile statuses.
13094 2017-06-30 Andrew Pinski <apinski@cavium.com>
13096 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13097 fold_stmt returned true.
13099 2017-06-30 Nathan Sidwell <nathan@acm.org>
13101 * ggc.h (empty_string): Delete.
13102 * cfgexpand.c (expand_asm_stmt): Use plain "".
13103 * optabs.c (expand_asm_memory_barrier): Likewise.
13104 * stringpool.c (empty_string): Delete.
13105 (digit_vector, digit_string): Delete.
13106 (ggc_alloc_string): Use plain "", don't optimize single digit
13107 strings. Use ggc_alloc_atomic.
13109 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13111 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13112 comparison set and one other set, use the cost of the non-comparison
13115 2017-06-30 Nathan Sidwell <nathan@acm.org>
13117 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13120 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13122 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13123 loops. Remove now unneeded calls to gimple_switch_set_label() that
13124 just set removed labels to NULL_TREE.
13126 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13128 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13129 (set_range_info): ...here. Only call set_range_info_raw if domain
13131 (set_nonzero_bits): Call set_range_info_raw.
13132 * tree-ssanames.h (set_range_info_raw): New.
13134 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13137 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13138 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13139 of nonimmediate_operand and <store_mask_constraint> instead of m
13140 for the input operand. For V8FI iterator, always split if input
13141 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13142 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13143 <store_mask_predicate> instead of register_operand and
13144 <store_mask_constraint> instead of v for the input operand. Make
13145 sure both operands aren't MEMs for if not <mask_applied>.
13147 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13149 * lto-wrapper.c (copy_file) Close both file descriptors before
13152 2017-06-30 Martin Liska <mliska@suse.cz>
13155 * multiple_target.c (create_dispatcher_calls): Make ifunc
13156 also for function that don't have calls or are not referenced.
13158 2017-06-30 Richard Biener <rguenther@suse.de>
13160 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13161 analyze the first scalar stmt. Move vector type computation
13162 for the BB case here from ...
13163 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13164 live operation processing in the SLP case properly.
13166 2017-06-30 Richard Biener <rguenther@suse.de>
13168 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13170 2017-06-30 Martin Liska <mliska@suse.cz>
13173 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13174 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13176 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13178 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13181 2017-06-30 Richard Biener <rguenther@suse.de>
13183 PR tree-optimization/81249
13184 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13185 condition reduction result to original scalar type.
13187 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13189 * profile-count.h (enum profile_quality): Fix typos and whitespace
13192 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13194 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13195 type for branch probabilities.
13197 2017-06-29 Julian Brown <julian@codesourcery.com>
13198 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13200 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13201 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13202 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13203 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13205 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13207 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13208 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13209 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13210 CC usage from generic code to here.
13211 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13212 CC usage into the target macros.
13214 2017-06-29 Maya Rashish <coypu@sdf.org>
13216 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13219 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13221 * arm/arm-builtins.c: Include profile-count.h
13222 * except.c (sjlj_emit_function_enter): Use
13223 profile_probability::unlikely.
13225 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13227 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13228 and tocrel_offset be pointer args rather than implicitly using
13230 (legitimate_constant_pool_address_p, rs6000_emit_move,
13231 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13232 tocrel_offset and use in toc_relative_expr_p call.
13233 (print_operand, print_operand_address): Use static tocrel_base_oac
13234 and tocrel_offset_oac.
13235 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13238 2017-06-29 Maya Rashish <coypu@sdf.org>
13240 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13242 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13244 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13245 objects, take into account only the alignment of 'op0' and 'mode1' if
13248 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13250 * ccmp.c (ccmp_tree_comparison_p): New function.
13251 (ccmp_candidate_p): Update to use above function.
13252 (get_compare_parts): New function.
13253 (expand_ccmp_next): Update to use new functions.
13254 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13256 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13257 take mode as argument.
13258 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13259 * expr.c (expand_expr_real_1): Pass mode as argument.
13261 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13263 * combine.c (combine_instructions): Print insns to dump_file, together
13266 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13268 * asan.c (asan_emit_stack_protection): Update.
13269 (create_cond_insert_point): Update.
13270 * auto-profile.c (afdo_propagate_circuit): Update.
13271 * basic-block.h (struct edge_def): Turn probability to
13272 profile_probability.
13273 (EDGE_FREQUENCY): Update.
13274 * bb-reorder.c (find_traces_1_round): Update.
13275 (better_edge_p): Update.
13276 (sanitize_hot_paths): Update.
13277 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13278 (make_single_succ_edge): Update.
13279 (check_bb_profile): Update.
13280 (dump_edge_info): Update.
13281 (update_bb_profile_for_threading): Update.
13282 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13283 probabilitycount to 0.
13284 * cfgbuild.c (compute_outgoing_frequencies): Update.
13285 * cfgcleanup.c (try_forward_edges): Update.
13286 (outgoing_edges_match): Update.
13287 (try_crossjump_to_edge): Update.
13288 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13289 (expand_gimple_tailcall): Update.
13290 (construct_init_block): Use make_single_succ_edge.
13291 (construct_exit_block): Use make_single_succ_edge.
13292 * cfghooks.c (verify_flow_info): Update.
13293 (redirect_edge_succ_nodup): Update.
13294 (split_edge): Update.
13295 (account_profile_record): Update.
13296 * cfgloopanal.c (single_likely_exit): Update.
13297 * cfgloopmanip.c (scale_loop_profile): Update.
13298 (set_zero_probability): Remove.
13299 (duplicate_loop_to_header_edge): Update.
13300 * cfgloopmanip.h (loop_version): Update prototype.
13301 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13302 (force_nonfallthru_and_redirect): Update.
13303 (update_br_prob_note): Update.
13304 (rtl_verify_edges): Update.
13305 (purge_dead_edges): Update.
13306 (rtl_lv_add_condition_to_bb): Update.
13307 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13308 * cgraphunit.c (init_lowered_empty_function): Update.
13309 (cgraph_node::expand_thunk): Update.
13310 * cilk-common.c: Include profile-count.h
13311 * dojump.c (inv): Remove.
13312 (jumpifnot): Update.
13313 (jumpifnot_1): Update.
13314 (do_jump_1): Update.
13316 (do_jump_by_parts_greater_rtx): Update.
13317 (do_compare_rtx_and_jump): Update.
13318 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13319 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13320 * dwarf2cfi.c: Include profile-count.h
13321 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13322 (sjlj_emit_dispatch_table): Likewise.
13323 * explow.c: Include profile-count.h
13324 * expmed.c (emit_store_flag_force): Update.
13325 (do_cmp_and_jump): Update.
13326 * expr.c (compare_by_pieces_d::generate): Update.
13327 (compare_by_pieces_d::finish_mode): Update.
13328 (emit_block_move_via_loop): Update.
13329 (store_expr_with_bounds): Update.
13330 (store_constructor): Update.
13331 (expand_expr_real_2): Update.
13332 (expand_expr_real_1): Update.
13333 * expr.h (try_casesi, try_tablejump): Update prototypes.
13334 * gimple-pretty-print.c (dump_probability): Update.
13335 (dump_profile): New.
13336 (dump_gimple_label): Update.
13337 (dump_gimple_bb_header): Update.
13338 * graph.c (draw_cfg_node_succ_edges): Update.
13339 * hsa-gen.c (convert_switch_statements): Update.
13340 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13341 (find_if_case_1): Update.
13342 (find_if_case_2): Update.
13343 * internal-fn.c (expand_arith_overflow_result_store): Update.
13344 (expand_addsub_overflow): Update.
13345 (expand_neg_overflow): Update.
13346 (expand_mul_overflow): Update.
13347 (expand_vector_ubsan_overflow): Update.
13348 * ipa-cp.c (good_cloning_opportunity_p): Update.
13349 * ipa-split.c (split_function): Use make_single_succ_edge.
13350 * ipa-utils.c (ipa_merge_profiles): Update.
13351 * loop-doloop.c (add_test): Update.
13352 (doloop_modify): Update.
13353 * loop-unroll.c (compare_and_jump_seq): Update.
13354 (unroll_loop_runtime_iterations): Update.
13355 * lra-constraints.c (lra_inheritance): Update.
13356 * lto-streamer-in.c (input_cfg): Update.
13357 * lto-streamer-out.c (output_cfg): Update.
13358 * mcf.c (adjust_cfg_counts): Update.
13359 * modulo-sched.c (sms_schedule): Update.
13360 * omp-expand.c (expand_omp_for_init_counts): Update.
13361 (extract_omp_for_update_vars): Update.
13362 (expand_omp_ordered_sink): Update.
13363 (expand_omp_for_ordered_loops): Update.
13364 (expand_omp_for_generic): Update.
13365 (expand_omp_for_static_nochunk): Update.
13366 (expand_omp_for_static_chunk): Update.
13367 (expand_cilk_for): Update.
13368 (expand_omp_simd): Update.
13369 (expand_omp_taskloop_for_outer): Update.
13370 (expand_omp_taskloop_for_inner): Update.
13371 * omp-simd-clone.c (simd_clone_adjust): Update.
13372 * optabs.c (expand_doubleword_shift): Update.
13373 (expand_abs): Update.
13374 (emit_cmp_and_jump_insn_1): Update.
13375 (expand_compare_and_swap_loop): Update.
13376 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13377 * predict.c (predictable_edge_p): Update.
13378 (edge_probability_reliable_p): Update.
13379 (set_even_probabilities): Update.
13380 (combine_predictions_for_insn): Update.
13381 (combine_predictions_for_bb): Update.
13382 (propagate_freq): Update.
13383 (estimate_bb_frequencies): Update.
13384 (force_edge_cold): Update.
13385 * profile-count.c (profile_count::dump): Add missing space into dump.
13386 (profile_count::debug): Add newline.
13387 (profile_count::differs_from_p): Explicitly convert to unsigned.
13388 (profile_count::stream_in): Update.
13389 (profile_probability::dump): New member function.
13390 (profile_probability::debug): New member function.
13391 (profile_probability::differs_from_p): New member function.
13392 (profile_probability::differs_lot_from_p): New member function.
13393 (profile_probability::stream_in): New member function.
13394 (profile_probability::stream_out): New member function.
13395 * profile-count.h (profile_count_quality): Rename to ...
13396 (profile_quality): ... this one.
13397 (profile_probability): New.
13398 (profile_count): Update.
13399 * profile.c (compute_branch_probabilities): Update.
13400 * recog.c (peep2_attempt): Update.
13401 * sched-ebb.c (schedule_ebbs): Update.
13402 * sched-rgn.c (find_single_block_region): Update.
13403 (compute_dom_prob_ps): Update.
13404 (schedule_region): Update.
13405 * sel-sched-ir.c (compute_succs_info): Update.
13406 * stmt.c (struct case_node): Update.
13407 (do_jump_if_equal): Update.
13408 (get_outgoing_edge_probs): Update.
13409 (conditional_probability): Update.
13410 (emit_case_dispatch_table): Update.
13411 (expand_case): Update.
13412 (expand_sjlj_dispatch_table): Update.
13413 (emit_case_nodes): Update.
13414 * targhooks.c: Update.
13415 * tracer.c (better_p): Update.
13416 (find_best_successor): Update.
13417 * trans-mem.c (expand_transaction): Update.
13418 * tree-call-cdce.c: Update.
13419 * tree-cfg.c (gimple_split_edge): Upate.
13420 (move_sese_region_to_fn): Upate.
13421 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13422 * tree-eh.c (lower_resx): Upate.
13423 (cleanup_empty_eh_move_lp): Upate.
13424 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13425 * tree-inline.c (copy_edges_for_bb): Update.
13426 (copy_cfg_body): Update.
13427 * tree-parloops.c (gen_parallel_loop): Update.
13428 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13429 (gimple_gen_time_profiler): Update.
13430 * tree-ssa-dce.c (remove_dead_stmt): Update.
13431 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13432 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13433 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13434 (unloop_loops): Update.
13435 (try_peel_loop): Update.
13436 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13437 * tree-ssa-loop-split.c (connect_loops): Update.
13438 (split_loop): Update.
13439 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13440 (hoist_guard): Update.
13441 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13442 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13443 (value_replacement): Update.
13444 * tree-ssa-reassoc.c (branch_fixup): Update.
13445 * tree-ssa-tail-merge.c (replace_block_by): Update.
13446 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13447 (create_edge_and_update_destination_phis): Update.
13448 (compute_path_counts): Update.
13449 (recompute_probabilities): Update.
13450 (update_joiner_offpath_counts): Update.
13451 (freqs_to_counts_path): Update.
13452 (duplicate_thread_path): Update.
13453 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13454 (struct switch_conv_info): Update.
13455 (gen_inbound_check): Update.
13456 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13457 (vect_do_peeling): Update.
13458 (vect_loop_versioning): Update.
13459 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13460 (optimize_mask_stores): Update.
13461 * ubsan.c (ubsan_expand_null_ifn): Update.
13462 * value-prof.c (gimple_divmod_fixed_value): Update.
13463 (gimple_divmod_fixed_value_transform): Update.
13464 (gimple_mod_pow2): Update.
13465 (gimple_mod_pow2_value_transform): Update.
13466 (gimple_mod_subtract): Update.
13467 (gimple_mod_subtract_transform): Update.
13468 (gimple_ic): Update.
13469 (gimple_stringop_fixed_value): Update.
13470 (gimple_stringops_transform): Update.
13471 * value-prof.h: Update.
13473 2017-06-29 Carl Love <cel@us.ibm.com>
13475 * config/rs6000/rs6000-c.c: Add support for built-in functions
13476 vector signed int vec_signed (vector float);
13477 vector signed long long vec_signed (vector double);
13478 vector signed int vec_signed2 (vector double, vector double);
13479 vector signed int vec_signede (vector double);
13480 vector signed int vec_signedo (vector double);
13481 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13482 instruction generator.
13483 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13484 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13485 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13487 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13488 vunsignede_v2df): Add define_expands.
13489 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13490 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13491 VEC_UNSIGNEDO): Add definitions.
13492 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13493 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13494 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13495 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13496 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13497 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13498 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13499 vec_unsignede, vec_unsignedo): Add builtin defines.
13500 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13502 * doc/extend.texi: Update the built-in documentation file for the
13503 new built-in functions.
13505 2017-06-29 Richard Biener <rguenther@suse.de>
13507 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13508 reduction chains to LOOP_VINFO_REDUCTIONS.
13509 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13510 SLP reductions after processing reduction chains.
13512 2017-06-29 Nathan Sidwell <nathan@acm.org>
13514 * builtins.c (fold_builtin_FUNCTION): Use
13515 lang_hooks.decl_printable_name.
13517 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13519 PR middle-end/81194
13520 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13521 with only one label.
13522 * stmt.c (expand_case): Assert NCASES is greater than one.
13524 2017-06-29 Richard Biener <rguenther@suse.de>
13526 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13528 (group_case_labels): Likewise.
13529 (find_taken_edge): Push sanity checking on val to workers...
13530 (find_taken_edge_cond_expr): ... here
13531 (find_taken_edge_switch_expr): ... and here, handle cases
13532 with just a default label.
13533 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13534 (group_case_labels): Likewise.
13535 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13536 group_case_labels does anything cleanup the CFG again.
13538 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13540 PR tree-optimization/81196
13541 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13542 exit condition comparing two IVs.
13544 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13546 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13547 profile to the dummy entry at the end of the list of architectures.
13548 * config/arm/arm-cpu-cdata.h: Regenerated.
13550 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13551 Michael Collison <michael.collison@arm.com>
13554 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13556 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13557 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13558 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13559 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13560 of shift when the shift amount is masked with constant equal to
13561 the size of the mode.
13562 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13565 2017-06-29 Martin Liska <mliska@suse.cz>
13567 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13569 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13572 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13575 (edge_growth_cache_entry::edge_growth_cache_entry): New
13577 (reset_edge_growth_cache): Update to use constructor.
13579 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13581 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13582 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13583 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13585 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13587 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13588 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13590 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13592 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13593 (*-linux-uclibc*): Add t-uclibc tmake_file.
13594 * config/t-musl: New.
13595 * config/t-uclibc: New.
13597 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13599 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13601 (gen_comm_data): Emit architectural setting of arch_prof.
13602 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13604 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13605 (armv8-m.base, armv8-m.main): Likewise.
13606 * arm-protos.h (arm_build_target): Add profile field.
13607 (arch_option): Likewise.
13608 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13610 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13611 arm_active_target.profile.
13613 2017-06-28 Richard Biener <rguenther@suse.de>
13615 PR middle-end/81227
13616 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13617 TYPE_OVERFLOW_WRAPS.
13618 * match.pd (negate_expr_p): Likewise.
13619 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13620 fold_build2, not fold_binary.
13622 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13624 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13625 Convert memory address to Pmode.
13626 (aarch64_print_operand): Assert MEM operands are always Pmode.
13628 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13631 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13632 Remove redundant if.
13633 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13634 * config/arm/aarch-common-protos.h
13635 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13636 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13638 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13641 * multiple_target.c (create_dispatcher_calls): Set the default
13642 clone to be static, not public.
13644 2017-06-28 Richard Biener <rguenther@suse.de>
13646 * tree-vect-loop.c (vectorizable_reduction): Move special
13647 cond reduction IV var creation ...
13648 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13649 parameter. Use STMT_VINFO_VECTYPE.
13650 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13653 2017-06-28 Martin Liska <mliska@suse.cz>
13656 * ipa-visibility.c (non_local_p): Handle visibility.
13658 2017-06-28 Martin Liska <mliska@suse.cz>
13661 * common.opt: Add IntegerRange to various options.
13662 * opt-functions.awk (integer_range_info): New function.
13663 * optc-gen.awk: Add integer_range_info to cl_options struct.
13664 * opts-common.c (decode_cmdline_option): Handle
13665 CL_ERR_INT_RANGE_ARG.
13666 (cmdline_handle_error): Likewise.
13667 * opts.c (print_filtered_help): Show valid interval in
13668 when --help is provided.
13669 * opts.h (struct cl_option): Add range_min and range_max fields.
13670 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13672 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13674 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13675 (x * C EQ/NE y * C): New transformation.
13677 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13679 * genmultilib (combination_space): Accept '+' in option names.
13681 2017-06-28 Martin Liska <mliska@suse.cz>
13684 * asan.c (instrument_derefs): Bail out inner references
13685 that are hard register variables.
13687 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13690 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13691 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13693 2017-06-28 Richard Biener <rguenther@suse.de>
13695 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13696 (vect_get_slp_defs): Adjust.
13697 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13699 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13701 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13702 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13703 (vectorizable_reduction): Adjust.
13704 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13706 (vect_get_slp_defs): Likewise.
13707 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13708 (vectorizable_bswap): Adjust.
13709 (vectorizable_call): Likewise.
13710 (vectorizable_conversion): Likewise.
13711 (vectorizable_assignment): Likewise.
13712 (vectorizable_shift): Likewise.
13713 (vectorizable_operation): Likewise.
13714 (vectorizable_store): Likewise.
13715 (vectorizable_condition): Likewise.
13716 (vectorizable_comparison): Likewise.
13718 2017-06-28 Michael Collison <michael.collison@arm.com>
13721 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13723 (arm_gen_movmemqi): Removed unused variable 'i'.
13724 Convert 'for' loop into 'while' loop.
13725 (arm_expand_prologue): Remove last unnecessary set of insn.
13726 (thumb_pop): Remove unused variable 'pushed_words'.
13727 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13729 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13731 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13732 * config/s390/s390-protos.h: Add prototype of
13733 s390_rel_address_ok_p.
13734 * config/s390/s390.c (s390_got_symbol): New function.
13735 (s390_rel_address_ok_p): New function.
13736 (legitimize_pic_address): Use s390_rel_address_ok_p.
13737 (s390_load_got): Use s390_got_symbol.
13738 (s390_option_override): Issue error if
13739 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13740 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13742 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13744 2017-06-27 Andrew Pinski <apinski@cavium.com>
13746 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13747 (X * copysign (1.0, X)): New pattern.
13748 (X * copysign (1.0, -X)): New pattern.
13749 (copysign (-1.0, CST)): New pattern.
13751 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13753 * genmultilib (combination_space): Remove variable.
13754 Validate reuse rules against regular expression for any sequence
13755 of multilib options in any order.
13757 2017-06-27 Michael Collison <michael.collison@arm.com>
13759 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13760 call aarch64_split_simd_combine.
13761 * (aarch64_combine_internal<mode>): Delete pattern.
13762 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13763 Allow register and subreg operands.
13765 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13767 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13768 specific need, just fallback on defaults.
13769 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13771 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13772 Olivier Hainque <hainque@adacore.com>
13774 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13776 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13777 targets. Pick a default if no particular attempt applied.
13778 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13781 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13783 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13784 (x86_64-wrs-vxworks7): Likewise.
13786 2017-06-27 Marek Polacek <polacek@redhat.com>
13789 * ubsan.c (instrument_null): Check get_base_address's result for null.
13791 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13793 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13795 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13797 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13798 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13799 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13800 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13801 New function types.
13802 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13803 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13804 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13805 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13806 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13807 BUILT_IN_FEUPDATEENV): New builtins.
13808 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13809 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13810 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13811 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13813 (builtin_structptr_types): Adjust size.
13814 * tree.c (builtin_structptr_types): Add four entries.
13816 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13817 Olivier Hainque <hainque@adacore.com>
13819 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13820 (TLS_SYM): New local macro, forcing reference to __tls__ on
13821 link command lines for VxWorks 7 RTPs, triggering initialization
13823 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13824 OS features TLS support, true for RTPs on VxWorks 7.
13825 * config/vxworks.c (vxworks_override_options): Setup emutls
13828 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13830 * predict.c (test_prediction_value_range): Use -1U instead of -1
13831 to avoid narrowing conversion warning.
13832 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
13833 to avoid narrowing conversion warning.
13834 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
13836 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
13838 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13840 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
13841 64bit configurations.
13842 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
13843 (SIZE_TYPE): Likewise.
13844 * config/vxworks.c (vxworks_emutls_var_fields): Use
13845 long_unsigned_type_node instead of unsigned_type_node as the offset
13846 field type, which is "pointer" mode in emutls.c.
13848 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13851 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
13853 PR middle-end/81207
13854 * gimple-fold.c (replace_call_with_call_and_fold): Handle
13855 gimple_vuse copying separately from gimple_vdef copying.
13857 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13859 * value-prof.c (free_hist): Remove call to memset and the enclosing if
13862 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13863 Olivier Hainque <hainque@adacore.com>
13865 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
13866 for all vxworks7 targets.
13867 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
13868 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
13869 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
13870 variations for VX6/VX7 and 32/64bits later on in ...
13871 (VXWORKS_LIB_SPEC): Leverage new macros.
13872 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
13873 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
13875 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
13877 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
13878 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
13880 2017-06-26 Carl Love <cel@us.ibm.com>
13882 * config/rs6000/rs6000-c.c: Add support for built-in functions
13883 vector bool char vec_reve (vector bool char);
13884 vector signed char vec_reve (vector signed char);
13885 vector unsigned char vec_reve (vector unsigned char);
13886 vector bool int vec_reve (vector bool int);
13887 vector signed int vec_reve (vector signed int);
13888 vector unsigned int vec_reve (vector unsigned int);
13889 vector bool long long vec_reve (vector bool long long);
13890 vector signed long long vec_reve (vector signed long long);
13891 vector unsigned long long vec_reve (vector unsigned long long);
13892 vector bool short vec_reve (vector bool short);
13893 vector signed short vec_reve (vector signed short);
13894 vector double vec_reve (vector double);
13895 vector float vec_reve (vector float);
13896 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
13897 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
13898 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
13899 (altivec_vreve): New pattern.
13900 * config/rs6000/altivec.h (vec_reve): New define.
13901 * doc/extend.texi (vec_rev): Update the built-in documentation file
13902 for the new built-in functions.
13904 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13906 PR tree-optimization/71815
13907 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
13909 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
13911 (slsr_process_phi): Likewise.
13912 (replace_uncond_cands_and_profitable_phis): Don't replace a
13913 multiply candidate with a stride of 1 (copy or cast).
13914 (phi_incr_cost): Call uses_consumed_by_stmt rather than
13916 (lowest_cost_path): Likewise.
13917 (total_savings): Likewise.
13919 2017-06-26 Richard Biener <rguenther@suse.de>
13922 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
13923 Use def_builtin_pure for all gather builtins.
13925 2017-06-26 Richard Biener <rguenther@suse.de>
13927 PR tree-optimization/81203
13928 * tree-tailcall.c (find_tail_calls): Do not move stmts into
13929 non-dominating BBs.
13931 2017-06-26 Marek Polacek <polacek@redhat.com>
13934 * doc/invoke.texi: Document -Wmultistatement-macros.
13936 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
13938 * doc/sourcebuild.texi (ARM-specific attributes): Document new
13939 arm_neon_ok_no_float_abi effective target.
13941 2017-06-26 Richard Biener <rguenther@suse.de>
13943 PR tree-optimization/80928
13944 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
13945 (copy_bbs): Set BB_DUPLICATED flag early.
13946 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
13948 (execute_on_shrinking_pred): Likewise.
13949 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
13950 BB_DUPLICATED blocks.
13951 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
13952 iterate over all PHIs considering removal of *gsi.
13954 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
13956 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
13959 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13961 * config/rs6000/rs6000-string.c: (expand_block_clear,
13962 do_load_for_compare, select_block_compare_mode,
13963 compute_current_alignment, expand_block_compare,
13964 expand_strncmp_align_check, expand_strn_compare,
13965 expand_block_move, rs6000_output_load_multiple)
13966 Move functions related to string/block move/compare
13967 to a separate file.
13968 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
13969 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
13970 for this function which is now used in two files.
13971 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
13972 * config.gcc: Add rs6000-string.o to extra_objs for
13973 targets powerpc*-*-* and rs6000*-*-*.
13975 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
13978 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
13979 32-bit, since indexed is not valid for DImode.
13980 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
13981 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
13982 (define_peephole2 for Altivec d-form load): Add 32-bit support.
13983 (define_peephole2 for Altivec d-form store): Likewise.
13986 * multiple_target.c (create_dispatcher_calls): Only create the
13987 dispatcher call if the function is the default clone of a
13988 versioned function.
13990 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
13992 PR middle-end/80902
13993 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
13994 a call, force the call to not be a tail call.
13996 2017-06-23 Jeff Law <law@redhat.com>
13998 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14001 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14003 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14004 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14005 handling for early expansion of vector shifts (sl,sr,sra,rl).
14006 (builtin_function_type): Add vector shift right instructions
14007 to the unsigned argument list.
14009 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14011 rtl-optimizatoin/79286
14012 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14013 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14014 trap. PIC register plus a const unspec without offset can never trap.
14016 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14018 * tree.h (builtin_structptr_type): New type.
14019 (builtin_structptr_types): Declare new array.
14020 * tree.c (builtin_structptr_types): New array.
14021 (free_lang_data, build_common_tree_nodes): Use it.
14023 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14026 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14029 2017-06-22 Matt Turner <mattst88@gmail.com>
14031 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14032 Lake models to skylake case. Assume skylake for unknown
14033 models with clflushopt.
14035 2017-06-22 Jeff Law <law@redhat.com>
14037 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14038 frame sizes that do not satisfy aarch64_uimm12_shift.
14040 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14042 * profile-count.h (apply_probability,
14043 apply_scale, probability_in): Fix checks for zero.
14045 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14047 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14048 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14050 2016-06-22 Richard Biener <rguenther@suse.de>
14052 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14053 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14054 REDUC_MAX_EXPR support.
14055 (vectorizable_reduction): Likewise.
14056 (vect_create_epilog_for_reduction): Likewise.
14058 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14060 * match.pd (A / (1 << B) -> A >> B): New.
14061 * generic-match-head.c: Include optabs-tree.h.
14062 * gimple-match-head.c: Likewise.
14063 * optabs-tree.h (target_supports_op_p): New.
14064 * optabs-tree.c (target_supports_op_p): New.
14066 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14068 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14069 $gcc_cv_ld --help output.
14070 (gcc_cv_ld_demangle): Likewise.
14071 (gcc_cv_ld_eh_frame_hdr): Likewise.
14072 (gcc_cv_ld_pie): Likewise.
14073 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14074 (gcc_cv_ld_buildid): Likewise.
14075 (gcc_cv_ld_sysroot): Likewise.
14076 (ld_bndplt_support): Likewise.
14077 (ld_pushpopstate_support): Likewise.
14078 * configure: Regenerate.
14079 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14081 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14084 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14085 operands indexes to avoid gap between operands and match_dups.
14087 2017-06-21 Andrew Pinski <apinski@cavium.com>
14089 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14090 Increment Arith_shift and Arith_shift_reg by 1.
14091 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14093 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14094 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14095 (aarch64_strip_extend): Add new argument and test for it.
14096 (aarch64_cheap_mult_shift_p): New function.
14097 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14098 add a cost if it is true.
14099 Update calls to aarch64_strip_extend.
14100 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14102 2017-06-21 Andrew Pinski <apinski@cavium.com>
14104 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14106 (thunderxt88): Likewise.
14107 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14108 (thunderx_prefetch_tune): New variable.
14109 (thunderx2t99_prefetch_tune): Update for the correct values.
14110 (thunderxt88_tunings): New variable.
14111 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14112 generic_prefetch_tune.
14113 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14115 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14117 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14118 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14119 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14120 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14121 (aarch64_atomic_cas<mode>, GPI): Likewise.
14123 2017-06-21 Martin Liska <mliska@suse.cz>
14125 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14126 statements on cold and hot labels.
14127 * predict.c (tree_estimate_probability_bb): Remove the
14128 prediction from this place.
14130 2017-06-21 Martin Liska <mliska@suse.cz>
14132 PR tree-optimization/79489
14133 * gimplify.c (maybe_add_early_return_predict_stmt): New
14135 (gimplify_return_expr): Call the function.
14136 * predict.c (tree_estimate_probability_bb): Remove handling
14138 * predict.def: Update comment about early return predictor.
14139 * gimple-predict.h (is_gimple_predict): New function.
14140 * predict.def: Change default value of early return to 66.
14141 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14143 * passes.def: Put pass_strip_predict_hints to the beginning of
14146 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14148 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14149 FUNCTION_DECL declarations.
14150 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14152 (dwaf2out_decl): Likewise.
14153 * godump.c (go_early_global_decl): Skip call to the real debug hook
14154 for FUNCTION_DECL declarations.
14155 * passes.c (rest_of_decl_compilation): Skip call to the
14156 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14157 -fdump-go-spec is passed.
14159 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14161 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14162 Reorder for compactness.
14163 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14164 (def_builtin_pure, def_builtin_pure2): New functions.
14165 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14167 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14169 * match.pd (nop_convert): New predicate.
14170 ((A +- CST1) +- CST2): Allow some NOP conversions.
14172 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14175 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14176 with ctors/dtors if GOVD_SHARED is set.
14178 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14180 * config/aarch64/aarch64.md (movti_aarch64):
14181 Emit mov rather than orr.
14182 (movtf_aarch64): Likewise.
14183 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14184 Emit mov rather than orr.
14186 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14188 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14189 Swap alternatives, make integer dup more expensive.
14191 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14193 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14194 Return true for non-tls symbols.
14196 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14198 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14199 (cortex-a75): Likewise.
14200 (cortex-a75.cortex-a55): Likewise.
14201 * config/aarch64/aarch64-tune.md: Regenerate.
14202 * doc/invoke.texi (-mtune): Document new values for -mtune.
14204 2017-06-21 Tom de Vries <tom@codesourcery.com>
14206 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14207 stack_size feature.
14208 (Effective-Target Keywords, Other attributes): Suggest using
14209 dg-add-options stack_size feature to get stack limit in stack_size
14210 effective target documentation.
14212 2017-06-21 Julian Brown <julian@codesourcery.com>
14213 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14215 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14216 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14217 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14219 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14220 attribute type list for neon_multiply.
14221 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14222 attribute type list for neon_multiply.
14223 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14224 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14225 attribute type list for neon_multiply.
14226 * config/arm/types.md (crypto_pmull): Add.
14227 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14228 attribute type list.
14230 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14232 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14235 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14237 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14238 to make sure not to dereference a NULL cost_classes_ptr pointer.
14240 2017-06-20 Carl Love <cel@us.ibm.com>
14242 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14243 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14244 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14245 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14246 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14247 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14248 VMULOSW): New enum "unspec" values.
14249 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14250 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14251 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14252 altivec_vmulosw): New patterns.
14253 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14254 VMULOSW): Add definitions.
14256 2017-06-20 Julia Koval <julia.koval@intel.com>
14258 * config/i386/i386.c: Fix rounding expand for new pattern.
14259 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14261 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14263 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14264 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14266 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14268 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14271 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14273 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14274 architecture, then by implementer ID.
14275 * config/aarch64/aarch64-tune.md: Regenerate.
14277 2017-06-20 Richard Biener <rguenther@suse.de>
14279 PR middle-end/81097
14280 * fold-const.c (split_tree): Fold to type before negating.
14282 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14284 * diagnostic-show-locus.c
14285 (selftest::test_fixit_deletion_affecting_newline): New function.
14286 (selftest::diagnostic_show_locus_c_tests): Call it.
14288 2017-06-20 Andreas Schwab <schwab@suse.de>
14291 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14294 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14296 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14297 __ARM_FEATURE_COPROC according to support.
14299 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14301 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14302 Rewritten to avoid overflow for > 32-bit pointers.
14305 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14306 by removing enum keyword.
14307 (ubsan_type_descriptor): Likewise. Formatting fix.
14310 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14311 splitter): Require TARGET_SSE2 in the condition.
14313 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14316 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14317 for doing vector set of SFmode on ISA 3.0.
14318 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14319 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14321 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14322 SFmode value into a V4SF variable that was extracted from another
14323 V4SF variable without converting the element to double precision
14324 and back to single precision vector format.
14325 (vsx_insert_extract_v4sf_p9_2): Likewise.
14327 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14329 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14330 in UWHI to avoid undefined overflow.
14333 * ubsan.h (enum ubsan_encode_value_phase): New.
14334 (ubsan_encode_value): Change second argument to
14335 enum ubsan_encode_value_phase with default value of
14336 UBSAN_ENCODE_VALUE_GENERIC.
14337 * ubsan.c (ubsan_encode_value): Change second argument to
14338 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14339 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14340 create_tmp_var_raw instead of create_tmp_var and use a
14342 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14343 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14344 ubsan_encode_value callers.
14347 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14348 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14349 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14351 2017-06-19 Richard Biener <rguenther@suse.de>
14353 PR middle-end/81118
14354 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14355 estimates if we changed anything.
14357 2017-06-19 Richard Biener <rguenther@suse.de>
14359 PR tree-optimization/80887
14360 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14361 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14362 simplified lookups, then reset mprts_hook.
14363 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14365 (try_to_simplify): Likewise.
14367 2017-06-19 Martin Liska <mliska@suse.cz>
14370 * gimplify.c (gimplify_switch_expr):
14371 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14373 2017-06-19 Martin Liska <mliska@suse.cz>
14375 * doc/install.texi: Document that PGO runs in 4 stages.
14377 2017-06-19 Martin Liska <mliska@suse.cz>
14380 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14381 to dispatcher function name.
14382 * multiple_target.c (replace_function_decl): New function.
14383 (create_dispatcher_calls): Redirect both edges and references.
14385 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14387 * profile-count.c (profile_count::dump): Dump quality.
14388 (profile_count::differs_from_p): Update for unsigned val.
14389 * profile-count.h (profile_count_quality): New enum.
14390 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14392 2017-06-19 Richard Biener <rguenther@suse.de>
14394 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14395 struct function as arg.
14396 (estimate_numbers_of_iterations): Export overload with loop arg.
14397 (free_numbers_of_iterations_estimates_loop): Use an overload of
14398 free_numbers_of_iterations_estimates instead.
14399 * tree-cfg.c (remove_bb): Adjust.
14400 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14401 * tree-parloops.c (gen_parallel_loop): Likewise.
14402 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14404 (tree_unroll_loops_completely): Likewise.
14405 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14406 Use an overload instead and export.
14407 (estimated_loop_iterations): Adjust.
14408 (max_loop_iterations): Likewise.
14409 (likely_max_loop_iterations): Likewise.
14410 (estimate_numbers_of_iterations): Take struct function as arg
14412 (loop_exits_before_overflow): Adjust.
14413 (free_numbers_of_iterations_estimates_loop): Use an overload.
14414 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14415 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14417 2017-06-19 Richard Biener <rguenther@suse.de>
14420 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14421 RANGE_EXPR conservatively.
14423 2017-06-16 Carl Love <cel@us.ibm.com>
14425 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14426 definitions for vec_float, vec_float2, vec_floato,
14427 vec_floate built-ins.
14428 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14429 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14431 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14432 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14433 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14434 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14435 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14436 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14437 vec_floato): Add builtin defines.
14438 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14439 Update the built-in documentation file for the new built-in
14442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14444 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14445 (mthumb): Mark as the negative of -marm.
14447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14449 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14451 (ARM Options, -mtune): Document that this accepts the same
14452 extension options as -mcpu.
14453 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14457 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14458 permitted extensions.
14460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14462 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14463 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14464 (armv8-m.main): Add option +nodsp.
14465 * config/arm/arm-cpu-cdata.h: Regenerated.
14467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14469 * config/arm/t-fuchsia: New file.
14470 * config.gcc (arm*-*-fuchsia*): Use it.
14472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14474 * config/arm/t-symbian: Rewrite for new option infrastructure.
14476 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14478 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14479 (MULTILIB_REQUIRED): Likewise.
14481 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14483 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14484 (MULTILIB_RESUE): Likewise.
14485 (MULTILIB_MATCHES): Likewise.
14486 (MULTLIB_REQUIRED): Likewise.
14488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14490 * config/arm/t-rtems: Rewrite for new option framework.
14492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14494 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14495 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14496 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14497 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14498 * config/arm/t-multilib: ... here.
14499 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14500 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14501 armv7-a and armv8*-a when A-profile libraries have not been built.
14502 * config/arm/t-rmprofile: Rewrite.
14504 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14506 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14507 with a backslash. Remove the backslash after substituting unescaped
14509 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14511 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14513 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14514 the driver to pass through the default setting of -mfloat-abi.
14515 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14517 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14518 (all_feat_combs): New rule.
14519 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14521 * config/arm/t-aprofile: Rewrite.
14523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14525 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14526 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14527 fpu is not specified by the user/command-line.
14528 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14529 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14530 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14531 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14532 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14533 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14537 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14538 * config/arm/t-arm-elf: Rewritten.
14540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14542 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14543 have some floating-point instructions.
14544 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14545 (TARGET_MAYBE_HARD_FLOAT): New macro.
14546 * config/arm/arm-builtins.c (arm_init_builtins): Use
14547 TARGET_MAYBE_HARD_FLOAT.
14548 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14550 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14552 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14553 (configargs.h): Include it.
14554 (arm_print_hint_for_fpu_option): New function.
14555 (arm_parse_fpu_option): New function.
14556 (candidate_extension): New class.
14557 (arm_canon_for_multilib): New function.
14558 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14559 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14560 (ARCH_CANONICAL_SPECS): New macro.
14561 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14565 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14566 are set after handling multilib fragments. Set target_cpu_default2
14569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14571 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14573 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14574 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14576 (check_arch): Likewise.
14577 * config/arm/arm.c (arm_configure_build_target): Handle
14578 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14579 options in the default.
14581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14583 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14584 when an option is an alias of another.
14585 * config/arm/parsecpu.awk (optalias): New parser token.
14586 (gen_comm_data): Mark non-alias options as such. Emit entries
14587 for extension aliases.
14588 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14589 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14590 (armv6kz, armv6zk, armv6t2): Likewise.
14591 (armv7): Make vfpv3-d16 an alias.
14592 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14594 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14595 Sort in canonical order.
14596 (armv8-a): Sort in canonical order.
14597 (armv8.1-a, armv8.2-a): Likewise.
14598 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14600 (cortex-a9): Sort in canonical order.
14601 * config/arm/arm.c (selftests.h): Include it.
14602 (arm_test_cpu_arch_data): New function.
14603 (arm_run_self_tests): New function.
14604 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14605 (targetm): Move declaration to the end of the file.
14606 * arm-cpu-cdata.h: Regenerated.
14608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14610 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14611 call to target_mode_check describing the type of option passed.
14612 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14613 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14614 arm_parse_cpu_option_name to match parameters against list of
14616 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14617 arm_arch_core_flags data structure.
14618 * config/arm/arm-cpu_cdata.h: Regenerated.
14620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14622 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14624 (arm_print_hint_for_cpu_option): Likewise.
14625 (arm_print_hint_for_arch_option): Likewise.
14626 (arm_parse_cpu_option_name): Likewise.
14627 (arm_parse_arch_option_name): Likewise.
14628 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14629 of entries in the all_fpus list.
14630 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14631 (arm_parse_cpu_option_name): Declare.
14632 (arm_parse_arch_option_name): Declare.
14633 (arm_parse_option_features): Declare.
14634 (arm_intialize_isa): Declare.
14635 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14637 (gen_comm_data): ... here. Make definitions non-static.
14638 * config/arm/arm-cpu-data.h: Regenerated.
14639 * config/arm/arm-cpu-cdata.h: Regenerated.
14641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14643 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14644 (cpu_arch_extension): New structure.
14645 (cpu_arch_option, arch_option, cpu_option): New structures.
14646 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14647 architecture types.
14648 (gen_data): Generate new format data tables.
14649 * config/arm/arm.c (cpu_tune): New structure.
14650 (cpu_option, processors): Delete.
14651 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14652 (arm_print_hint_for_cpu_option): ... this and ...
14653 (arm_print_hint_for_arch_option): ... this.
14654 (arm_parse_arch_cpu_name): Delete. Replace with ...
14655 (arm_parse_cpu_option_name): ... this and ...
14656 (arm_parse_arch_option_name): ... this.
14657 (arm_unrecognized_feature): Change type of target parameter to
14659 (arm_parse_arch_cpu_features): Delete. Replace with ...
14660 (arm_parse_option_features): ... this.
14661 (arm_configure_build_target): Rework to use new configuration data
14663 (arm_print_tune_info): Rework for new configuration data tables.
14664 * config/arm/arm-cpu-data.h: Regenerated.
14665 * config/arm/arm-cpu.h: Regenerated.
14667 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14669 * Makefile.in (OBJS): Move sbitmap.o from here ...
14670 (OBJS-libcommon): ... to here.
14672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14674 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14675 (ISA_ALL_CRYPTO): New macro.
14676 (ISA_ALL_SIMD): New macro
14677 (ISA_ALL_FP): New macro.
14678 * config/arm/arm.c (fpu_bitlist): Update initializer.
14679 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14681 (arm9e): Add fpu. Add option for nofp
14682 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14683 (arm926ej-s, arm1026ej-s): Likewise.
14684 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14685 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14686 neon-fp16, neon-vfpv4, nofp and nosimd.
14687 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14688 (cortex-a8): Add fpu. Add option for nofp.
14689 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14690 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14691 (cortex-r4f): Add fpu.
14692 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14693 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14695 (cortex-r8): Likewise.
14696 (cortex-m4): Add fpu. Add option for nofp.
14697 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14698 (cortex-a17.cortex-a7): Likewise.
14699 (cortex-a32): Add fpu. Add options for crypto and nofp.
14700 (cortex-a35, cortex-a53): Likewise.
14701 (cortex-a57): Add fpu. Add option for crypto.
14702 (cortex-a72, cortex-a73): Likewise.
14703 (exynos-m1): Likewise.
14704 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14705 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14706 (cortex-m33): Add fpu. Add option for nofp.
14707 * config/arm/arm-cpu-cdata.h: Regenerated
14708 * config/arm/arm-cpu-data.h: Regenerated.
14710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14712 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14713 (armv5te, armv5tej): Likewise.
14714 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14715 (armv7): Add options fp and vfpv3-d16.
14716 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14717 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14719 (armv7ve): Likewise.
14720 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14721 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14722 (armv8-a): Add nocrypto option.
14723 (armv8.1-a, armv8.2-a): Likewise.
14724 (armv8-m.main): add options fp, fp.dp and nofp.
14726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14728 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14730 (armv8-a+crc): Delete.
14731 (armv8.1-a): Add options simd, crypto and nofp.
14732 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14733 (armv8.2-a+fp16): Delete.
14734 (armv8-m.main): Add option dsp.
14735 (armv8-m.main+dsp): Delete.
14736 (cortex-a8): Add fpu. Add nofp option.
14737 (cortex-a9): Add fpu. Add nofp and nosimd options.
14738 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14739 link to main cpu and architecture data structures.
14740 (gen_comm_data): Only put isa attributes from the main architecture
14742 (option): New statement for architecture and CPU entries.
14743 * arm.c (struct cpu_option): New structure.
14744 (struct processors): Add entry for options.
14745 (arm_unrecognized_feature): New function.
14746 (arm_parse_arch_cpu_name): Ignore any characters after the first
14748 (arm_parse_arch_cpu_feature): New function.
14749 (arm_configure_build_target): Separate out any CPU and architecture
14750 features and parse separately. Don't error out if -mfpu=auto is
14751 used with only an architecture string.
14752 (arm_print_asm_arch_directives): New function.
14753 (arm_file_start): Call it.
14754 * config/arm/arm-cpu-cdata.h: Regenerated.
14755 * config/arm/arm-cpu-data.h: Likewise.
14756 * config/arm/arm-tables.opt: Likewise.
14758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14760 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14761 assembler when it is not -mfpu=auto.
14763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14765 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14766 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14767 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14768 (ASM_CPU_SPEC): Rewrite.
14769 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14770 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14771 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14772 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14773 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14774 copied string is NUL-terminated. Also strip any characters prefixed
14776 (arm_rewrite_selected_arch): New function.
14777 (arm_rewrite_march): New function.
14779 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14781 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14782 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14783 (march, mcpu, mtune): Convert to string-based options.
14784 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14785 (arm_parse_arch_cpu_name): New function.
14786 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14787 identify selected architecture or CPU.
14788 (arm_option_save): New function.
14789 (TARGET_OPTION_SAVE): Redefine.
14790 (arm_option_restore): Restore string options.
14791 (arm_option_print): Print string options.
14793 2017-06-16 Martin Sebor <msebor@redhat.com>
14795 PR tree-optimization/80933
14796 PR tree-optimization/80934
14797 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14798 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14799 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14800 (gimple_fold_builtin): Call them.
14802 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14804 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14805 as unlikely; update profile.
14807 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14809 * predict.c (force_edge_cold): Handle declaring edges impossible
14812 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14814 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14816 (try_unroll_loop_completely): Fix reporting.
14818 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14820 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14822 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14825 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14826 if given a non-constant argument for an intrinsic which requires a
14829 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14831 * profile.c (compare_freqs): New function.
14832 (branch_prob): Sort edge list.
14833 (find_spanning_tree): Assume that the list is priority sorted.
14835 2017-06-16 Richard Biener <rguenther@suse.de>
14837 PR tree-optimization/81090
14838 * passes.def (pass_record_bounds): Remove.
14839 * tree-pass.h (make_pass_record_bounds): Likewise.
14840 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
14841 make_pass_record_bounds): Likewise.
14842 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
14843 not free niter estimates at the beginning but at the end.
14844 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
14846 2017-06-16 Richard Biener <rguenther@suse.de>
14848 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
14849 initializer to workaround ICE in host GCC 4.8.
14851 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14853 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
14855 (clone_inlined_nodes): Update.
14857 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14859 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
14860 prefetch settings, and enable prefetching by default at -O3.
14862 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14864 * config/aarch64/aarch64.c (aarch64_override_options_internal):
14865 Set flag_prefetch_loop_arrays according to tuning data.
14867 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
14869 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
14870 New tune structure.
14871 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
14872 [Unrelated to main purpose of the patch] Place the pointer field last
14873 to enable type checking errors when tune structure are wrongly merged.
14874 * config/aarch64/aarch64.c (generic_prefetch_tune,)
14875 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
14876 (thunderx2t99_prefetch_tune): New tune constants.
14877 (tune_params *_tunings): Update all tunings (no functional change).
14878 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
14879 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
14880 from tunings structures.
14882 2017-06-16 Jakub Jelinek <jakub@redhat.com>
14885 * ubsan.c (instrument_null): Add T argument, use it instead
14886 of computing it based on IS_LHS.
14887 (instrument_object_size): Likewise.
14888 (pass_ubsan::execute): Adjust instrument_null and
14889 instrument_object_size callers to pass gimple_get_lhs or
14890 gimple_assign_rhs1 result to it. Use instrument_null instead of
14891 calling get_base_address and instrument_mem_ref. Handle
14892 aggregate call arguments for object-size sanitization.
14894 2017-06-16 Yury Gribov <tetra2005@gmail.com>
14896 PR tree-optimization/81089
14897 * tree-vrp.c (is_masked_range_test): Validate operands of
14900 2017-06-15 Martin Sebor <msebor@redhat.com>
14903 * dumpfile.c (dump_register): Avoid calling memset to initialize
14904 a class with a default ctor.
14905 * gcc.c (struct compiler): Remove const qualification.
14906 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
14907 * hash-table.h: Ditto.
14908 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
14910 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
14911 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
14913 * params.h (struct param_info): Make struct members non-const.
14914 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
14915 with default initialization.
14916 * vec.h (vec_copy_construct, vec_default_construct): New helper
14918 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
14919 with vec_copy_construct.
14920 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
14921 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
14922 * doc/invoke.texi (-Wclass-memaccess): Document.
14924 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14926 * emit-rtl.h (is_leaf): Update comment about local
14927 register allocator.
14929 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14932 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
14933 for a variable to have a section before checking if the section has a
14935 Set section to.persistent if persistent attribute is set.
14936 Warn if .persistent attribute is used on an automatic variable.
14938 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
14940 PR rtl-optimization/80474
14941 * reorg.c (update_block): Do not ignore instructions in a delay slot.
14943 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
14945 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
14948 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
14950 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
14951 (casesi): Emit bounds checking as RTL.
14952 (casesi_internal_mips16_<mode>): Remove bounds checking.
14954 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
14956 * config/xtensa/xtensa.c (xtensa_option_override): Append
14957 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
14958 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
14959 xtensa_doloop_hooks): Define unconditionally.
14960 (xtensa_reorg_loops): Only call reorg_loops in the presence of
14962 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
14963 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
14964 for it in xtensa_option_override.
14965 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
14966 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
14968 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
14970 * doc/cppopts.texi: Document '-' special value to -MF.
14972 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
14974 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
14975 (cortex_a53_fconst): Likewise.
14976 (cortex_a53_fpmul): Likewise.
14977 (cortex_a53_f_load_64): Likewise.
14978 (cortex_a53_f_load_many): Likewise.
14979 (cortex_a53_advsimd_alu): Likewise.
14980 (cortex_a53_advsimd_alu_q): Likewise.
14981 (cortex_a53_advsimd_mul): Likewise.
14982 (cortex_a53_advsimd_mul_q): Likewise.
14983 (fpmac bypass): Add new bypass for fpmac-fpmac case.
14984 Add missing fmul, r2f_cvt and fconst cases.
14986 2017-06-14 Richard Biener <rguenther@suse.de>
14988 PR middle-end/81088
14989 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
14991 (fold_binary_loc): When associating do not treat pre-existing
14992 TREE_OVERFLOW on literal constants as a reason to allow
14993 TREE_OVERFLOW on associated literal constants.
14995 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
14997 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
14998 (MASK_FEATURES): New macro.
14999 * config/sparc/sparc.c (sparc_option_override): Remove the special
15000 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15002 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15004 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15005 a division of 0 if non-call exceptions are enabled.
15007 2017-06-14 Andrew Pinski <apinski@cavium.com>
15008 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15011 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15012 Improve vector initialization code gen for only variable case.
15014 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15016 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15018 2017-06-14 Richard Biener <rguenther@suse.de>
15020 PR tree-optimization/81083
15021 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15024 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15026 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15027 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15028 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15029 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15030 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15031 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15033 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15035 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15036 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15038 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15040 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15042 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15044 * config/rs6000/t-rtems: Don't handle SPE.
15046 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15048 * config/rs6000/t-linux: Don't handle SPE.
15050 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15052 * config/rs6000/eabispe.h: Delete file.
15054 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15056 * config/rs6000/t-spe: Delete file.
15058 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15060 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15061 (rs6000_legitimate_offset_address_p): Return false for anything in
15062 V2SImode or V2SFmode.
15064 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15066 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15067 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15068 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15070 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15071 (rs6000_legitimate_offset_address_p): Ditto.
15072 (rs6000_emit_move): Ditto.
15073 (rs6000_init_builtins): Remove V4HI_type_node.
15075 2017-06-13 Martin Liska <mliska@suse.cz>
15078 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15079 (gate_asan): Likewise.
15080 * asan.h (asan_no_sanitize_address_p): Remove the function.
15081 (sanitize_flags_p): New function.
15082 * builtins.def: Fix coding style.
15083 * common.opt: Use renamed enum value.
15084 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15085 * doc/extend.texi: Document no_sanitize attribute.
15086 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15087 to SANITIZE_UNDEFINED_NONDEFAULT.
15088 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15089 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15090 Use sanitize_flags_p.
15091 * gimplify.c (gimplify_function_tree): Likewise.
15092 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15093 * opts.c (parse_no_sanitize_attribute): New function.
15094 (common_handle_option): Use renamed enum value.
15095 * opts.h (parse_no_sanitize_attribute): Declare.
15096 * tree.c (sanitize_flags_p): New function.
15097 * tree.h: Declared here.
15098 * tsan.c: Use sanitize_flags_p.
15099 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15100 (instrument_mem_ref): Likewise.
15101 (instrument_bool_enum_load): Likewise.
15102 (do_ubsan_in_current_function): Remove the function.
15103 (pass_ubsan::execute): Use sanitize_flags_p.
15104 * ubsan.h: Remove do_ubsan_in_current_function
15105 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15106 (dump_function_to_file): Use it here.
15108 2017-06-13 Martin Jambor <mjambor@suse.cz>
15110 PR tree-optimization/80803
15111 PR tree-optimization/81063
15112 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15113 (propagate_subaccesses_across_link): Enqueue subtree whenever
15114 necessary instead of relying on the caller.
15116 2017-06-13 Martin Jambor <mjambor@suse.cz>
15118 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15119 that have a first_link.
15120 (sort_and_splice_var_accesses): Do not check first_link before
15122 (subtree_mark_written_and_enqueue): Likewise.
15123 (propagate_all_subaccesses): Likewise and do not stop at first
15124 parent with a first_link.
15126 2017-06-13 Martin Jambor <mjambor@suse.cz>
15128 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15131 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15133 * match.pd: New pattern.
15135 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15137 * tree-vrp.c (is_masked_range_test): New function.
15138 (register_edge_assert_for): Determine ranges for
15141 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15143 PR tree-optimization/67328
15144 * fold-const.c (maskable_range_p): New function.
15145 (build_range_check): Generate bittests if possible.
15147 2017-06-13 Martin Liska <mliska@suse.cz>
15149 * gimple-pretty-print.c (dump_probability): Add new argument.
15150 (dump_edge_probability): Dump both probability and count.
15151 (dump_gimple_label): Likewise.
15152 (dump_gimple_bb_header): Likewise.
15154 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15157 * config/avr/avr-devices.c: Fix indentation.
15158 * config/avr/gen-avr-mmcu-specs.c: Dito.
15160 2017-06-13 Richard Biener <rguenther@suse.de>
15162 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15163 instead get vector type from stmt_info.
15164 (vectorizable_reduction): Adjust. Remove dead code.
15166 2017-06-13 Richard Biener <rguenther@suse.de>
15168 PR middle-end/81065
15169 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15170 case of C * (x * C2 + C3).
15171 (fold_addr_of_array_ref_difference): Properly fold index difference.
15173 2017-06-12 David S. Miller <davem@davemloft.net>
15176 * config/sparc/sparc.md (return expander): Emit frame blockage if
15177 function uses alloca.
15179 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15181 * combine.c (make_field_assignment): Check len rather than the mode
15182 precision when calling force_to_mode.
15184 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15186 Support multilibs and devices that see flash in RAM address range.
15189 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15190 (avr_mcu_t) <flash_pm_offset>: New field.
15191 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15192 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15193 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15194 (AVR_TINY_PM_OFFSET): Remove macro.
15195 * config/avr/avr.opt (-mshort-calls): New option.
15196 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15197 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15198 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15199 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15200 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15201 instead of avr_arch->have_jmp_call.
15202 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15203 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15204 avr_arch->flash_pm_offset to define.
15205 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15206 new field flash_pm_offset. Add entry for avrxmega3.
15207 (avr_texinfo): Add entry for avrxmega3.
15208 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15209 attiny212, attiny214,
15210 attiny412, attiny414, attiny416, attiny417,
15211 attiny814, attiny816, attiny817,
15212 attiny1614, attiny1616, attiny1617,
15213 attiny3214, attiny3216, attiny3217.
15214 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15215 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15216 (avr_print_operand_address) [AVR_TINY]: Same.
15217 (avr_asm_init_sections) <readonly_data_section>: Only patch
15218 callback if avr_arch->flash_pm_offset = 0.
15219 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15220 for rodata if avr_arch->flash_pm_offset != 0.
15221 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15222 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15223 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15224 (m_options): Append opt_rcall.
15225 (m_dirnames): Append dir_rcall.
15226 * config/avr/t-multilib: Regenerate.
15228 * configure.ac [target=avr]: Check whether avrxmega3 default
15229 linker description file works as needed.
15230 * configure: Regenerate.
15231 * doc/avr-mmcu.texi: Regenerate.
15232 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15233 <__AVR_ARCH__>: Document avrxmega3 and 103.
15234 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15235 <__AVR_SHORT_CALLS__>: Document it.
15236 <__AVR_PM_BASE_ADDRESS__>: Document it.
15237 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15238 (AVR Variable Attributes) <progmem>: Document this is
15239 not needed for avrxmega3.
15240 (AVR Named Address Spaces) <__flash>: Dito.
15242 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15244 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15246 2017-06-12 Doug Rupp <rupp@adacore.com>
15248 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15249 Append vxworks-stdint.h to the tm_file list.
15250 * config/vxworks-stdint.h: New file.
15252 2017-06-12 Martin Liska <mliska@suse.cz>
15254 PR tree-optimization/81041
15255 * tree-profile.c (gimple_gen_ic_func_profiler):
15256 Create an extra BB in profile-generate
15257 (gimple_gen_time_profiler): Likewise.
15259 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15261 PR tree-optimization/81003
15262 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15263 (update_range_test): Use it instead of force_gimple_operand_gsi.
15265 2017-06-12 Richard Biener <rguenther@suse.de>
15267 PR tree-optimization/81053
15268 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15269 with backedge value not defined in loop. Simplify def stmt
15272 2017-06-11 Tom de Vries <tom@codesourcery.com>
15275 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15277 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15278 nvptx_cannot_force_const_mem.
15280 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15282 * opts.c (finish_options): Move test for flag_split_stack after
15283 it has been initialized.
15285 2017-06-11 Jason Merrill <jason@redhat.com>
15287 * tree.h (id_equal): New.
15288 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15289 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15290 instead of strcmp of IDENTIFIER_POINTER.
15292 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15294 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15295 (mark_all_inlined_calls_cdtor): Fix formating.
15296 (inline_transform): Rescale profile before inlining.
15298 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15300 * cgraph.h (cgraph_edge::clone): Update prototype.
15301 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15302 (cgraph_node::create_clone): Update.
15303 (cgraph_node::create_version_clone): Update.
15304 * tree-inline.c (copy_bb): Update.
15305 (expand_call_inline): Update.
15307 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15309 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15310 factored out from ...
15311 (rs6000_emit_prologue): ... here.
15313 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15315 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15316 factored out from ...
15317 (rs6000_emit_prologue): ... here.
15319 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15321 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15323 (handle_missing_profiles): Fix computation of tp_first_run.
15324 (counts_to_freqs): Do not touch freqs when count is 0.
15326 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15328 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15331 2017-06-10 Tom de Vries <tom@codesourcery.com>
15333 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15334 attributes): Document signal effective target.
15336 2017-06-10 Tom de Vries <tom@codesourcery.com>
15338 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15339 Document effective target stack_size.
15341 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15343 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15344 to the edit_context if they can be auto-applied.
15346 2017-06-9 Ian Lance Taylor <iant@golang.org>
15348 * opts.c (finish_options): If -fsplit-stack, disable implicit
15349 -forder-blocks-and-partition.
15350 * doc/invoke.texi (Optimize Options): Document that when using
15351 -fsplit-stack -forder-blocks-and-partition is not implicitly
15354 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15356 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15357 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15358 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15359 * builtins.def (abort, trap, unreachable): Declare cold.
15360 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15361 * tree-core.h (ECF_COLD): New.
15362 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15363 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15365 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15367 * predict.c (unlikely_executed_stmt_p): Cleanup.
15369 2017-06-09 Richard Biener <rguenther@suse.de>
15371 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15372 model if the ref is always written to.
15374 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15376 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15378 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15380 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15383 2017-06-09 Tom de Vries <tom@codesourcery.com>
15386 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15387 "target cannot support label values" when encountering LABEL_REF.
15389 2017-06-09 Martin Liska <mliska@suse.cz>
15391 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15392 (gimple_gen_ic_func_profiler): Emit direct comparison
15393 of __gcov_indirect_call_callee with NULL.
15394 (gimple_gen_time_profiler): Change probability from
15395 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15397 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15399 * profile.c (edge_gcov_counts): Turn to pointer.
15400 (compute_branch_probabilities, compute_branch_probabilities): Update.
15401 (branch_prob): Do not clear edge_gcov_count.
15402 * profile.h (edge_gcov_counts): Turn to pointer.
15403 (edge_gcov_count): Update.
15405 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15407 * gimple.h (gimple_check_failed): Mark cold.
15409 2017-06-09 Richard Biener <rguenther@suse.de>
15411 PR tree-optimization/66623
15412 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15413 refactor check_reduction into two parts, properly computing
15414 whether we have to check reduction validity for outer loop
15417 2017-06-09 Richard Biener <rguenther@suse.de>
15419 PR tree-optimization/79483
15420 * graphite-scop-detection.c (order): New global.
15421 (get_order): Compute bb to order mapping that satisfies code
15422 generation constraints.
15423 (cmp_pbbs): New helper.
15424 (build_scops): Start domwalk at entry block, sort generated
15427 2017-06-09 Richard Biener <rguenther@suse.de>
15429 PR middle-end/81007
15430 * ipa-polymorphic-call.c
15431 (ipa_polymorphic_call_context::restrict_to_inner_class):
15432 Skip FIELD_DECLs with error_mark_node type.
15433 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15436 2017-06-09 Martin Liska <mliska@suse.cz>
15438 * predict.c (struct branch_predictor): New struct.
15439 (test_prediction_value_range): New test.
15440 (predict_c_tests): New function.
15441 * selftest-run-tests.c (selftest::run_tests): Run the function.
15442 * selftest.h: Declare new tests.
15444 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15447 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15448 gen_add3_insn did not fail.
15449 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15450 r0, construct that number in a temporary reg and add that reg to r0.
15451 If asked to put the result in r0 as well, fail.
15453 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15455 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15456 for early expansion of vec_eqv.
15458 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15460 PR middle-end/81005
15461 * ubsan.c (instrument_null): Avoid pointless code temporary.
15462 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15464 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15468 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15471 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15474 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15476 * predict.c (unlikely_executed_edge_p): Move ahead.
15477 (probably_never_executed_edge_p): Use it.
15479 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15481 PR middle-end/79988
15482 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15483 gimple_call_builtin_p call.
15485 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15487 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15488 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15489 rtl_check_failed_type2, rtl_check_failed_code1,
15490 rtl_check_failed_code2, rtl_check_failed_code_mode,
15491 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15492 rtvec_check_failed_bounds, rtl_check_failed_flag,
15493 _fatal_insn_not_found, _fatal_insn): Likewise.
15494 * tree.h (tree_contains_struct_check_failed,
15495 tree_check_failed, tree_not_check_failed,
15496 tree_class_check_failed, tree_range_check_failed,
15497 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15498 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15499 tree_operand_check_failed, omp_clause_check_failed,
15500 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15503 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15505 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15506 flag_branch_probabilities.
15507 * ipa-inline.c (edge_badness): Likewise.
15508 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15509 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15510 * predict.c (maybe_hot_frequency_p): Likewise.
15511 (probably_never_executed): Likewise.
15512 * sched-ebb.c (schedule_ebbs): Likewise.
15513 * sched-rgn.c (find_single_block_region): Likewise.
15514 * tracer.c (tail_duplicate): Likewise.
15516 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15518 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15519 longer requires x_flag_profile_use.
15521 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15523 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15524 instead of flag_reorder_blocks_and_partition.
15525 * dbxout.c (dbxout_function_end): Likewise.
15526 * dwarf2out.c (gen_subprogram_die): Likewise.
15527 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15528 * hw-doloop.c (reorg_loops): Likewise.
15529 * varasm.c (assemble_start_function,
15530 assemble_end_function): Likewise.
15531 (decide_function_section): Do not check for
15532 flag_reorder_blocks_and_partition.
15534 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15536 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15538 (chkp_get_hard_register_fake_addr_expr): Ditto.
15539 (chkp_build_addr_expr): Add check for hard reg case.
15540 (chkp_parse_array_and_component_ref): Ditto.
15541 (chkp_find_bounds_1): Ditto.
15542 (chkp_process_stmt): Don't generate bounds store for
15545 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15547 * predict.c (maybe_hot_bb_p): Do not check profile status.
15548 (maybe_hot_edge_p): Likewise.
15549 (probably_never_executed): Check for zero counts even if profile
15551 (unlikely_executed_edge_p): New function.
15552 (unlikely_executed_stmt_p): New function.
15553 (unlikely_executed_bb_p): New function.
15554 (set_even_probabilities): Use unlikely predicates.
15555 (combine_predictions_for_bb): Likewise.
15556 (predict_paths_for_bb): Likewise.
15557 (predict_paths_leading_to_edge): Likewise.
15558 (determine_unlikely_bbs): New function.
15559 (estimate_bb_frequencies): Use it.
15560 (compute_function_frequency): Use zero counts even if profile is
15562 * profile-count.h: Fix typo.
15564 2017-08-08 Julia Koval <julia.koval@intel.com>
15566 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15567 _mm512_mask_cvtsepi16_storeu_epi8,
15568 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15569 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15570 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15571 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15572 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15573 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15574 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15575 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15576 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15577 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15578 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15579 __builtin_ia32_pmovuswb256mem_mask,
15580 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15581 __builtin_ia32_pmovwb512mem_mask): New builtins.
15583 2017-08-08 Julia Koval <julia.koval@intel.com>
15585 PR target/73350,80862
15586 * config/i386/subst.md (round): Fix round pattern.
15587 * config/i386/i386.c (ix86_erase_embedded_rounding):
15588 Fix erasing rounding for the fixed pattern.
15590 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15592 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15594 2017-06-08 Martin Liska <mliska@suse.cz>
15596 PR gcov-profile/80911
15597 * gcov.c (block_info::block_info): New constructor.
15599 2017-06-07 Carl Love <cel@us.ibm.com>
15601 * config/rs6000/rs6000-c: The return type of the following
15602 built-in functions was implemented as int not long long. Fix sign
15603 of return value for the unsigned version of vec_mulo and vec_mule.
15604 vector unsigned long long vec_bperm (vector unsigned long long,
15605 vector unsigned char)
15606 vector signed long long vec_mule (vector signed int,
15608 vector unsigned long long vec_mule (vector unsigned int,
15609 vector unsigned int)
15610 vector signed long long vec_mulo (vector signed int,
15612 vector unsigned long long vec_mulo (vector unsigned int,
15613 vector unsigned int)
15614 * doc/extend.texi: Fix the documentation for the built-in
15617 2017-06-07 Carl Love <cel@us.ibm.com>
15620 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15623 2017-06-07 Carl Love <cel@us.ibm.com>
15625 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15626 support, Generate doublehv for signed int/float for BE case only.
15628 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15630 * doc/invoke.texi (mcx16): Rewrite.
15632 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15634 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15635 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15636 *mov<mode>_softfloat, and an anonymous splitter): Use
15637 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15639 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15641 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15643 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15644 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15645 (rs6000_debug_reg_global): Adjust.
15646 (rs6000_init_hard_regno_mode_ok): Adjust.
15647 (rs6000_dbx_register_number): Adjust.
15648 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15649 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15650 Remove SPE_ACC and SPEFSCR.
15651 (REG_ALLOC_ORDER): Ditto.
15652 (FRAME_POINTER_REGNUM): Change to 111.
15653 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15654 (REG_CLASS_NAMES): Ditto.
15655 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15656 (REGISTER_NAMES): Ditto.
15657 (ADDITIONAL_REG_NAMES): Ditto.
15658 (rs6000_reg_names): Ditto.
15659 * config/rs6000/rs6000.md: Renumber some register number
15662 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15664 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15666 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15667 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15669 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15670 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15671 Delete the SPE high registers.
15672 (REG_ALLOC_ORDER): Ditto.
15673 (enum reg_class): Remove SPE_HIGH_REGS.
15674 (REG_CLASS_NAMES): Ditto.
15675 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15676 (REGISTER_NAMES): Ditto.
15677 (rs6000_reg_names): Ditto.
15678 * doc/tm.texi.in: Remove SPE as example.
15679 * doc/tm.texi: Regenerate.
15681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15683 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15684 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15685 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15686 * config/rs6000/rs6000.md (type): Remove "brinc".
15688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15690 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15691 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15692 * config/rs6000/linuxspe.h: Delete file.
15693 * config/rs6000/rs6000.md: Don't include spe.md.
15694 * config/rs6000/spe.h: Delete file.
15695 * config/rs6000/spe.md: Delete file.
15696 * config/rs6000/t-rs6000: Remove spe.md.
15698 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15700 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15701 (reg_or_none500mem_operand): Delete.
15702 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15703 instead of reg_or_none500mem_operand.
15705 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15707 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15708 handling of SPE flags.
15709 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15711 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15713 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15715 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15716 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15717 paired_divv2sf3): Similar.
15718 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15719 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15720 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15722 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15723 Rename the paired_* instruction patterns.
15724 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15726 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15727 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15728 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15729 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15730 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15731 PAIRED_VECTOR_MODE.
15732 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15733 (spe_func_has_64bit_regs_p): Delete.
15734 (spe_expand_predicate_builtin): Delete.
15735 (spe_expand_evsel_builtin): Delete.
15736 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15737 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15738 (invalid_e500_subreg): Delete.
15739 (rs6000_legitimize_address): Always force_reg op2 as well, for
15740 paired single memory accesses.
15741 (rs6000_member_type_forces_blk): Delete.
15742 (rs6000_spe_function_arg): Delete.
15743 (rs6000_expand_unop_builtin): Delete SPE handling.
15744 (rs6000_expand_binop_builtin): Ditto.
15745 (spe_expand_stv_builtin): Delete.
15746 (bdesc_2arg_spe): Delete.
15747 (spe_expand_builtin): Delete.
15748 (spe_expand_predicate_builtin): Delete.
15749 (spe_expand_evsel_builtin): Delete.
15750 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15751 (spe_init_builtins): Delete.
15752 (spe_func_has_64bit_regs_p): Delete.
15753 (savres_routine_name): Delete "info" parameter. Adjust callers.
15754 (rs6000_emit_stack_reset): Ditto.
15755 (rs6000_dwarf_register_span): Delete.
15756 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15757 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15758 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15760 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15762 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15763 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15764 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15765 mulv2sf3, divv2sf3): Delete expanders.
15767 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15769 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15771 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15773 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15774 * config/rs6000/rs6000.c: Ditto.
15776 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15778 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15779 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15780 comparison_operator.
15782 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15784 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15785 * config/rs6000/rs6000.opt: Ditto.
15786 * config/rs6000/t-rtems: Ditto.
15788 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15790 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15791 TARGET_E500_SINGLE by 0, simplify.
15792 * config/rs6000/rs6000.c: Ditto.
15793 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15794 (spe_build_register_parallel): Delete.
15795 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15796 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15797 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15798 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15799 (E500_CONVERT): Delete.
15800 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15804 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15805 * config/rs6000/dfp.md: Ditto.
15806 (negdd2, *negdd2_fpr): Merge.
15807 (absdd2, *absdd2_fpr): Merge.
15808 (negtd2, *negtd2_fpr): Merge.
15809 (abstd2, *abstd2_fpr): Merge.
15810 * config/rs6000/e500.h: Delete file.
15811 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15812 TARGET_FPRS by 1 and simplify.
15813 * config/rs6000/rs6000-c.c: Ditto.
15814 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15815 TARGET_DF_SPE by 0.
15816 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15818 * config/rs6000/rs6000.md: Ditto.
15819 (floatdidf2, *floatdidf2_fpr): Merge.
15820 (move_from_CR_gt_bit): Delete.
15821 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15822 (E500_CR_IOR_COMPARE): Delete.
15823 (All patterns that require !TARGET_FPRS): Delete.
15824 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15826 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15828 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15830 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15832 * graphds.c (add_edge): Intitialize edge's attached data.
15833 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
15834 pointer parameter. Call pointed function on each edge during
15835 graph traversing. Skip traversing the edge when the function
15837 (graphds_dfs, graphds_scc): Ditto.
15838 (for_each_edge): New parameter. Pass the new parameter to callback
15840 * graphds.h (skip_edge_callback): New function pointer type.
15841 (graphds_dfs, graphds_scc): New function pointer parameter.
15842 (graphds_edge_callback, for_each_edge): New parameter.
15844 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15846 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
15847 out code checking if runtime alias check is possible to below ...
15848 Call the new function.
15849 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
15850 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
15852 2017-06-07 Marek Polacek <polacek@redhat.com>
15855 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
15856 TYPE_OVERFLOW_WRAPS check.
15858 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15860 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
15861 if versioning is required.
15862 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
15863 peeling with the check for versioning.
15865 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15867 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
15868 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
15869 Set true to new parameter if new ssa variable is defined.
15870 (vect_gen_vector_loop_niters): Refactor. Set range information
15871 for the new vector loop bound variable.
15872 (vect_do_peeling): Ditto.
15874 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15876 * tree-affine.c (ssa.h): Include header file.
15877 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
15878 has wrapping overflow behavior.
15880 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15882 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
15884 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15886 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
15887 (tree_to_aff_combination): ... here.
15889 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15891 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
15892 reg_pressure model function.
15893 (ivopts_global_cost_for_size): Delete.
15894 (determine_set_costs, iv_ca_recount_cost): Call new model function
15895 ivopts_estimate_reg_pressure.
15897 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15899 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
15900 expensive than udiv. Remove floating point cases from mod.
15902 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15904 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
15905 Increase idiv cost.
15907 2017-06-07 Tamar Christina <tamar.christina@arm.com>
15909 * config/aarch64/aarch64.md
15910 (copysignsf3): Fix mask generation.
15912 2017-06-07 Jakub Jelinek <jakub@redhat.com>
15914 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
15916 (class dump_manager): Add register_dumps method.
15917 * dumpfile.c: Include langhooks.h.
15918 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
15919 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
15920 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
15921 (dump_manager::dump_register): Start with 512 entries instead of 32.
15922 (dump_manager::register_dumps): New method.
15923 * toplev.c (general_init): Instead of invoking register_dumps
15924 langhook, invoke register_dumps method on the dump manager.
15925 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
15928 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
15930 * doc/md.texi: Clarify the restrictions on a define_insn condition.
15931 Say that # requires an associated define_split to exist, and that
15932 the define_split must be suitable for use after register allocation.
15934 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
15936 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
15937 (compute_outgoing_frequencies): Also initialize zero counts.
15938 (find_many_sub_basic_blocks): Do not produce uninitialized profile
15939 around loops; preserve more of profile when nothing changes.
15941 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
15943 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
15945 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
15946 * config/arm/arm-cpu-cdata.h: Regenerate.
15947 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
15948 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
15949 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
15950 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
15951 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
15953 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
15955 * config/arm/t-rmprofile: Likewise.
15956 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
15958 2017-06-06 David S. Miller <davem@davemloft.net>
15961 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
15962 blockage if function uses alloca.
15964 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15966 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
15967 New "uid" fields to hold pretty-print IDs of group and ref.
15968 Memory references are now identified as <group_id>:<ref_id>
15969 instead of using [random] addresses.
15970 (dump_mem_details): Simplify, no functional change.
15971 (dump_mem_ref): Simplify and make output more concise.
15972 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
15973 (find_or_create_group): Initialize group uid.
15974 (record_ref): Initialize ref uid. Improve debug output.
15975 (prune_group_by_reuse, should_issue_prefetch_p,)
15976 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
15977 (mark_nontemporal_store, determine_loop_nest_reuse):
15978 Improve debug output.
15980 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15982 * dbgcnt.def (prefetch): New debug counter.
15983 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
15984 (schedule_prefetches): Stop issueing prefetches if debug counter
15987 2017-06-06 Tom de Vries <tom@codesourcery.com>
15989 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
15990 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
15992 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15994 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
15995 Use aarch64_reg_or_zero predicate for operand 4.
15996 (aarch64_compare_and_swap<mode> define_insn_and_split):
15997 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
15998 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16000 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16002 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16003 (arm_compute_save_core_reg_mask): This.
16004 (thumb1_compute_save_reg_mask): Rename into ...
16005 (thumb1_compute_save_core_reg_mask): This.
16006 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16007 (arm_compute_frame_layout): Likewise.
16009 2017-06-06 Richard Biener <rguenther@suse.de>
16011 PR tree-optimization/80974
16012 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16013 keep or clear leaders SSA info.
16015 2017-06-06 Tom de Vries <tom@codesourcery.com>
16017 * config/nvptx/nvptx.c (split_mode_p): New function.
16018 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16020 2017-06-06 Tom de Vries <tom@codesourcery.com>
16022 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16024 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16027 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16030 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16032 * shrink-wrap.c (handle_simple_exit): Update profile.
16033 (try_shrink_wrapping): Upate profile.
16035 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16037 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16038 (tree_guess_outgoing_edge_probabilities): New.
16039 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16040 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16042 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16044 * ipa-split.c (split_function): Initialize return bb profile.
16046 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16048 * profile.c (compute_branch_probabilities): Also initialize
16049 EXIT_BLOCK profile.
16051 2017-06-06 Richard Biener <rguenther@suse.de>
16053 PR tree-optimization/80928
16054 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16055 (vect_analyze_loop_operations): Properly guard analysis for
16057 (vect_transform_loop): Likewise.
16058 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16059 (vect_model_induction_cost): Do not cost for pure SLP.
16060 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16061 of induction in inner loop vectorization.
16062 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16063 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16064 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16066 (vect_analyze_slp_cost_1): Cost induction.
16067 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16068 (vect_get_slp_vect_defs): Likewise.
16069 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16070 (vect_transform_stmt): Handle SLP reductions.
16071 * tree-vectorizer.h (vectorizable_induction): Adjust.
16073 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16075 * config/rs6000/rs6000.c (make_resolver_func): Update
16076 init_lowered_empty_function call.
16078 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16080 * doc/invoke.texi: Document the -fprofile-abs-path option.
16081 * common.opt (fprofile-abs-path): New option.
16082 * gcov-io.h (gcov_write_filename): Declare.
16083 * gcov-io.c (gcov_write_filename): New function.
16084 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16085 * profile.c (output_location): Likewise.
16087 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16089 * shring-wrap.c: Revert accidental commit.
16091 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16093 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16095 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16097 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16099 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16100 profile in callgraph edge.
16101 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16102 (apply_scale): Likewise.
16103 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16104 Also scale profile when inlining function with zero profile.
16105 (initialize_cfun): Update exit block profile even when it is zero.
16106 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16107 when profile is read.
16109 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16111 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16112 (CLONE_*): New constants to define the processors we can generate
16113 code for with the target_clone attribute.
16114 (rs6000_clone_map): New array to identify which clone processors
16115 the current program is running on.
16116 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16117 target_clone attribute.
16118 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16119 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16120 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16121 (cpu_expand_builtin): Add support for target_clone attribute.
16122 (rs6000_valid_attribute_p): Allow "default" attribute.
16123 (get_decl_name): New debug function to simplify printing the
16124 current function name in debugging statements.
16125 (rs6000_clone_priority): New functions to support the target_clone
16126 attribute, and be able to generate code to switch between ISA 2.05
16127 through ISA 3.0 (power6 through power9).
16128 (rs6000_compare_version_priority): Likewise.
16129 (rs6000_get_function_versions_dispatcher): Likewise.
16130 (make_resolver_func): Likewise.
16131 (add_condition_to_bb): Likewise.
16132 (dispatch_function_versions): Likewise.
16133 (rs6000_generate_version_dispatcher_body): Likewise.
16134 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16135 (fusion_gpr_load_p): Fix a spacing issue.
16136 * doc/extend.texi (Common Function Attributes): Document that the
16137 PowerPC supports the target_clone attribute.
16139 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16141 * config/arm/arm.h: explain F symbol found in description of ARM
16142 register allocation in its legend.
16144 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16146 * config/mips/frame-header-opt.c: Include profile-count.h.
16147 * config/riscv/riscv.c: Include profile-count.h
16149 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16151 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16153 (sm_set_flag_if_changed): Add bbs field.
16154 (execute_sm_if_changed_flag_set): Pass BBS.
16155 (execute_sm): Update.
16157 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16159 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16162 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16164 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16165 (peephole2): New peephole2 to emit the above.
16166 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16168 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16170 * config/aarch64/aarch64.c (define_peephole2 above
16171 *sub_<shift>_<mode>): New peephole.
16173 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16175 * config/i386/i386.c (make_resolver_func): Update.
16176 * Makefile.in: Add profile-count.h and profile-count.o
16177 * auto-profile.c (afdo_indirect_call): Update to new API.
16178 (afdo_set_bb_count): Update.
16179 (afdo_propagate_edge): Update.
16180 (afdo_propagate_circuit): Update.
16181 (afdo_calculate_branch_prob): Update.
16182 (afdo_annotate_cfg): Update.
16183 * basic-block.h: Include profile-count.h
16184 (struct edge_def): Turn count to profile_count.
16185 (struct basic_block_def): Likewie.
16186 (REG_BR_PROB_BASE): Move to profile-count.h
16187 (RDIV): Move to profile-count.h
16188 * bb-reorder.c (max_entry_count): Turn to profile_count.
16189 (find_traces): Update.
16190 (rotate_loop):Update.
16191 (connect_traces):Update.
16192 (sanitize_hot_paths):Update.
16193 * bt-load.c (migrate_btr_defs): Update.
16194 * cfg.c (RDIV): Remove.
16195 (init_flow): Use alloc_block.
16196 (alloc_block): Uninitialize count.
16197 (unchecked_make_edge): Uninitialize count.
16198 (check_bb_profile): Update.
16199 (dump_edge_info): Update.
16200 (dump_bb_info): Update.
16201 (update_bb_profile_for_threading): Update.
16202 (scale_bbs_frequencies_int): Update.
16203 (scale_bbs_frequencies_gcov_type): Update.
16204 (scale_bbs_frequencies_profile_count): New.
16205 * cfg.h (update_bb_profile_for_threading): Update.
16206 (scale_bbs_frequencies_profile_count): Declare.
16207 * cfgbuild.c (compute_outgoing_frequencies): Update.
16208 (find_many_sub_basic_blocks): Update.
16209 * cfgcleanup.c (try_forward_edges): Update.
16210 (try_crossjump_to_edge): Update.
16211 * cfgexpand.c (expand_gimple_tailcall): Update.
16212 (construct_exit_block): Update.
16213 * cfghooks.c (verify_flow_info): Update.
16214 (dump_bb_for_graph): Update.
16215 (split_edge): Update.
16216 (make_forwarder_block): Update.
16217 (duplicate_block): Update.
16218 (account_profile_record): Update.
16219 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16220 (get_estimated_loop_iterations): Update.
16221 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16222 (single_likely_exit): Update.
16223 * cfgloopmanip.c (scale_loop_profile): Update.
16225 (set_zero_probability): Update.
16226 (lv_adjust_loop_entry_edge): Update.
16227 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16228 (purge_dead_edges): Update.
16229 (rtl_account_profile_record): Update.
16230 * cgraph.c (cgraph_node::create): Uninitialize count.
16231 (symbol_table::create_edge): Uninitialize count.
16232 (cgraph_update_edges_for_call_stmt_node): Update.
16233 (cgraph_edge::dump_edge_flags): Update.
16234 (cgraph_node::dump): Update.
16235 (cgraph_edge::maybe_hot_p): Update.
16236 * cgraph.h: Include profile-count.h
16237 (create_clone), create_edge, create_indirect_edge): Update.
16238 (cgraph_node): Turn count to profile_count.
16239 (cgraph_edge0: Likewise.
16240 (make_speculative, clone): Update.
16241 (create_edge): Update.
16242 (init_lowered_empty_function): Update.
16243 * cgraphclones.c (cgraph_edge::clone): Update.
16244 (duplicate_thunk_for_node): Update.
16245 (cgraph_node::create_clone): Update.
16246 * cgraphunit.c (cgraph_node::analyze): Update.
16247 (cgraph_node::expand_thunk): Update.
16248 * final.c (dump_basic_block_info): Update.
16249 * gimple-streamer-in.c (input_bb): Update.
16250 * gimple-streamer-out.c (output_bb): Update.
16251 * graphite.c (print_global_statistics): Update.
16252 (print_graphite_scop_statistics): Update.
16253 * hsa-brig.c: Include basic-block.h.
16254 * hsa-dump.c: Include basic-block.h.
16255 * hsa-gen.c (T sum_slice): Update.
16256 (convert_switch_statements):Update.
16257 * hsa-regalloc.c: Include basic-block.h.
16258 * ipa-chkp.c (chkp_produce_thunks): Update.
16259 * ipa-cp.c (struct caller_statistics): Update.
16260 (init_caller_stats): Update.
16261 (gather_caller_stats): Update.
16262 (ipcp_cloning_candidate_p): Update.
16263 (good_cloning_opportunity_p): Update.
16264 (get_info_about_necessary_edges): Update.
16265 (dump_profile_updates): Update.
16266 (update_profiling_info): Update.
16267 (update_specialized_profile): Update.
16268 (perhaps_add_new_callers): Update.
16269 (decide_about_value): Update.
16270 (ipa_cp_c_finalize): Update.
16271 * ipa-devirt.c (struct odr_type_warn_count): Update.
16272 (struct decl_warn_count): Update.
16273 (struct final_warning_record): Update.
16274 (possible_polymorphic_call_targets): Update.
16275 (ipa_devirt): Update.
16276 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16277 * ipa-icf.c (sem_function::merge): Update.
16278 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16279 * ipa-inline.c (compute_uninlined_call_time): Update.
16280 (compute_inlined_call_time): Update.
16281 (want_inline_small_function_p): Update.
16282 (want_inline_self_recursive_call_p): Update.
16283 (edge_badness): Update.
16284 (lookup_recursive_calls): Update.
16285 (recursive_inlining): Update.
16286 (inline_small_functions): Update.
16287 (dump_overall_stats): Update.
16288 (dump_inline_stats): Update.
16289 * ipa-profile.c (ipa_profile_generate_summary): Update.
16290 (ipa_propagate_frequency): Update.
16291 (ipa_profile): Update.
16292 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16293 * ipa-utils.c (ipa_merge_profiles): Update.
16294 * loop-doloop.c (doloop_modify): Update.
16295 * loop-unroll.c (report_unroll): Update.
16296 (unroll_loop_runtime_iterations): Update.
16297 * lto-cgraph.c (lto_output_edge): Update.
16298 (lto_output_node): Update.
16299 (input_node): Update.
16300 (input_edge): Update.
16301 (merge_profile_summaries): Update.
16302 * lto-streamer-in.c (input_cfg): Update.
16303 * lto-streamer-out.c (output_cfg): Update.
16304 * mcf.c (create_fixup_graph): Update.
16305 (adjust_cfg_counts): Update.
16306 (sum_edge_counts): Update.
16307 * modulo-sched.c (sms_schedule): Update.
16308 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16309 * predict.c (maybe_hot_count_p): Update.
16310 (probably_never_executed): Update.
16311 (dump_prediction): Update.
16312 (combine_predictions_for_bb): Update.
16313 (propagate_freq): Update.
16314 (handle_missing_profiles): Update.
16315 (counts_to_freqs): Update.
16316 (rebuild_frequencies): Update.
16317 (force_edge_cold): Update.
16318 * predict.h: Include profile-count.h
16319 (maybe_hot_count_p, counts_to_freqs): UPdate.
16320 * print-rtl-function.c: Do not include cfg.h
16321 * print-rtl.c: Include basic-block.h
16322 * profile-count.c: New file.
16323 * profile-count.h: New file.
16324 * profile.c (is_edge_inconsistent): Update.
16325 (correct_negative_edge_counts): Update.
16326 (is_inconsistent): Update.
16327 (set_bb_counts): Update.
16328 (read_profile_edge_counts): Update.
16329 (compute_frequency_overlap): Update.
16330 (compute_branch_probabilities): Update; Initialize and deinitialize
16332 (branch_prob): Update.
16333 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16334 (edge_gcov_count): New.
16335 (bb_gcov_count): New.
16336 * shrink-wrap.c (try_shrink_wrapping): Update.
16337 * tracer.c (better_p): Update.
16338 * trans-mem.c (expand_transaction): Update.
16339 (ipa_tm_insert_irr_call): Update.
16340 (ipa_tm_insert_gettmclone_call): Update.
16341 * tree-call-cdce.c: Update.
16342 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16343 (gimple_duplicate_sese_tail): Update.
16344 (gimple_account_profile_record): Update.
16345 (execute_fixup_cfg): Update.
16346 * tree-inline.c (copy_bb): Update.
16347 (copy_edges_for_bb): Update.
16348 (initialize_cfun): Update.
16349 (freqs_to_counts): Update.
16350 (copy_cfg_body): Update.
16351 (expand_call_inline): Update.
16352 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16353 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16354 (try_unroll_loop_completely): Update.
16355 (try_peel_loop): Update.
16356 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16357 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16358 * tree-ssa-loop-split.c (connect_loops): Update.
16359 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16360 * tree-ssa-reassoc.c (branch_fixup): Update.
16361 * tree-ssa-tail-merge.c (replace_block_by): Update.
16362 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16363 (compute_path_counts): Update.
16364 (update_profile): Update.
16365 (recompute_probabilities): Update.
16366 (update_joiner_offpath_counts): Update.
16367 (estimated_freqs_path): Update.
16368 (freqs_to_counts_path): Update.
16369 (clear_counts_path): Update.
16370 (ssa_fix_duplicate_block_edges): Update.
16371 (duplicate_thread_path): Update.
16372 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16373 (struct switch_conv_info): Update.
16374 * tree-tailcall.c (decrease_profile): Update.
16375 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16376 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16377 * value-prof.c (check_counter): Update.
16378 (gimple_divmod_fixed_value): Update.
16379 (gimple_mod_pow2): Update.
16380 (gimple_mod_subtract): Update.
16381 (gimple_ic_transform): Update.
16382 (gimple_stringop_fixed_value): Update.
16383 * value-prof.h (gimple_ic): Update.
16385 2017-06-02 Carl Love <cel@us.ibm.com>
16387 * config/rs6000/rs6000-c: Add support for built-in functions
16388 vector double vec_doublee (vector signed int);
16389 vector double vec_doublee (vector unsigned int);
16390 vector double vec_doublee (vector float);
16391 vector double vec_doubleh (vector signed int);
16392 vector double vec_doubleh (vector unsigned int);
16393 vector double vec_doubleh (vector float);
16394 vector double vec_doublel (vector signed int);
16395 vector double vec_doublel (vector unsigned int);
16396 vector double vec_doublel (vector float);
16397 vector double vec_doubleo (vector signed int);
16398 vector double vec_doubleo (vector unsigned int);
16399 vector double vec_doubleo (vector float);.
16400 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16401 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16403 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16404 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16405 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16407 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16408 vec_doublel, vec_doubleh.
16409 * doc/extend.texi: Update the built-in documentation file for the
16410 new built-in functions.
16412 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16415 * ipa-inline-analysis.c (free_growth_caches): Set
16416 edge_removal_hook_holder to NULL after removing it.
16418 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16420 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16421 comparision with zero.
16423 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16424 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16425 for early expansion of vec_min and vec_max builtins.
16426 (builtin_function_type): Add min/max unsigned variants to those
16427 identified as having unsigned arguments.
16429 2017-06-02 Olivier Hainque <hainque@adacore.com>
16431 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16433 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16435 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16436 Use VALL_F16 iterator rather than VALL.
16438 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16440 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16441 Emit CBNZ inside loop when doing a strong exchange and comparing
16442 against zero. Generate the CC flags after the loop.
16444 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16446 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16447 (dl_section_ref): New.
16448 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16449 On AIX, append an expression to subtract the size of the
16450 section length to dl_section_ref.
16452 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16454 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16455 for early expansion of vector absolute builtins.
16457 2017-06-02 Richard Biener <rguenther@suse.de>
16459 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16460 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16462 2017-06-02 Richard Biener <rguenther@suse.de>
16464 PR tree-optimization/80948
16465 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16466 stmt order as well.
16468 2017-06-02 Richard Biener <rguenther@suse.de>
16470 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16472 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16473 for inductions as relevant.
16475 2017-06-02 Richard Biener <rguenther@suse.de>
16477 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16478 (vectorizable_induction): ... this. Remove dead code.
16480 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16482 * builtins. (expand_builtin_alloca): Remove second parameter and
16483 infer its value from the first parameter instead.
16484 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16486 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16488 PR rtl-optimization/80903
16489 * loop-doloop.c (add_test): Unshare sequence.
16491 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16493 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16495 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16497 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16499 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16500 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16501 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16502 (xlogue_layout::get_stub_rtx): Make static.
16503 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16504 (xlogue_layout::compute_stub_managed_regs): Rename to...
16505 (xlogue_layout::count_stub_managed_regs): ...this.
16506 (xlogue_layout::is_stub_managed_reg): New function.
16507 (xlogue_layout::m_stub_names): Rename to...
16508 (xlogue_layout::s_stub_names): ...this, make static.
16509 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16510 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16511 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16512 xlogue_layout::s_stub_names): Instantiate statics.
16513 (stub_managed_regs): Remove.
16514 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16515 (disable_call_ms2sysv_xlogues): Rename to...
16516 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16517 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16519 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16520 change after reload_completed.
16521 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16523 (ix86_expand_prologue): Likewise.
16524 (ix86_expand_epilogue): Likewise.
16525 (ix86_expand_split_stack_prologue): Likewise.
16526 (ix86_compute_frame_layout): Remove frame parameter ...
16527 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16528 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16530 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16531 (ix86_frame): Move from here ...
16532 * config/i386/i386.h (ix86_frame): ... to here.
16533 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16534 complete ix86_frame data structure instead. Remove some_ld_name.
16536 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16538 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16539 symbols that hold a DECL_VALUE_EXPR.
16541 2017-06-01 Martin Jambor <mjambor@suse.cz>
16543 PR tree-optimization/80898
16544 * tree-sra.c (process_subtree_disqualification): Removed.
16545 (disqualify_candidate): Do not acll
16546 process_subtree_disqualification.
16547 (subtree_mark_written_and_enqueue): New function.
16548 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16549 RHS has been disqualified and re-queue LHS if necessary. Apart
16550 from that, ignore disqualified RHS.
16552 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16554 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16555 address fetch for z10 or later.
16557 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16559 * config/arc/arc.md (tst_movb): Add guard when splitting.
16561 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16563 * config/arc/arc.c (arc_can_eliminate): Test against
16564 arc_frame_pointer_needed.
16566 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16568 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16569 to prevent store reordering.
16570 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16571 (type): Add block type.
16572 (stack_tie): Define special instruction to be used in
16575 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16577 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16578 constraint. It is not valid for the pattern.
16579 (noncommutative_binary_comparison): Likewise.
16581 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16583 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16586 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16588 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16589 be used by the reg-alloc.
16591 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16593 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16594 reg-alloc when having mul64 or mul32x16 instructions.
16595 (mulsidi3): Likewise.
16596 (umulsidi3): Likewise.
16597 (mulsi32x16): New pattern.
16598 (mulsi64): Likewise.
16599 (mulsidi64): Likewise.
16600 (umulsidi64): Likewise.
16601 (MUL32x16_REG): Define.
16602 (mul64_600): Use MUL32x16_REG.
16603 (mac64_600): Likewise.
16604 (umul64_600): Likewise.
16605 (umac64_600): Likewise.
16607 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16609 * config/arc/arc.md (mulsi3_700): Make it commutative.
16611 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16613 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16615 (*sign_extendsidi2_insn): Likewise for movstosw.
16617 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16619 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16620 the type of the input discriminant value. Convert the
16621 discriminant value of signedness vary.
16623 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16625 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16626 Add to -Wall section.
16628 2017-06-01 Richard Biener <rguenther@suse.de>
16630 PR middle-end/66313
16631 * fold-const.c (fold_plusminus_mult_expr): If the factored
16632 factor may be zero use a wrapping type for the inner operation.
16633 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16634 and handle moved defs.
16635 (process_assignment): Properly guard the unary op case. Return a
16636 tri-state indicating that moving the stmt before the call may allow
16637 to continue. Pass through to_move.
16638 (find_tail_calls): Handle moving unrelated defs before
16641 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16644 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16645 splitter result in the canonical way.
16647 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16649 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16650 also for 32bit target. Update insn attributes.
16651 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16653 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16655 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16656 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16657 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16658 (_mm_maskz_min_ss): New intrinsics.
16660 2017-05-31 Martin Liska <mliska@suse.cz>
16662 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16663 Change comment style to one we normally use.
16664 (vectorizable_reduction): Likewise.
16665 (vectorizable_induction): Likewise.
16666 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16667 (vectorizable_call): Likewise.
16668 (vectorizable_simd_clone_call): Likewise.
16669 (vectorizable_conversion): Likewise.
16670 (vectorizable_assignment): Likewise.
16671 (vectorizable_shift): Likewise.
16672 (vectorizable_operation): Likewise.
16673 (vectorizable_store): Likewise.
16674 (vectorizable_load): Likewise.
16675 * tree-vectorizer.h: Likewise.
16677 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16679 * passes.c (emergency_dump_function): New.
16680 * tree-pass.h (emergency_dump_function): Declare.
16681 * plugin.c (plugins_internal_error_function): Remove.
16682 * plugin.h (plugins_internal_error_function): Remove declaration.
16683 * toplev.c (internal_error_function): New static function. Use it...
16684 (general_init): ...here.
16686 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16688 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16689 (arc_rtx_costs): Add costs for new patterns.
16690 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16691 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16693 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16695 * tree-ssa-strlen.c (get_next_strinfo): New function.
16696 (get_stridx_plus_constant): Use it.
16697 (zero_length_string): Likewise.
16698 (adjust_related_strinfos): Likewise.
16699 (adjust_last_stmt): Likewise.
16701 2017-05-31 Richard Biener <rguenther@suse.de>
16704 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16705 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16707 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16709 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16710 loop_vinfo argument and use of dependence distance vectors.
16711 Check instead whether the two references differ only in their
16712 initial value and assume that they have the same alignment if the
16713 difference is a multiple of the vector alignment.
16714 (vect_analyze_data_refs_alignment): Update call accordingly.
16716 2017-05-31 Martin Liska <mliska@suse.cz>
16719 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16721 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16723 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16724 (create_intersect_range_checks): Move from ...
16725 * tree-data-ref.c (create_intersect_range_checks_index)
16726 (create_intersect_range_checks): ... to here.
16727 (create_runtime_alias_checks): New function factored from ...
16728 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16729 here. Call above function.
16730 * tree-data-ref.h (create_runtime_alias_checks): New function.
16732 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16734 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16735 segment length for dr_b and compute it in wide_int.
16737 2017-05-31 Richard Biener <rguenther@suse.de>
16739 PR tree-optimization/80906
16740 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16741 and pass through iv_map.
16742 (copy_bb_and_scalar_dependences): Adjust.
16743 (translate_pending_phi_nodes): Likewise.
16744 (copy_loop_close_phi_args): Handle code-generating IVs instead
16747 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16749 * diagnostic-color.c (color_dict): Add "type-diff".
16750 (parse_gcc_colors): Update comment.
16751 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16752 -fdiagnostics-show-template-tree and -fno-elide-type.
16753 (GCC_COLORS): Add type-diff to example.
16755 (-fdiagnostics-show-template-tree): New.
16756 (-fno-elide-type): New.
16757 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16758 the pp_format_decoder callback. Call any m_format_postprocessor's
16760 (pretty_printer::pretty_printer): Initialize
16761 m_format_postprocessor.
16762 (pretty_printer::~pretty_printer): Delete any
16763 m_format_postprocessor.
16764 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16765 (class format_postprocessor): New class.
16766 (struct pretty_printer::format_decoder): Document the new parameters.
16767 (struct pretty_printer::m_format_postprocessor): New field.
16768 * tree-diagnostic.c (default_tree_printer): Update for new
16769 bool and const char ** params.
16770 * tree-diagnostic.h (default_tree_printer): Likewise.
16772 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16774 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16775 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16776 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16777 rs6000_gen_cell_microcode code.
16778 (rs6000_final_prescan_insn): Delete.
16779 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16780 "warn-cell-microcode" entries.
16781 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16782 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16783 throughout. Change cc_reg_not_micro_cr0_operand to
16784 cc_reg_not_cr0_operand throughout.
16785 (*extendhi<mode>2_noload): Delete.
16786 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16787 (mwarn-cell-microcode): Delete.
16788 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16789 -mgen-cell-microcode and -mwarn-cell-microcode.
16791 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16794 * config/i386/constraints.md (Yd): New constraint.
16796 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16797 and (?Yd, r) alternatives. Update insn attributes.
16798 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16799 and (?*Yd, r) alternatives. Update insn attributes.
16800 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16802 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16804 * gimplify.c (gimplify_modify_expr): Don't create a
16805 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16808 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16810 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16812 2017-05-30 Richard Biener <rguenther@suse.de>
16814 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16815 and reduc_def fields.
16816 (STMT_VINFO_REDUC_TYPE): New define.
16817 (STMT_VINFO_REDUC_DEF): Likewise.
16818 (vect_force_simple_reduction): Adjust prototype.
16819 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16820 (vect_is_simple_reduction): Remove check_reduction argument.
16821 (vect_force_simple_reduction): Adjust and set
16822 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16823 (vectorizable_reduction): Do not re-do reduction analysis
16824 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16825 * tree-parloops.c (gather_scalar_reductions): Adjust.
16827 2017-05-30 Richard Biener <rguenther@suse.de>
16829 PR middle-end/80901
16830 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
16833 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
16835 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
16836 Introduce unknown_misalignment parameter and remove vf.
16837 (vect_peeling_hash_get_lowest_cost):
16838 Pass unknown_misalignment parameter.
16839 (vect_enhance_data_refs_alignment):
16840 Fix unsupportable data ref treatment.
16842 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16844 * tree-vect-data-refs.c (vect_get_data_access_cost):
16845 Workaround for SLP handling.
16846 (vect_enhance_data_refs_alignment):
16847 Compute costs for doing no peeling at all, compare to the best
16848 peeling costs so far and avoid peeling if cheaper.
16850 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16852 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
16853 Return peeling info and set costs to zero for unlimited cost
16855 (vect_enhance_data_refs_alignment): Also inspect all datarefs
16856 with unknown misalignment. Compute and costs for unknown
16857 misalignment, compare them to the costs for known misalignment
16858 and choose the cheapest for peeling.
16860 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16862 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
16863 (vect_get_peeling_costs_all_drs): Create function.
16864 (vect_peeling_hash_get_lowest_cost):
16865 Use vect_get_peeling_costs_all_drs.
16866 (vect_peeling_supportable): Create function.
16867 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
16869 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
16871 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
16872 DR_HAS_NEGATIVE_STEP.
16873 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
16874 (vect_enhance_data_refs_alignment): Use.
16875 (vect_duplicate_ssa_name_ptr_info): Use.
16876 * tree-vectorizer.h (dr_misalignment): Use.
16877 (known_alignment_for_access_p): Use.
16879 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16882 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
16884 (has_section_name): New function.
16886 2017-05-30 Martin Liska <mliska@suse.cz>
16889 * auto-profile.c (get_function_decl_from_block): Fix
16892 2017-05-30 Richard Biener <rguenther@suse.de>
16894 PR middle-end/80876
16895 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
16897 2017-05-30 Martin Liska <mliska@suse.cz>
16899 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
16900 * dumpfile.h (struct dump_file_info): Remove ctors.
16902 2017-05-30 Martin Liska <mliska@suse.cz>
16904 * predict.def: Fix GNU coding style.
16906 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16908 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16909 Mark 'to' argument with ATTRIBUTE_UNUSED.
16911 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
16913 * config/xtensa/xtensa.c (xtensa_emit_call): Use
16914 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
16915 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
16918 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
16920 * doc/install.texi (Options specification): Restore entry of
16921 --enable-sjlj-exceptions.
16923 2017-05-27 Michael Eager <eager@eagercon.com>
16926 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
16928 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
16930 * config/microblaze/microblaze.h
16931 (FIXED_REGISTERS): Update in macro.
16932 (CALL_USED_REGISTERS): Update in macro.
16934 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
16936 * doc/install.texi: Add links to macOS binary distributions.
16938 2017-05-27 Jakub Jelinek <jakub@redhat.com>
16942 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
16944 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
16946 2017-05-26 Martin Liska <mliska@suse.cz>
16948 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
16950 2017-05-26 Martin Liska <mliska@suse.cz>
16952 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
16953 always leading ';; '.
16954 (dump_bb_info): Likewise.
16955 (brief_dump_cfg): Likewise.
16956 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
16957 * dumpfile.c: Remove usage of TDF_VERBOSE.
16958 * dumpfile.h (enum dump_kind): Likewise.
16959 (dump_gimple_bb_header): Do not use TDF_COMMENT.
16960 * print-tree.c (debug_verbose): Remove.
16961 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
16962 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
16963 * tree-diagnostic.c (default_tree_printer): Replace
16964 TDF_DIAGNOSTIC with TDF_SLIM.
16966 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16968 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
16969 in parameter loop, rather than loop_vinfo.
16970 (create_intersect_range_checks): Ditto.
16971 (vect_create_cond_for_alias_checks): Update call to above functions.
16973 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16975 PR tree-optimization/80815
16976 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
16977 for merging runtime alias checks. Handle negative DR_STEPs.
16979 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16981 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
16983 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
16984 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
16985 out code pruning runtime alias checks.
16986 * tree-data-ref.c (prune_runtime_alias_test_list): New function
16987 factored out from above.
16988 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16990 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
16992 (prune_runtime_alias_test_list): New decalaration.
16994 2017-05-26 Bin Cheng <bin.cheng@arm.com>
16996 * tree-vect-data-refs.c (compare_tree): Rename and move ...
16997 * tree-data-ref.c (data_ref_compare_tree): ... to here.
16998 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
16999 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17000 (operator==, comp_dr_with_seg_len_pair): Ditto.
17001 (vect_prune_runtime_alias_test_list): Ditto.
17003 2017-05-26 Martin Liska <mliska@suse.cz>
17006 * params.def: Bound partial-inlining-entry-probability param.
17008 2017-05-26 Marek Polacek <polacek@redhat.com>
17011 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17014 2017-05-26 Richard Biener <rguenther@suse.de>
17016 PR tree-optimization/80842
17017 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17020 2017-05-26 Richard Biener <rguenther@suse.de>
17022 PR tree-optimization/80844
17023 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17025 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17027 * doc/md.texi (Machine Constraints): Update x86 family
17028 machine constraints section to match 'config/i386/constraints.md'.
17030 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17032 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17034 2017-05-25 Nathan Sidwell <nathan@acm.org>
17036 * doc/invoke.texi (--enable-languages): Update documentation.
17038 2017-05-25 Martin Liska <mliska@suse.cz>
17040 * dumpfile.c: Add TDF_FOLDING.
17041 * dumpfile.h (enum dump_kind): Likewise.
17042 * genmatch.c (dt_simplify::gen_1): Use it.
17044 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17046 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17048 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17050 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17051 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17053 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17055 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17056 * match.pd (X == C): Rewrite it here.
17057 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17058 with_certain_nonzero_bits2): New predicates.
17059 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17061 2017-05-24 Nathan Sidwell <nathan@acm.org>
17063 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17066 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17069 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17071 * config/powerpcspe: New port. Files are copied from the rs6000
17072 port, with "rs6000" in filenames replaced by "powerpcspe".
17074 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17076 PR rtl-optimization/80754
17077 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17079 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17081 * config/sparc/sparc.md (length): Return the correct value for -mflat
17082 sibcalls to match output_sibcall.
17084 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17088 * config/rs6000/rs6000.c (struct machine_function): Add new field
17090 (rs6000_get_separate_components): Init that field, use it.
17091 (rs6000_components_for_bb): Use the field.
17093 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17095 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17097 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17099 PR middle-end/80823
17100 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17102 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17105 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17106 address against address_operand predicate.
17107 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17109 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17111 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17112 parameters passed indirectly.
17114 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17116 * config/i386/i386.md (*movdi_internal): Remove SSE4
17117 alternative 18 (?r, *v). Update insn attributes.
17118 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17119 Update insn attributes.
17120 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17121 Update insn attributes.
17122 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17123 alternative 1 (r, v). Remove isa attribute.
17124 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17125 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17126 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17128 2017-05-23 Tom de Vries <tom@codesourcery.com>
17130 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17133 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17135 * cgraphunit.c (symbol_table::process_new_functions): Update.
17136 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17137 (inline_generate_summary): Rename to ...
17138 (ipa_fn_summary_generate): ... this one.
17139 (inline_read_summary): Rename to ...
17140 (ipa_fn_summary_read): ... this one.
17141 (inline_write_summary): Rename to ...
17142 (ipa_fn_summary_write): ... this one.
17143 (inline_free_summary): Rename to ...
17144 (ipa_free_fn_summary): ... this one.
17145 (pass_data_local_fn_summary, pass_local_fn_summary,
17146 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17147 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17148 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17149 make_pass_ipa_fn_summary): New.
17150 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17151 inline_write_summary, inline_free_summary): Remove.
17152 (ipa_free_fn_summary) : New.
17153 * ipa-inline.c (ipa_inline): Update.
17154 (pass_ipa_inline): Do not generate summaries.
17155 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17157 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17158 and add pass_ipa_fn_summary.
17159 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17161 (make_pass_inline_parameters): Remove.
17163 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17165 * omp-low.c (struct omp_context): Remove "default_kind" member.
17168 * omp-offload.c (execute_oacc_device_lower): Remove the
17169 parallelism dimensions function attributes for unparallelized
17170 OpenACC kernels constructs.
17172 2017-05-23 Martin Liska <mliska@suse.cz>
17174 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17176 (cgraph_edge::make_speculative): Likewise.
17177 (cgraph_edge::resolve_speculation): Likewise.
17178 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17179 (cgraph_node::dump): Likewise.
17180 * cgraph.h: Likewise.
17181 * cgraphunit.c (analyze_functions): Likewise.
17182 (symbol_table::compile): Likewise.
17183 * ipa-cp.c (print_all_lattices): Likewise.
17184 (determine_versionability): Likewise.
17185 (initialize_node_lattices): Likewise.
17186 (ipcp_verify_propagated_values): Likewise.
17187 (estimate_local_effects): Likewise.
17188 (update_profiling_info): Likewise.
17189 (create_specialized_node): Likewise.
17190 (perhaps_add_new_callers): Likewise.
17191 (decide_about_value): Likewise.
17192 (decide_whether_version_node): Likewise.
17193 (identify_dead_nodes): Likewise.
17194 (ipcp_store_bits_results): Likewise.
17195 * ipa-devirt.c (dump_targets): Likewise.
17196 (ipa_devirt): Likewise.
17197 * ipa-icf.c (sem_item::dump): Likewise.
17198 (sem_function::equals): Likewise.
17199 (sem_variable::equals): Likewise.
17200 (sem_item_optimizer::read_section): Likewise.
17201 (sem_item_optimizer::execute): Likewise.
17202 (congruence_class::dump): Likewise.
17203 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17204 (dump_inline_summary): Likewise.
17205 (estimate_node_size_and_time): Likewise.
17206 (inline_analyze_function): Likewise.
17207 * ipa-inline-transform.c (inline_call): Likewise.
17208 * ipa-inline.c (report_inline_failed_reason): Likewise.
17209 (want_early_inline_function_p): Likewise.
17210 (edge_badness): Likewise.
17211 (update_edge_key): Likewise.
17212 (inline_small_functions): Likewise.
17213 * ipa-profile.c (ipa_profile): Likewise.
17214 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17215 (ipa_make_edge_direct_to_target): Likewise.
17216 (remove_described_reference): Likewise.
17217 (ipa_impossible_devirt_target): Likewise.
17218 (propagate_controlled_uses): Likewise.
17219 (ipa_print_node_params): Likewise.
17220 (ipcp_transform_function): Likewise.
17221 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17222 (propagate_pure_const): Likewise.
17223 * ipa-reference.c (generate_summary): Likewise.
17224 (read_write_all_from_decl): Likewise.
17225 (propagate): Likewise.
17226 (ipa_reference_read_optimization_summary): Likewise.
17227 * ipa-utils.c (ipa_merge_profiles): Likewise.
17228 * ipa.c (walk_polymorphic_call_targets): Likewise.
17229 (symbol_table::remove_unreachable_nodes): Likewise.
17230 (ipa_single_use): Likewise.
17231 * passes.c (execute_todo): Likewise.
17232 * predict.c (drop_profile): Likewise.
17233 * symtab.c (symtab_node::get_dump_name): New function.
17234 (symtab_node::dump_name): Likewise.
17235 (symtab_node::dump_asm_name): Likewise.
17236 (symtab_node::dump_references): Likewise.
17237 (symtab_node::dump_referring): Likewise.
17238 (symtab_node::dump_base): Likewise.
17239 (symtab_node::debug_symtab): Likewise.
17240 * tree-sra.c (convert_callers_for_node): Likewise.
17241 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17242 * value-prof.c (init_node_map): Likewise.
17244 2017-05-23 Martin Liska <mliska@suse.cz>
17246 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17247 and symtab_node::debug_symtab to symbol_table::debug.
17248 * cgraphunit.c (analyze_functions): Use the renamed function.
17249 (symbol_table::compile): Likewise.
17250 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17251 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17252 * passes.c (execute_todo): Likewise.
17253 * symtab.c (symbol_table::dump): New function.
17254 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17256 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17258 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17259 that nonconst implies exec.
17261 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17263 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17264 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17265 (inline_edge_summary_vec): Turn into ...
17266 (ipa_call_summaries): ... this one.
17267 (redirect_to_unreachable, edge_set_predicate,
17268 evaluate_properties_for_edge, inline_summary_alloc,
17269 reset_ipa_call_summary, reset_inline_summary,
17270 inline_summary_t::duplicate): Update.
17271 (inline_edge_duplication_hook): Turn to ...
17272 (ipa_call_summary_t::duplicate): ... this one.
17273 (inline_edge_removal_hook): Turn to ...
17274 (ipa_call_summary_t::remove): ... this one.
17275 (dump_inline_edge_summary): Turn to ...
17276 (dump_ipa_call_summary): ... this one.
17277 (estimate_function_body_sizes): Update.
17278 (inline_update_callee_summaries): Update.
17279 (remap_edge_change_prob): Update.
17280 (remap_edge_summaries): Update.
17281 (inline_merge_summary): Update.
17282 (do_estimate_edge_time): Update.
17283 (inline_generate_summary): Update.
17284 (inline_read_section): Update.
17285 (inline_read_summary): Update.
17286 (inline_free_summary): Update.
17287 * ipa-inline.c (can_inline_edge_p): Update.
17288 (compute_inlined_call_time): Update.
17289 (want_inline_small_function_p): Update.
17290 (edge_badness): Update.
17291 (early_inliner): Update.
17292 * ipa-inline.h (inline_edge_summary): Turn to ...
17293 (ipa_call_summary): ... this one.
17294 (ipa_call_summary_t): New class.
17295 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17296 (ipa_call_summaries): New.
17297 (inline_edge_summary): Remove.
17298 (estimate_edge_growth): Update.
17299 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17300 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17301 * ipa-split.c (execute_split_functions): Update.
17302 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17304 2017-05-23 Tom de Vries <tom@codesourcery.com>
17306 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17307 attributes): Document rdrand effective target.
17309 2017-05-23 Tom de Vries <tom@codesourcery.com>
17311 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17312 attributes): Sort alphabetically.
17314 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17316 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17318 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17321 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17322 V2DF/V2DI splat into two separate patterns, one that handles
17323 registers, and the other that only handles memory. Drop support
17324 for splatting from a GPR on ISA 2.07 and then splitting the
17325 splat into direct move and splat.
17326 (vsx_splat_<mode>_reg): Likewise.
17327 (vsx_splat_<mode>_mem): Likewise.
17329 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17331 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17333 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17335 PR middle-end/80809
17336 * omp-low.c (finish_taskreg_remap): New function.
17337 (finish_taskreg_scan): If unit size of ctx->record_type
17338 is non-constant, unshare the size expression and replace
17339 decls in it with possible outer var refs.
17341 PR middle-end/80809
17342 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17343 GOVD_SHARED rather than GOVD_PRIVATE with it.
17344 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17345 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17347 PR middle-end/80853
17348 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17349 as last argument to build_outer_var_ref for pointer bases of array
17350 section reductions.
17352 2017-05-19 Martin Sebor <msebor@redhat.com>
17354 * print-tree.c (print_node): Print DECL_READ_P flag.
17356 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17358 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17359 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17360 * cgraph.c: Likewise.
17361 * cgraphunit.c: Likewise.
17362 * gengtype.c: Likewise.
17363 * ipa-cp.c: Likewise.
17364 * ipa-devirt.c: Likewise.
17365 * ipa-icf.c: Likewise.
17366 * ipa-predicate.c: Likewise.
17367 * ipa-profile.c: Likewise.
17368 * ipa-prop.c: Likewise.
17369 * ipa-split.c: Likewise.
17371 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17372 edge_predicate_pool, dump_inline_hints,
17373 inline_summary::account_size_time, redirect_to_unreachable,
17374 edge_set_predicate, set_hint_predicate,
17375 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17376 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17377 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17378 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17379 ipa_call_summary_t::remove, initialize_growth_caches,
17380 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17381 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17382 mark_modified, unmodified_parm_1, unmodified_parm,
17383 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17384 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17385 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17386 will_be_nonconstant_predicate, record_modified_bb_info,
17387 get_minimal_bb, record_modified, param_change_prob,
17388 phi_result_unknown_predicate, predicate_for_phi_result,
17389 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17390 estimate_function_body_sizes, compute_inline_parameters,
17391 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17392 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17393 inline_update_callee_summaries, remap_edge_change_prob,
17394 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17395 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17396 inline_analyze_function, inline_summary_t::insert,
17397 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17398 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17399 inline_free_summary): Move to ipa-fnsummary.h
17400 (predicate_t): Remove.
17401 * ipa-fnsummary.c: New file.
17402 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17403 (enum inline_hints_vals, inline_hints, agg_position_info,
17404 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17405 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17406 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17407 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17408 inline_read_summary, inline_write_summary, inline_free_summary,
17409 inline_analyze_function, initialize_inline_failed,
17410 inline_merge_summary, inline_update_overall_summary,
17411 compute_inline_parameters): Move to ipa-fnsummary.h
17412 * ipa-fnsummary.h: New file.
17413 * ipa-inline-transform.h: Include ipa-inline.h.
17414 * ipa-inline.c: LIkewise.
17416 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17418 * ipa-inline.c (edge_badness): Use inlined_time instead of
17419 inline_summaries->get.
17421 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17423 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17425 2017-05-22 Nathan Sidwell <nathan@acm.org>
17427 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17428 (fdump-lang): Document 'raw' option.
17429 * dumpfile.h (TDI_tu): Delete.
17430 * dumpfile.c (dump_files): Remove translation-unit.
17431 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17433 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17435 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17436 command option from $(AWK) call.
17437 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17438 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17439 [FORMAT]: Remove handling of variable.
17440 * config/avr/t-multilib: Regenerate.
17442 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17444 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17446 (dump_inline_summary): Do not print self_time.
17447 (estimate_function_body_sizes): Do not set self_time.
17448 (compute_inline_parameters): Likewise.
17449 (inline_read_section, inline_write_summary): Do not stream self_time.
17450 * ipa-inline.h (inline_summary): Drop self_time.
17452 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17454 * ipa-inline-analysis.c (account_size_time): Rename to ...
17455 (inline_summary::account_size_time): ... this one.
17456 (reset_ipa_call_summary): Turn to ...
17457 (ipa_call_summary::reset): ... this one.
17458 (reset_inline_summary): Turn to ...
17459 (inline_summary::reset): ... this one.
17460 (inline_summary_t::remove): Update.
17461 (inline_summary_t::duplicate): Update.
17462 (ipa_call_summary_t::remove): Update.
17463 (dump_inline_summary): Update.
17464 (estimate_function_body_sizes): Update.
17465 (compute_inline_parameters): Update.
17466 (estimate_node_size_and_time): Update.
17467 (inline_merge_summary): Update.
17468 (inline_update_overall_summary): Update.
17469 (inline_read_section): Update.
17470 (inline_write_summary): Update.
17471 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17472 add account_size_time and reset member functions.
17473 (ipa_call_summary): Add reset function.
17474 * ipa-predicate.h (predicate::operator &): Constify.
17476 2017-05-22 Richard Biener <rguenther@suse.de>
17478 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17480 2017-05-19 Jason Merrill <jason@redhat.com>
17482 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17484 2017-05-19 Marek Polacek <polacek@redhat.com>
17487 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17488 TYPE_OVERFLOW_WRAPS checks.
17490 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17492 * tree-core.h (enum omp_clause_default_kind): Add
17493 "OMP_CLAUSE_DEFAULT_PRESENT".
17494 * tree-pretty-print.c (dump_omp_clause): Handle it.
17495 * gimplify.c (enum gimplify_omp_var_data): Add
17496 "GOVD_MAP_FORCE_PRESENT".
17497 (gimplify_adjust_omp_clauses_1): Map it to
17498 "GOMP_MAP_FORCE_PRESENT".
17499 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17501 * gimplify.c (oacc_default_clause): Clarify.
17503 2017-05-19 Nathan Sidwell <nathan@acm.org>
17505 LANG_HOOK_REGISTER_DUMPS
17506 * toplev.c (general_init): Call register dump lang hook.
17507 * doc/invoke.texi: Document -fdump-lang option family.
17508 * dumpfile.c (dump_files): Remove class dump here.
17509 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17510 * dumpfile.h (tree_dump_index): Remove TDI_class.
17511 * langhooks-def.h (lhd_register_dumps): Declare.
17512 (LANG_HOOKS_REGISTER_DUMPS): Define.
17513 (LANG_HOOKS_INITIALIZER): Add it.
17514 * langhooks.c (lhd_register_dumps): Define.
17515 * langhooks.h (struct lang_hooks): Add register_dumps.
17517 2017-05-19 Nathan Sidwell <nathan@acm.org>
17519 * context.h (context::set_passes): New.
17520 * context.c (context::context): Do not create pass manager.
17521 * toplev.c (general_init): Create pass manager here.
17523 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17525 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17526 use this splitter if two add or or instructions would also work for
17527 the constant we want to generate.
17529 2017-05-19 Richard Biener <rguenther@suse.de>
17532 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17533 predicate evaluation.
17535 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17537 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17539 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17542 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17544 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17545 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17546 (inline_edge_summary_vec): Turn into ...
17547 (ipa_call_summaries): ... this one.
17548 (redirect_to_unreachable, edge_set_predicate,
17549 evaluate_properties_for_edge, inline_summary_alloc,
17550 reset_ipa_call_summary, reset_inline_summary,
17551 inline_summary_t::duplicate): Update.
17552 (inline_edge_duplication_hook): Turn to ...
17553 (ipa_call_summary_t::duplicate): ... this one.
17554 (inline_edge_removal_hook): Turn to ...
17555 (ipa_call_summary_t::remove): ... this one.
17556 (dump_inline_edge_summary): Turn to ...
17557 (dump_ipa_call_summary): ... this one.
17558 (estimate_function_body_sizes): Update.
17559 (inline_update_callee_summaries): Update.
17560 (remap_edge_change_prob): Update.
17561 (remap_edge_summaries): Update.
17562 (inline_merge_summary): Update.
17563 (do_estimate_edge_time): Update.
17564 (inline_generate_summary): Update.
17565 (inline_read_section): Update.
17566 (inline_read_summary): Update.
17567 (inline_free_summary): Update.
17568 * ipa-inline.c (can_inline_edge_p): Update.
17569 (compute_inlined_call_time): Update.
17570 (want_inline_small_function_p): Update.
17571 (edge_badness): Update.
17572 (early_inliner): Update.
17573 * ipa-inline.h (inline_edge_summary): Turn to ...
17574 (ipa_call_summary): ... this one.
17575 (ipa_call_summary_t): New class.
17576 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17577 (ipa_call_summaries): New.
17578 (inline_edge_summary): Remove.
17579 (estimate_edge_growth): Update.
17580 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17581 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17582 * ipa-split.c (execute_split_functions): Update.
17583 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17585 2017-05-19 Richard Biener <rguenther@suse.de>
17587 PR middle-end/80764
17588 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17590 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17592 * config/rs6000/rs6000.c (struct machine_function): Add field
17593 fpr_is_wrapped_separately.
17594 (rs6000_get_separate_components): Use 64 components. Handle the
17595 new FPR components.
17596 (rs6000_components_for_bb): Handle the FPR components.
17597 (rs6000_emit_prologue_components): Handle the FPR components.
17598 (rs6000_emit_epilogue_components): Handle the FPR components.
17599 (rs6000_set_handled_components): Handle the FPR components.
17600 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17601 that are already separately shrink-wrapped.
17602 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17603 that are already separately shrink-wrapped.
17605 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17608 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17611 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17612 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17613 cases where the register allocator uses a move and an offsettable
17614 memory operation to/from a FPR register on ISA 2.06/2.07.
17615 (define_peephole2 for Altivec d-form store): Likewise.
17617 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17620 * config/i386/mmx.md (*mov<mode>_internal): Enable
17621 alternatives 11, 12, 13 and 14 also for 32bit targets.
17622 Remove alternatives 15, 16, 17 and 18.
17623 * config/i386/sse.md (vec_concatv2di): Change
17624 alternative (!x, *y) to (x, ?!*Yn).
17626 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17628 * dumpfile.h (enum dump_kind): Remove stray comma.
17630 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17632 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17633 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17634 predicate::num_conditions
17635 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17636 (CHANGED): turn into predicate::changed.
17637 (agg_position_info): Move to ipa-predicate.h
17638 (add_condition, predicate::add_clause, predicate::operator &=,
17639 predicate::or_with, predicate::evaluate, predicate::probability,
17640 dump_condition, dump_clause, predicate::dump,
17641 predicate::remap_after_duplication, predicate::remap_after_inlining,
17642 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17643 (evaluate_conditions_for_known_args): Update.
17644 (set_cond_stmt_execution_predicate): Update.
17645 * ipa-inline.h: Include ipa-predicate.h
17646 (condition, inline_param_summary, conditions, agg_position_info,
17647 predicate): Move to ipa-predicate.h
17648 * ipa-predicate.c: New file.
17649 * ipa-predicate.h: New file.
17651 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17653 * final.c (leaf_function_p): Check we are not in a sequence.
17655 2017-05-18 Martin Liska <mliska@suse.cz>
17657 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17658 * dumpfile.c (dump_register): Use new enum dump_kind.
17659 (get_dump_file_name): Likewise.
17660 (dump_enable_all): Likewise.
17661 (dump_switch_p_1): Likewise.
17662 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17663 * dumpfile.h (enum dump_kind): New enum type.
17664 (struct dump_file_info): Create constructor and
17665 format fields and comments.
17666 * passes.c (pass_manager::register_one_dump_file):
17668 * statistics.c (statistics_early_init): Likewise.
17669 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17670 TDF_TREE with TDF_SLIM.
17671 (gather_memory_references_ref): Likewise.
17673 2017-05-18 Martin Liska <mliska@suse.cz>
17675 * vec.h (struct vnull): Use it.
17677 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17679 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17680 (true_predicate, false_predicate, true_predicate_p,
17681 false_predicate_p): Remove.
17682 (single_cond_predicate, not_inlined_predicate): Turn to member function
17684 (add_condition): Update.
17685 (add_clause): Turn to...
17686 (predicate::add_clause): ... this one; update; allow passing NULL
17688 (and_predicates): Turn to ...
17689 (predicate::operator &=): ... this one.
17690 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17691 (or_predicates): Turn to ...
17692 (predicate::or_with): ... this one.
17693 (evaluate_predicate): Turn to ...
17694 (predicate::evaluate): ... this one.
17695 (predicate_probability): Turn to ...
17696 (predicate::probability): ... this one.
17697 (dump_condition): Update.
17698 (dump_predicate): Turn to ...
17699 (predicate::dump): ... this one.
17700 (account_size_time): Update.
17701 (edge_set_predicate): Update.
17702 (set_hint_predicate): UPdate.
17703 (evaluate_conditions_for_known_args): Update.
17704 (evaluate_properties_for_edge): Update.
17705 (remap_predicate_after_duplication): Turn to...
17706 (predicate::remap_after_duplication): ... this one.
17707 (remap_hint_predicate_after_duplication): Update.
17708 (inline_summary_t::duplicate): UPdate.
17709 (dump_inline_edge_summary): Update.
17710 (dump_inline_summary): Update.
17711 (set_cond_stmt_execution_predicate): Update.
17712 (set_switch_stmt_execution_predicate): Update.
17713 (compute_bb_predicates): Update.
17714 (will_be_nonconstant_expr_predicate): Update.
17715 (will_be_nonconstant_predicate): Update.
17716 (phi_result_unknown_predicate): Update.
17717 (predicate_for_phi_result): Update.
17718 (array_index_predicate): Update.
17719 (estimate_function_body_sizes): Update.
17720 (estimate_node_size_and_time): Update.
17721 (estimate_ipcp_clone_size_and_time): Update.
17722 (remap_predicate): Rename to ...
17723 (predicate::remap_after_inlining): ... this one.
17724 (remap_hint_predicate): Update.
17725 (inline_merge_summary): Update.
17726 (inline_update_overall_summary): Update.
17727 (estimate_size_after_inlining): Update.
17728 (read_predicate): Rename to ...
17729 (predicate::stream_in): ... this one.
17730 (read_inline_edge_summary): Update.
17731 (write_predicate): Rename to ...
17732 (predicate::stream_out): ... this one.
17733 (write_inline_edge_summary): Update.
17734 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17735 (clause_t): Turn to uint32_t
17736 (predicate): Turn to class; implement constructor and operators
17738 (size_time_entry): Update.
17739 (inline_summary): Update.
17740 (inline_edge_summary): Update.
17742 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17744 * fold-const.c (fold_binary_loc): Move transformation...
17745 * match.pd (C - X CMP X): ... here.
17747 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17749 * config/sparc/sparc.c (sparc_option_override): Set function
17750 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17751 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17753 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17755 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17757 2017-05-18 Marek Polacek <polacek@redhat.com>
17760 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17761 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17762 gimple_assign_load_p.
17764 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17766 PR middle-end/80692
17767 * real.c (do_compare): Give decimal_do_compare preference over
17768 comparing just the signs.
17770 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17772 * doc/md.texi (Canonicalization of Instructions): Describe the
17773 canonical form of instructions that inherently set a condition
17776 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17778 PR middle-end/80775
17779 * tree-cfg.c: Move deletion of unreachable case statements to after
17780 the merging of consecutive case labels.
17782 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17784 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17785 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17786 restoring of callee-saved registers.
17788 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17790 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17791 * config/visium/visium.c (single_set_and_flags): Likewise.
17792 * config/visium/visium.md (Substitutions): Likewise.
17794 2017-05-17 Martin Liska <mliska@suse.cz>
17796 * cfg.c: Introduce dump_flags_t type and
17797 use it instead of int type.
17799 * cfghooks.c: Likewise.
17800 * cfghooks.h (struct cfg_hooks): Likewise.
17801 * cfgrtl.c: Likewise.
17802 * cfgrtl.h: Likewise.
17803 * cgraph.c (cgraph_node::get_body): Likewise.
17804 * coretypes.h: Likewise.
17805 * domwalk.c: Likewise.
17806 * domwalk.h: Likewise.
17807 * dumpfile.c (struct dump_option_value_info): Likewise.
17808 (dump_enable_all): Likewise.
17809 (dump_switch_p_1): Likewise.
17810 (opt_info_switch_p): Likewise.
17811 * dumpfile.h (enum tree_dump_index): Likewise.
17812 (struct dump_file_info): Likewise.
17813 * genemit.c: Likewise.
17814 * generic-match-head.c: Likewise.
17815 * gengtype.c (open_base_files): Likewise.
17816 * gimple-pretty-print.c: Likewise.
17817 * gimple-pretty-print.h: Likewise.
17818 * graph.c (print_graph_cfg): Likewise.
17819 * graphite-scop-detection.c (dot_all_sese): Likewise.
17820 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17821 * loop-unroll.c (report_unroll): Likewise.
17822 * passes.c (pass_manager::register_one_dump_file): Likewise.
17823 * print-tree.c: Likewise.
17824 * statistics.c: Likewise.
17825 * tree-cfg.c: Likewise.
17826 * tree-cfg.h: Likewise.
17827 * tree-dfa.c: Likewise.
17828 * tree-dfa.h: Likewise.
17829 * tree-dump.c (dump_function): Likewise.
17830 * tree-dump.h (struct dump_info): Likewise.
17831 * tree-pretty-print.c: Likewise.
17832 * tree-pretty-print.h: Likewise.
17833 * tree-ssa-live.c: Likewise.
17834 * tree-ssa-live.h: Likewise.
17835 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
17836 * tree-vect-loop.c: Likewise.
17837 * tree-vect-slp.c: Likewise.
17839 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
17840 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17842 PR tree-optimization/80457
17843 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
17844 of all arguments to a statement as scalar_to_vec operations.
17845 (vectorizable_call): Adjust call to vect_model_simple_cost for
17847 (vectorizable_conversion): Likewise.
17848 (vectorizable_assignment): Likewise.
17849 (vectorizable_shift): Likewise.
17850 (vectorizable_operation): Likewise.
17851 (vectorizable_comparison): Likewise.
17852 (vect_is_simple_cond): Record the def types for operands.
17853 (vectorizable_condition): Likewise, call vect_model_simple_cost.
17854 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
17855 for statement argument count.
17857 2017-05-16 Carl Love <cel@us.ibm.com>
17859 * config/rs6000/rs6000-c: Add support for built-in functions
17860 vector unsigned long long vec_bperm (vector unsigned long long,
17861 vector unsigned char)
17862 vector signed long long vec_mule (vector signed int,
17864 vector unsigned long long vec_mule (vector unsigned int,
17865 vector unsigned int)
17866 vector signed long long vec_mulo (vector signed int,
17868 vector unsigned long long vec_mulo (vector unsigned int,
17869 vector unsigned int)
17870 vector signed char vec_sldw (vector signed char,
17871 vector signed char,
17873 vector unsigned char vec_sldw (vector unsigned char,
17874 vector unsigned char,
17876 vector signed short vec_sldw (vector signed short,
17877 vector signed short,
17879 vector unsigned short vec_sldw (vector unsigned short,
17880 vector unsigned short,
17882 vector signed int vec_sldw (vector signed int,
17885 vector unsigned int vec_sldw (vector unsigned int,
17886 vector unsigned int,
17888 vector signed long long vec_sldw (vector signed long long,
17889 vector signed long long,
17891 vector unsigned long long vec_sldw (vector unsigned long long,
17892 vector unsigned long long,
17894 * config/rs6000/rs6000-c: Add support for built-in functions
17895 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
17896 * config/rs6000/altivec.h: Add defintion for vec_sldw.
17897 * doc/extend.texi: Update the built-in documentation for the
17898 new built-in functions.
17900 2017-05-16 Marek Polacek <polacek@redhat.com>
17904 * tree.c (save_expr): Don't fold the expression.
17906 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
17908 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
17909 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
17910 and (?*y,m). Update insn attributes.
17912 2017-05-16 Martin Liska <mliska@suse.cz>
17914 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
17915 flags argument of print_gimple_stmt, print_gimple_expr,
17916 print_generic_stmt and print_generic_expr.
17917 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
17918 * coretypes.h: Likewise.
17919 * except.c (dump_eh_tree): Likewise.
17920 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
17921 * gimple-pretty-print.h: Likewise.
17922 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
17923 (backprop::push_to_worklist): Likewise.
17924 (backprop::pop_from_worklist): Likewise.
17925 (backprop::process_use): Likewise.
17926 (backprop::intersect_uses): Likewise.
17927 (note_replacement): Likewise.
17928 * gimple-ssa-store-merging.c
17929 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
17930 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
17931 (pass_store_merging::execute): Likewise.
17932 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
17933 (ssa_base_cand_dump_callback): Likewise.
17934 (dump_incr_vec): Likewise.
17935 (replace_refs): Likewise.
17936 (replace_mult_candidate): Likewise.
17937 (create_add_on_incoming_edge): Likewise.
17938 (create_phi_basis): Likewise.
17939 (insert_initializers): Likewise.
17940 (all_phi_incrs_profitable): Likewise.
17941 (introduce_cast_before_cand): Likewise.
17942 (replace_one_candidate): Likewise.
17943 * gimplify.c (gimplify_expr): Likewise.
17944 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
17945 (set_rename): Likewise.
17946 (rename_uses): Likewise.
17947 (copy_loop_phi_nodes): Likewise.
17948 (add_close_phis_to_merge_points): Likewise.
17949 (copy_loop_close_phi_args): Likewise.
17950 (copy_cond_phi_args): Likewise.
17951 (graphite_copy_stmts_from_block): Likewise.
17952 (translate_pending_phi_nodes): Likewise.
17953 * graphite-poly.c (print_pdr): Likewise.
17954 (dump_gbb_cases): Likewise.
17955 (dump_gbb_conditions): Likewise.
17956 (print_scop_params): Likewise.
17957 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
17958 (build_cross_bb_scalars_use): Likewise.
17959 (gather_bbs::before_dom_children): Likewise.
17960 * hsa-dump.c (dump_hsa_immed): Likewise.
17961 * ipa-cp.c (print_ipcp_constant_value): Likewise.
17962 (get_replacement_map): Likewise.
17963 * ipa-inline-analysis.c (dump_condition): Likewise.
17964 (estimate_function_body_sizes): Likewise.
17965 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
17966 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
17967 * ipa-prop.c (ipa_dump_param): Likewise.
17968 (ipa_print_node_jump_functions_for_edge): Likewise.
17969 (ipa_modify_call_arguments): Likewise.
17970 (ipa_modify_expr): Likewise.
17971 (ipa_dump_param_adjustments): Likewise.
17972 (ipa_dump_agg_replacement_values): Likewise.
17973 (ipcp_modif_dom_walker::before_dom_children): Likewise.
17974 * ipa-pure-const.c (check_stmt): Likewise.
17975 (pass_nothrow::execute): Likewise.
17976 * ipa-split.c (execute_split_functions): Likewise.
17977 * omp-offload.c (dump_oacc_loop_part): Likewise.
17978 (dump_oacc_loop): Likewise.
17979 * trans-mem.c (tm_log_emit): Likewise.
17980 (tm_memopt_accumulate_memops): Likewise.
17981 (dump_tm_memopt_set): Likewise.
17982 (dump_tm_memopt_transform): Likewise.
17983 * tree-cfg.c (gimple_verify_flow_info): Likewise.
17984 (print_loop): Likewise.
17985 * tree-chkp-opt.c (chkp_print_addr): Likewise.
17986 (chkp_gather_checks_info): Likewise.
17987 (chkp_get_check_result): Likewise.
17988 (chkp_remove_check_if_pass): Likewise.
17989 (chkp_use_outer_bounds_if_possible): Likewise.
17990 (chkp_reduce_bounds_lifetime): Likewise.
17991 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
17992 (chkp_mark_completed_bounds): Likewise.
17993 (chkp_register_incomplete_bounds): Likewise.
17994 (chkp_mark_invalid_bounds): Likewise.
17995 (chkp_maybe_copy_and_register_bounds): Likewise.
17996 (chkp_build_returned_bound): Likewise.
17997 (chkp_get_bound_for_parm): Likewise.
17998 (chkp_build_bndldx): Likewise.
17999 (chkp_get_bounds_by_definition): Likewise.
18000 (chkp_generate_extern_var_bounds): Likewise.
18001 (chkp_get_bounds_for_decl_addr): Likewise.
18002 * tree-chrec.c (chrec_apply): Likewise.
18003 * tree-data-ref.c (dump_data_reference): Likewise.
18004 (dump_subscript): Likewise.
18005 (dump_data_dependence_relation): Likewise.
18006 (analyze_overlapping_iterations): Likewise.
18007 * tree-inline.c (expand_call_inline): Likewise.
18008 (tree_function_versioning): Likewise.
18009 * tree-into-ssa.c (dump_defs_stack): Likewise.
18010 (dump_currdefs): Likewise.
18011 (dump_names_replaced_by): Likewise.
18012 (dump_update_ssa): Likewise.
18013 (update_ssa): Likewise.
18014 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18015 * tree-parloops.c (build_new_reduction): Likewise.
18016 (try_create_reduction_list): Likewise.
18017 (ref_conflicts_with_region): Likewise.
18018 (oacc_entry_exit_ok_1): Likewise.
18019 (oacc_entry_exit_single_gang): Likewise.
18020 * tree-pretty-print.h: Likewise.
18021 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18022 (get_scalar_evolution): Likewise.
18023 (add_to_evolution): Likewise.
18024 (get_loop_exit_condition): Likewise.
18025 (analyze_evolution_in_loop): Likewise.
18026 (analyze_initial_condition): Likewise.
18027 (analyze_scalar_evolution): Likewise.
18028 (instantiate_scev): Likewise.
18029 (number_of_latch_executions): Likewise.
18030 (gather_chrec_stats): Likewise.
18031 (final_value_replacement_loop): Likewise.
18032 (scev_const_prop): Likewise.
18033 * tree-sra.c (dump_access): Likewise.
18034 (disqualify_candidate): Likewise.
18035 (create_access): Likewise.
18036 (reject): Likewise.
18037 (maybe_add_sra_candidate): Likewise.
18038 (create_access_replacement): Likewise.
18039 (analyze_access_subtree): Likewise.
18040 (analyze_all_variable_accesses): Likewise.
18041 (sra_modify_assign): Likewise.
18042 (initialize_constant_pool_replacements): Likewise.
18043 (find_param_candidates): Likewise.
18044 (decide_one_param_reduction): Likewise.
18045 (replace_removed_params_ssa_names): Likewise.
18046 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18047 * tree-ssa-copy.c (dump_copy_of): Likewise.
18048 (copy_prop_visit_cond_stmt): Likewise.
18049 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18050 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18051 (record_equivalences_from_stmt): Likewise.
18052 * tree-ssa-dse.c (compute_trims): Likewise.
18053 (delete_dead_call): Likewise.
18054 (delete_dead_assignment): Likewise.
18055 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18056 (forward_propagate_into_cond): Likewise.
18057 (pass_forwprop::execute): Likewise.
18058 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18059 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18061 (move_computations_worker): Likewise.
18062 (execute_sm): Likewise.
18063 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18064 (remove_exits_and_undefined_stmts): Likewise.
18065 (remove_redundant_iv_tests): Likewise.
18066 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18067 (adjust_iv_update_pos): Likewise.
18068 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18069 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18070 (value_replacement): Likewise.
18071 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18072 * tree-ssa-pre.c (print_pre_expr): Likewise.
18073 (get_representative_for): Likewise.
18074 (create_expression_by_pieces): Likewise.
18075 (insert_into_preds_of_block): Likewise.
18076 (eliminate_insert): Likewise.
18077 (eliminate_dom_walker::before_dom_children): Likewise.
18078 (eliminate): Likewise.
18079 (remove_dead_inserted_code): Likewise.
18080 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18081 * tree-ssa-reassoc.c (get_rank): Likewise.
18082 (eliminate_duplicate_pair): Likewise.
18083 (eliminate_plus_minus_pair): Likewise.
18084 (eliminate_not_pairs): Likewise.
18085 (undistribute_ops_list): Likewise.
18086 (eliminate_redundant_comparison): Likewise.
18087 (update_range_test): Likewise.
18088 (optimize_range_tests_var_bound): Likewise.
18089 (optimize_vec_cond_expr): Likewise.
18090 (rewrite_expr_tree): Likewise.
18091 (rewrite_expr_tree_parallel): Likewise.
18092 (linearize_expr): Likewise.
18093 (break_up_subtract): Likewise.
18094 (linearize_expr_tree): Likewise.
18095 (attempt_builtin_powi): Likewise.
18096 (attempt_builtin_copysign): Likewise.
18097 (transform_stmt_to_copy): Likewise.
18098 (transform_stmt_to_multiply): Likewise.
18099 (dump_ops_vector): Likewise.
18100 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18101 (print_scc): Likewise.
18102 (set_ssa_val_to): Likewise.
18103 (visit_reference_op_store): Likewise.
18104 (visit_use): Likewise.
18105 (sccvn_dom_walker::before_dom_children): Likewise.
18106 (run_scc_vn): Likewise.
18107 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18109 (expr_hash_elt::print): Likewise.
18110 (const_and_copies::pop_to_marker): Likewise.
18111 (const_and_copies::record_const_or_copy_raw): Likewise.
18112 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18113 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18114 (dump_predicates): Likewise.
18115 (find_uninit_use): Likewise.
18116 (warn_uninitialized_phi): Likewise.
18117 (pass_late_warn_uninitialized::execute): Likewise.
18118 * tree-ssa.c (verify_vssa): Likewise.
18119 (verify_ssa): Likewise.
18120 (maybe_optimize_var): Likewise.
18121 * tree-vrp.c (dump_value_range): Likewise.
18122 (dump_all_value_ranges): Likewise.
18123 (dump_asserts_for): Likewise.
18124 (register_edge_assert_for_2): Likewise.
18125 (vrp_visit_cond_stmt): Likewise.
18126 (vrp_visit_switch_stmt): Likewise.
18127 (vrp_visit_stmt): Likewise.
18128 (vrp_visit_phi_node): Likewise.
18129 (simplify_cond_using_ranges_1): Likewise.
18130 (fold_predicate_in): Likewise.
18131 (evrp_dom_walker::before_dom_children): Likewise.
18132 (evrp_dom_walker::push_value_range): Likewise.
18133 (evrp_dom_walker::pop_value_range): Likewise.
18134 (execute_early_vrp): Likewise.
18136 2017-05-16 Richard Biener <rguenther@suse.de>
18138 * dwarf2out.c (loc_list_from_tree_1): Do not create
18139 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18141 2017-05-16 Richard Biener <rguenther@suse.de>
18143 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18145 (note_variable_value_in_expr): If we resolved the decl ref
18146 do not push to the stack.
18148 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18150 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18151 operations in fast-math mode.
18152 (vaddq_f16): Likewise.
18153 (vmul_f16): Likewise.
18154 (vmulq_f16): Likewise.
18155 (vsub_f16): Likewise.
18156 (vsubq_f16): Likewise.
18157 * config/arm/neon.md (add<mode>3): New.
18159 (fma:<VH:mode>3): New. Also remove outdated comment.
18162 2017-05-16 Martin Liska <mliska@suse.cz>
18166 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18167 (odr_types_equivalent_p): Likewise.
18169 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18171 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18173 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18176 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18177 non-interunit SSE move alternatives with '?'.
18178 (zero-extendsidi peephole2): New peephole to skip intermediate
18179 general register in SSE zero-extend sequence.
18181 2017-05-15 Jeff Law <law@redhat.com>
18183 * reorg.c (relax_delay_slots): Create a new variable to hold
18184 the temporary target rather than clobbering TARGET_LABEL.
18186 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18187 missing argument to extract_bit_field call.
18188 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18190 2017-05-15 Martin Liska <mliska@suse.cz>
18193 * gcc.c (process_command): Do not allow empty argument of -o option.
18195 2017-05-15 Renlin Li <renlin.li@arm.com>
18197 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18198 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18199 * config/aarch64/constraints.md (Usf): Add long call check.
18200 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18201 (call_value): Likewise.
18202 (sibcall): Likewise.
18203 (sibcall_value): Likewise.
18205 (call_value_insn): New.
18206 (sibcall_insn): Update rtx pattern.
18207 (sibcall_value_insn): Likewise.
18208 (call_internal): Remove.
18209 (call_value_internal): Likewise.
18210 (sibcall_internal): Likewise.
18211 (sibcall_value_internal): Likewise.
18212 (call_reg): Likewise.
18213 (call_symbol): Likewise.
18214 (call_value_reg): Likewise.
18215 (call_value_symbol): Likewise.
18217 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18220 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18222 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18224 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18225 compatible with CCGOCmode and with CCZmode.
18227 2017-05-14 Martin Sebor <msebor@redhat.com>
18229 PR middle-end/77671
18230 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18231 (gimple_fold_builtin_snprintf): Same.
18232 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18233 (gimple_fold_builtin_snprintf): Same.
18234 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18235 of character types.
18236 (is_call_safe): New function.
18237 (try_substitute_return_value): Call it.
18238 (try_simplify_call): New function.
18239 (pass_sprintf_length::handle_gimple_call): Call it.
18241 2017-05-14 Martin Sebor <msebor@redhat.com>
18243 PR middle-end/80669
18244 * builtins.c (expand_builtin_stpncpy): Simplify.
18246 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18248 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18249 * config/i386/i386.h
18250 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18251 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18252 (struct machine_function): Add new members call_ms2sysv,
18253 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18254 (struct machine_frame_state): New fields sp_realigned and
18255 sp_realigned_offset.
18256 * config/i386/i386.c
18257 (enum xlogue_stub): New enum.
18258 (enum xlogue_stub_sets): New enum.
18259 (class xlogue_layout): New class.
18260 (struct ix86_frame): New fields stack_realign_allocate_offset,
18261 stack_realign_offset and outlined_save_offset. Modify comments to
18262 detail stack layout when using out-of-line stubs.
18263 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18264 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18265 -mcall-ms2sysv-xlogues.
18266 (stub_managed_regs): New static variable.
18267 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18268 registers managed by out-of-line stub.
18269 (disable_call_ms2sysv_xlogues): New function.
18270 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18271 m->call_ms2sysv when appropriate and compute frame layout for
18273 (sp_valid_at, fp_valid_at): New inline functions.
18274 (choose_basereg): New function.
18275 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18277 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18278 Use align parameter of choose_baseaddr to generated aligned SSE movs
18280 (pro_epilogue_adjust_stack): Modify to track
18281 machine_frame_state::sp_realigned.
18282 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18283 (ix86_nsaved_sseregs): Likewise.
18284 (ix86_emit_save_regs): Likewise.
18285 (ix86_emit_save_regs_using_mov): Likewise.
18286 (ix86_emit_save_sse_regs_using_mov): Likewise.
18287 (get_scratch_register_on_entry): Likewise.
18288 (gen_frame_set): New function.
18289 (gen_frame_load): Likewise.
18290 (gen_frame_store): Likewise.
18291 (emit_outlined_ms2sysv_save): Likewise.
18292 (emit_outlined_ms2sysv_restore): Likewise.
18293 (ix86_expand_prologue): Modify stack re-alignment code and call
18294 emit_outlined_ms2sysv_save when appropriate.
18295 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18296 parameter rtx_insn *insn, which allows the function to be used to only
18297 generate the notes.
18298 (ix86_expand_epilogue): Modify validity checks of frame and stack
18299 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18300 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18301 * config/i386/predicates.md
18302 (save_multiple): New predicate.
18303 (restore_multiple): Likewise.
18304 * config/i386/sse.md
18305 (save_multiple<mode>): New pattern.
18306 (save_multiple_realign<mode>): Likewise.
18307 (restore_multiple<mode>): Likewise.
18308 (restore_multiple_and_return<mode>): Likewise.
18309 (restore_multiple_leave_return<mode>): Likewise.
18310 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18312 2017-05-14 Julia Koval <julia.koval@intel.com>
18314 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18315 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18316 (__builtin_ia32_xsetbv): New builtins.
18317 * config/i386/i386.c (ix86_expand_special_args_builtin):
18319 (ix86_expand_builtin): Special expand for new intrinsics.
18320 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18321 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18322 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18324 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18326 * cfganal.c (inverted_post_order_compute): Change argument type
18328 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18329 * df-core.c (rest_of_handle_df_initialize): Adjust.
18330 (rest_of_handle_df_finish): Likewise.
18331 (df_analyze_1): Likewise.
18332 (df_analyze): Likewise.
18333 (loop_inverted_post_order_compute): Change argument to be a vec *.
18334 (df_analyze_loop): Adjust.
18335 (df_get_n_blocks): Likewise.
18336 (df_get_postorder): Likewise.
18337 * df.h (struct df_d): Change field to be a vec.
18338 * lcm.c (compute_laterin): Adjust.
18339 (compute_available): Likewise.
18340 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18341 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18342 * tree-ssa-pre.c (compute_antic): Likewise.
18344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18346 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18347 (depth_first_search::depth_first_search): Change structure init
18348 function to this constructor.
18349 (depth_first_search::add_bb): Rename function to this member.
18350 (depth_first_search::execute): Likewise.
18351 (flow_dfs_compute_reverse_finish): Adjust.
18353 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18355 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18356 (longest_simple_path): Likewise.
18357 * shrink-wrap.c (spread_components): Likewise.
18358 (disqualify_problematic_components): Likewise.
18359 (emit_common_heads_for_components): Likewise.
18360 (emit_common_tails_for_components): Likewise.
18361 (insert_prologue_epilogue_for_components): Likewise.
18363 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18365 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18368 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18370 * df-core.c (df_set_blocks): Start using auto_bitmap.
18371 (df_compact_blocks): Likewise.
18372 * df-problems.c (df_rd_confluence_n): Likewise.
18373 * df-scan.c (df_insn_rescan_all): Likewise.
18374 (df_process_deferred_rescans): Likewise.
18375 (df_update_entry_block_defs): Likewise.
18376 (df_update_exit_block_uses): Likewise.
18377 (df_entry_block_bitmap_verify): Likewise.
18378 (df_exit_block_bitmap_verify): Likewise.
18379 (df_scan_verify): Likewise.
18380 * lra-constraints.c (lra_constraints): Likewise.
18381 (undo_optional_reloads): Likewise.
18382 (lra_undo_inheritance): Likewise.
18383 * lra-remat.c (calculate_gen_cands): Likewise.
18384 (do_remat): Likewise.
18385 * lra-spills.c (assign_spill_hard_regs): Likewise.
18386 (spill_pseudos): Likewise.
18387 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18388 (bitmap_set_subtract_values): Likewise.
18390 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18392 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18393 management with auto_bitmap.
18394 (fix_inter_tick): Likewise.
18395 (fix_recovery_deps): Likewise.
18396 * ira.c (add_store_equivs): Likewise.
18397 (find_moveable_pseudos): Likewise.
18398 (split_live_ranges_for_shrink_wrap): Likewise.
18399 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18400 (rtx_reuse_manager::seen_def_p): Likewise.
18401 (rtx_reuse_manager::set_seen_def): Likewise.
18402 * print-rtl.h (class rtx_reuse_manager): Likewise.
18404 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18406 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18408 (migrate_btr_def): Likewise.
18409 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18410 * df-core.c (loop_post_order_compute): Likewise.
18411 (loop_inverted_post_order_compute): Likewise.
18412 * hsa-common.h: Likewise.
18413 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18414 * init-regs.c (initialize_uninitialized_regs): Likewise.
18415 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18416 (inline_small_functions): Likewise.
18417 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18418 * ira.c (combine_and_move_insns): Likewise.
18419 (build_insn_chain): Likewise.
18420 * loop-invariant.c (find_invariants): Likewise.
18421 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18422 * predict.c (tree_predict_by_opcode): Likewise.
18423 (predict_paths_leading_to): Likewise.
18424 (predict_paths_leading_to_edge): Likewise.
18425 (estimate_loops_at_level): Likewise.
18426 (estimate_loops): Likewise.
18427 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18428 (spread_components): Likewise.
18429 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18430 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18431 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18432 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18433 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18434 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18435 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18436 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18437 (mark_threaded_blocks): Likewise.
18438 (thread_through_all_blocks): Likewise.
18439 * tree-ssa.c (verify_ssa): Likewise.
18440 (execute_update_addresses_taken): Likewise.
18441 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18443 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18445 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18447 (post_order_compute): Likewise.
18448 (inverted_post_order_compute): Likewise.
18449 (pre_and_rev_post_order_compute_fn): Likewise.
18451 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18453 * genrecog.c (int_set::int_set): Explicitly construct our
18454 auto_vec base class.
18455 * vec.h (auto_vec::auto_vec): New constructor.
18457 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18459 * bitmap.h (class auto_bitmap): New constructor taking
18460 bitmap_obstack * argument.
18462 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18464 * bitmap.h (class auto_bitmap): Change type of m_bits to
18465 bitmap_head, and adjust ctor / dtor and member operators.
18467 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18469 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18470 when returned register mode doesn't match original mode.
18472 2017-05-12 Jeff Law <law@redhat.com>
18473 Jakub Jelinek <jakub@redhat.com>
18475 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18476 we look for cc setter after the compare-elim changes.
18477 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18478 within the vector to match what compare-elim now expects.
18479 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18480 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18482 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18483 after the compare-elim changes.
18484 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18485 the vector to match what compare-elim now expects.
18486 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18487 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18488 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18489 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18490 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18492 * config/visium/visium.c (single_set_and_flags): Fix where
18493 we look for cc setter after the compare-elim changes.
18494 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18495 with the vector to match what compare-elim now expects.
18496 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18497 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18498 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18499 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18500 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18501 (neg<mode>2_insn_set_overflow): Likewise.
18503 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18505 PR middle-end/79794
18506 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18507 maybe_expand_insn call, set ops[0].target. If still set after call,
18508 set alt_rtl. Add extra arg to recursive calls.
18509 (extract_bit_field): Add alt_rtl argument. Pass to
18511 * expmed.h (extract_bit_field): Fix prototype.
18512 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18513 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18514 to extract_bit_field_calls.
18515 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18516 Pass alt_rtl to extract_bit_field calls.
18517 * calls.c (store_unaligned_arguments_into_psuedos)
18518 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18519 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18521 * optabs.h (struct expand_operand): Add target bitfield.
18523 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18525 * compare-elim.c (try_eliminate_compare): Canonicalize
18526 operation with embedded compare to
18527 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18528 (set (reg) (operation)].
18530 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18532 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18535 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18536 cost of adding a carry flag for ADC instruction.
18537 [case MINUS]: Ignore the cost of subtracting a carry flag
18538 for SBB instruction.
18540 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18542 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18544 * config/rs6000/bmiintrin.h: New file.
18545 * config/rs6000/bmi2intrin.h: New file.
18546 * config/rs6000/x86intrin.h: New file.
18548 2017-05-12 Jeff Law <law@redhat.com>
18550 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18553 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18555 PR middle-end/80707
18556 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18558 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18560 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18561 early expansion of vector divide builtins.
18562 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18563 builtins identified as having unsigned arguments.
18565 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18567 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18568 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18569 expansion of vector logical operations (and, andc, or, xor,
18572 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18574 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18575 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18577 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18579 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18580 early GIMPLE expansion of vector multiplies.
18582 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18584 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18585 TARGET_HAVE_MOVT conditional.
18586 (movt splitter): Likewise.
18588 2017-05-12 Richard Biener <rguenther@suse.de>
18590 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18591 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18592 Fold all stmts not inplace.
18594 2017-05-12 Richard Biener <rguenther@suse.de>
18596 PR tree-optimization/80713
18597 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18598 inserted_exprs bit for not removed stmts.
18600 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18602 PR middle-end/69921
18603 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18604 parallelized" attribute for parallelized OpenACC kernels.
18605 * omp-offload.c (execute_oacc_device_lower): Use it.
18607 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18608 Set "oacc kernels" attribute.
18609 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18610 parameter. Adjust all users.
18611 (oacc_fn_attrib_kernels_p): Remove function.
18612 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18613 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18614 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18615 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18616 assert "oacc kernels" attribute is set.
18618 2017-05-11 Carl Love <cel@us.ibm.com>
18620 * config/rs6000/rs6000-c: Add support for built-in functions
18621 vector unsigned char vec_popcnt (vector signed char)
18622 vector unsigned char vec_popcnt (vector unsigned char)
18623 vector unsigned short vec_popcnt (vector signed short)
18624 vector unsigned short vec_popcnt (vector unsigned short)
18625 vector unsigned int vec_popcnt (vector signed int)
18626 vector unsigned int vec_popcnt (vector unsigned int)
18627 vector unsigned long long vec_popcnt (vector signed long long)
18628 vector unsigned long long vec_popcnt (vector unsigned long long)
18629 vector signed long long vec_slo (vector signed long long,
18630 vector signed char)
18631 vector signed long long vec_slo (vector signed long long,
18632 vector unsigned char)
18633 vector unsigned long long vec_slo (vector unsigned long long,
18634 vector signed char)
18635 vector unsigned long long vec_slo (vector unsigned long long,
18636 vector unsigned char)
18637 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18638 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18639 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18640 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18641 * doc/extend.texi: Update the built-in documentation file for the
18642 new built-in functions.
18644 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18646 * attribs.h (sorted_attr_string): Move machine independent
18647 functions for target clone support from the i386 port to common
18648 code. Rename ix86_function_versions to common_function_versions.
18649 Rename make_name to make_unique_name.
18650 (common_function_versions): Likewise.
18651 (make_unique_name): Likewise.
18652 (make_dispatcher_decl): Likewise.
18653 (is_function_default_version): Likewise.
18654 * attribs.c (attr_strcmp): Likewise.
18655 (sorted_attr_string): Likewise.
18656 (common_function_versions): Likewise.
18657 (make_unique_name): Likewise.
18658 (make_dispatcher_decl): Likewise.
18659 (is_function_default_version): Likewise.
18660 * config/i386/i386.c (attr_strcmp): Likewise.
18661 (sorted_attr_string): Likewise.
18662 (ix86_function_versions): Likewise.
18663 (make_name): Likewise.
18664 (make_dispatcher_decl): Likewise.
18665 (is_function_default_version): Likewise.
18666 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18668 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18671 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18672 Account for direct move costs for vec_construct of integer
18675 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18678 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18679 (UNSPEC_STX_ATOMIC): Ditto.
18680 (loaddi_via_sse): New insn.
18681 (storedi_via_sse): Ditto.
18682 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18683 Update corresponding peephole2 patterns.
18684 (atomic_storedi_fpu): Ditto.
18686 2017-05-11 Julia Koval <julia.koval@intel.com>
18688 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18689 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18691 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18692 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18693 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18695 2017-05-11 Nathan Sidwell <nathan@acm.org>
18697 * graphite-poly.c: Include dumpfile.h.
18699 * dumpfle.h (dump_function): Declare here ...
18700 * tree-dump.h (dump_function): ... not here.
18701 * dumpfile.c: #include tree-cfg.h.
18702 (dump_function): Move here from ...
18703 * tree-dump.c (dump_function): ... here.
18704 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18705 * graphite-poly.c: Don't include tree-dump.h.
18706 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18707 * print-tree.c: Likewise.
18708 * stor-layout.c: Likewise.
18709 * tree-nested.c: Likewise.
18711 * dumpfile.c (dump_start): Use TDF_FLAGS.
18712 (dump_enable_all): Fix TDF_KIND check thinko.
18714 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18716 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18717 array entries to represent two legal parameterizations of the
18718 overloaded __builtin_cmpb function, as represented by the
18719 P6_OV_BUILTIN_CMPB constant.
18720 (altivec_resolve_overloaded_builtin): Add special case handling
18721 for the __builtin_cmpb function, as represented by the
18722 P6_OV_BUILTIN_CMPB constant.
18723 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18724 (BU_P6_64BIT_2): New macro.
18725 (BU_P6_OVERLOAD_2): New macro
18726 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18727 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18728 (CMPB): Add overload support to represent both 32-bit and 64-bit
18729 compare-bytes function.
18730 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18731 support for TARGET_CMPB.
18732 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18733 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18734 documentation of the __builtin_cmpb overloaded built-in function.
18736 2017-05-11 Richard Biener <rguenther@suse.de>
18738 PR tree-optimization/80705
18739 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18740 bases are not vectorizable.
18742 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18744 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18745 when counting register pressure.
18747 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18749 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18750 (iv_ca_more_deps): Renamed to ...
18751 (iv_ca_compare_deps): ... this.
18752 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18754 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18756 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18758 (determine_group_iv_costs): ... here.
18759 (find_inv_vars_cb): Record inv var if it's not recorded before.
18761 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18763 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18764 (get_shiftadd_cost): Ditto.
18766 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18768 * tree-ssa-address.c: Include header file.
18769 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18771 (add_to_parts): Refactor.
18772 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18773 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18776 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18778 PR tree-optimization/53090
18779 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18781 (extract_cond_operands): Detect condition with IV on both sides
18782 and return COMP_IV_EXPR_2.
18783 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18784 (rewrite_use_compare): Simplify by removing call to function
18785 extract_cond_operands.
18787 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18789 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18790 (extract_cond_operands): Detect condition comparing against non-
18791 invariant bound and return appropriate enum value.
18792 (find_interesting_uses_cond): Update use of extract_cond_operands.
18793 Handle its return value accordingly.
18794 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18796 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18798 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18799 nonlinear iv_use computation in loop invariant sensitive way.
18801 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18803 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18804 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18808 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18809 (dump_cand): Support iv_cand.inv_exprs.
18810 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18812 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18815 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18817 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18819 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18820 as local function. Include necessary header files.
18821 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18823 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18825 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18829 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18830 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
18831 RSHIFT_EXPR and BIT_NOT_EXPR.
18833 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18835 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
18836 (adjust_setup_cost): New parameter supporting round up adjustment.
18837 (struct address_cost_data): Delete.
18838 (force_expr_to_var_cost): Don't bound cost with spill_cost.
18839 (split_address_cost, ptr_difference_cost): Delete.
18840 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
18841 (struct ainc_cost_data): New struct.
18842 (get_address_cost_ainc): New function.
18843 (get_address_cost, get_computation_cost): Reimplement.
18844 (determine_group_iv_cost_address): Record inv_expr for all uses of
18846 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
18847 (iv_ca_has_deps): Reimplemented to ...
18848 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
18850 (iv_ca_extend): Call iv_ca_more_deps.
18852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18854 * tree-ssa-address.c (struct mem_address): Move to header file.
18855 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
18856 * tree-ssa-address.h (struct mem_address): Move from C file.
18857 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
18859 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18861 * tree-affine.h (aff_combination_type): New interface.
18862 (aff_combination_zero_p): Remove static.
18863 (aff_combination_const_p): New interface.
18864 (aff_combination_singleton_var_p): New interfaces.
18866 2017-05-11 Richard Biener <rguenther@suse.de>
18868 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18869 Skip unreachable blocks and destinations.
18870 (eliminate): Move stmt removal and fixup ...
18871 (fini_eliminate): ... here. Skip inserted exprs.
18872 (pass_pre::execute): Move fini_pre after fini_eliminate.
18873 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
18874 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
18875 PRE to get rid of dead code that has invalid SSA form and
18876 split critical edges again.
18878 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18880 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
18882 2017-05-11 Richard Biener <rguenther@suse.de>
18884 * passes.c (execute_function_todo): Verify loops if they are
18885 said to be up-to-date.
18886 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
18887 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
18889 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
18892 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
18893 handle calling assemble_external ourself.
18896 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
18897 modes with zero size. Enhance comment.
18899 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18901 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
18902 built-ins for vec_xl and vec_xst with short and char pointer
18905 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18907 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
18908 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
18909 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
18910 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
18911 (_mm_maskz_min_round_ss): New intrinsics.
18912 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
18913 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
18914 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
18915 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
18916 (__builtin_ia32_minss_mask_round): New builtins.
18917 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18918 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18919 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
18921 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
18922 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
18924 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
18927 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
18929 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
18930 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
18931 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
18932 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
18933 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
18934 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
18935 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
18936 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18937 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
18938 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
18939 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
18940 (__builtin_ia32_mulss_mask_round): New builtins.
18941 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
18942 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
18943 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
18945 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
18946 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
18948 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
18951 2017-05-10 Julia Koval <julia.koval@intel.com>
18953 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
18954 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
18955 (_mm256_setr_m128i): New intrinsics.
18957 2017-05-10 Julia Koval <julia.koval@intel.com>
18959 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
18960 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
18961 (_mm_maskz_rcp14_ss): New intrinsics.
18962 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
18963 (__builtin_ia32_rcp14ss_mask): New builtins.
18964 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
18966 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
18968 PR tree-optimization/51513
18969 * tree-cfg.c (gimple_seq_unreachable_p): New function.
18970 (assert_unreachable_fallthru_edge_p): Use it.
18971 (group_case_labels_stmt): Likewise.
18972 * tree-cfg.h: Prototype it.
18973 * stmt.c: Include cfghooks.h and tree-cfg.h.
18974 (emit_case_dispatch_table) <gap_label>: New local variable.
18975 Use it to fill dispatch table gaps.
18976 Test for default_label before updating probabilities.
18977 (expand_case) <default_label>: Remove unneeded initialization.
18978 Test for unreachable default case statement and remove its edge.
18979 Set default_label accordingly.
18980 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
18982 2017-05-10 Carl Love <cel@us.ibm.com>
18984 * config/rs6000/rs6000-c: Add support for built-in functions
18985 vector signed char vec_neg (vector signed char)
18986 vector signed short int vec_neg (vector short int)
18987 vector signed int vec_neg (vector signed int)
18988 vector signed long long vec_neg (vector signed long long)
18989 vector float vec_neg (vector float)
18990 vector double vec_neg (vector double)
18991 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
18993 * config/rs6000/altivec.h: Add define for vec_neg
18994 * doc/extend.texi: Update the built-in documentation for the
18995 new built-in functions.
18997 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18999 PR tree-optimization/77644
19000 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19002 2017-05-10 Nathan Sidwell <nathan@acm.org>
19004 * dumpfile.h (TDI_lang_all): New.
19005 (TDF_KIND): New. Renumber others
19006 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19008 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19010 (get_dump_file_name): Adjust suffix generation.
19011 (dump_enable_all): Use TDF_KIND.
19012 * doc/invoke.texi (-fdump-lang-all): Document.
19014 * dumpfile.h: Tabify.
19016 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19019 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19020 Move member access before delete.
19022 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19024 * tree-inline.c (expand_call_inline): Split block at stmt
19027 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19030 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19031 are now unused after splitting mov{sf,sd}_hardfloat.
19032 (f32_lr2): Likewise.
19033 (f32_lm): Likewise.
19034 (f32_lm2): Likewise.
19035 (f32_li): Likewise.
19036 (f32_li2): Likewise.
19037 (f32_lv): Likewise.
19038 (f32_sr): Likewise.
19039 (f32_sr2): Likewise.
19040 (f32_sm): Likewise.
19041 (f32_sm2): Likewise.
19042 (f32_si): Likewise.
19043 (f32_si2): Likewise.
19044 (f32_sv): Likewise.
19045 (f32_dm): Likewise.
19046 (f32_vsx): Likewise.
19047 (f32_av): Likewise.
19048 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19049 For movsf, order stores so the VSX stores occur before the GPR
19050 store which encourages the register allocator to use a traditional
19051 FPR instead of a GPR. For movsd, order the stores so that the GPR
19052 store comes before the VSX stores to allow the power6 to work.
19053 This is due to the power6 not having a 32-bit integer store
19054 instruction from a FPR.
19055 (movsf_hardfloat): Likewise.
19056 (movsd_hardfloat): Likewise.
19058 2017-05-09 Martin Sebor <msebor@redhat.com>
19060 PR translation/80280
19061 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19064 PR translation/80280
19065 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19066 data member added in r247778.
19067 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19069 2017-05-09 Nathan Sidwell <nathan@acm.org>
19071 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19073 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19076 2017-05-09 Marek Polacek <polacek@redhat.com>
19078 * doc/invoke.texi: Fix typo.
19080 2017-05-09 Richard Biener <rguenther@suse.de>
19082 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19083 (vrp_val_is_min): Likewise.
19084 (set_value_range_to_value): Likewise.
19085 (set_value_range_to_nonnegative): Likewise.
19086 (gimple_assign_nonzero_p): Likewise.
19087 (gimple_stmt_nonzero_p): Likewise.
19088 (vrp_int_const_binop): Likewise. Remove unreachable case.
19089 (adjust_range_with_scev): Adjust comments.
19090 (compare_range_with_value): Likewise.
19091 (extract_range_from_phi_node): Likewise.
19092 (test_for_singularity): Likewise.
19094 2017-05-09 Richard Biener <rguenther@suse.de>
19096 * tree-vrp.c (get_single_symbol): Add assert that we don't
19097 get overflowed constants as invariant part.
19098 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19099 checks. Use wi::cmp instead of recursing for integer constants.
19100 (compare_values): Just ignore whether we assumed undefined
19101 overflow instead of failing the compare.
19102 (extract_range_for_var_from_comparison_expr): Add comment before the
19103 TREE_NO_WARNING sets.
19104 (test_for_singularity): Likewise.
19105 (extract_range_from_comparison): Do not disable optimization
19106 when we assumed undefined overflow.
19107 (extract_range_basic): Remove init of unused var.
19109 2017-05-09 Richard Biener <rguenther@suse.de>
19111 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19112 (extract_range_from_multiplicative_op_1): Adjust.
19113 (extract_range_from_binary_expr_1): Use int_const_binop.
19115 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19118 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19119 rs6000_store_data_bypass_p in seven define_bypass directives and
19120 in several comments.
19121 * config/rs6000/rs6000-protos.h: Add prototype for
19122 rs6000_store_data_bypass_p function.
19123 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19124 function implements slightly different (rs6000-specific) semantics
19125 than store_data_bypass_p, returning false rather than aborting
19126 with assertion error when arguments do not satisfy the
19127 requirements of store data bypass.
19128 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19129 rs6000_store_data_bypass_p.
19131 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19133 * config/xtensa/xtensa-protos.h
19134 (xtensa_initial_elimination_offset): New declaration.
19135 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19136 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19137 macro definition, add case for FRAME_POINTER_REGNUM when
19138 FRAME_GROWS_DOWNWARD.
19139 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19140 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19141 xtensa_initial_elimination_offset.
19143 2017-05-08 Nathan Sidwell <nathan@acm.org>
19145 * doc/invoke.texi: Alphabetize -fdump options.
19147 2017-05-08 Martin Sebor <msebor@redhat.com>
19149 PR translation/80280
19150 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19152 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19154 * target.def (compute_frame_layout): New optional target hook.
19155 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19156 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19157 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19159 * reload1.c (verify_initial_elim_offsets): Likewise.
19160 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19161 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19162 (arm_get_frame_offsets): Split up into this ...
19163 (arm_compute_frame_layout): ... and this function.
19165 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19167 * config/aarch64/constraints.md (Usa): New constraint.
19168 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19170 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19172 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19173 with_multilib_list after it has been checked.
19175 2017-05-08 Richard Biener <rguenther@suse.de>
19177 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19178 (bitmap_set_subtract_values): Likewise.
19180 2017-05-08 Richard Biener <rguenther@suse.de>
19182 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19183 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19185 (gimple_stmt_nonzero_warnv_p): Rename to ...
19186 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19188 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19189 (extract_range_basic): Adjust, do not disable propagation on
19190 strict overflow sensitive simplification.
19191 (vrp_visit_cond_stmt): Likewise.
19193 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19195 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19196 body size unconditionally.
19198 2017-05-07 Jeff Law <law@redhat.com>
19201 2017-05-06 Jeff Law <law@redhat.com>
19202 PR tree-optimization/78496
19203 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19206 PR tree-optimization/78496
19207 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19208 (simplify_stmt_using_ranges): Call it.
19209 (vrp_dom_walker::before_dom_children): Extract equivalences
19210 from an ASSERT_EXPR with an equality comparison against a
19213 2017-05-06 Jeff Law <law@redhat.com>
19215 PR tree-optimization/78496
19216 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19219 PR tree-optimization/78496
19220 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19221 (simplify_stmt_using_ranges): Call it.
19222 (vrp_dom_walker::before_dom_children): Extract equivalences
19223 from an ASSERT_EXPR with an equality comparison against a
19226 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19228 * lra-constraints.c (lra_copy_reg_equiv): New function.
19229 (split_reg): Use it to copy equivalence information from the
19230 original register to the spill register.
19232 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19234 PR rtl-optimization/75964
19235 * simplify-rtx.c (simplify_const_relational_operation): Remove
19236 invalid handling of comparisons of integer ABS.
19238 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19240 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19241 initialize to zero.
19242 (init_regs): Remove declaration.
19243 (function_arg_advance_32): Initialize error_p as boolean variable.
19245 2017-05-05 Nathan Sidwell <nathan@acm.org>
19247 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19248 lines. Use for (;;).
19250 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19252 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19253 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19254 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19255 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19256 VF=2 that require versioning.
19258 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19260 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19263 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19265 * diagnostic.h (diagnostic_override_option_index): Convert from
19266 macro to inline function.
19268 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19270 * diagnostic.c (last_module_changed_p): New function.
19271 (set_last_module): New function.
19272 (diagnostic_report_current_module): Convert macro usage to
19273 the above functions.
19274 * diagnostic.h (diagnostic_context::last_module): Strengthen
19275 from const line_map * to const line_map_ordinary *.
19276 (diagnostic_last_module_changed): Delete macro.
19277 (diagnostic_set_last_module): Delete macro.
19279 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19281 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19282 with diagnostic_report_diagnostic.
19283 (diagnostic_n_impl_richloc): Likewise.
19284 * diagnostic.h (report_diagnostic): Delete macro.
19285 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19286 with diagnostic_report_diagnostic.
19287 * substring-locations.c (format_warning_va): Likewise.
19289 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19291 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19292 save/restor of format_spec. Move option-printing code to...
19293 (print_option_information): ...this new function, and
19294 reimplement by simply printing to the pretty_printer,
19295 rather than appending to the format string.
19297 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19299 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19300 handling logic into...
19301 (update_effective_level_from_pragmas): ...this new function.
19303 2017-05-04 Andrew Waterman <andrew@sifive.com>
19305 * config/riscv/riscv.opt (mstrict-align): New option.
19306 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19307 (SLOW_UNALIGNED_ACCESS): Define.
19308 (riscv_slow_unaligned_access): Declare.
19309 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19311 (riscv_slow_unaligned_access): New variable.
19312 (rocket_tune_info): Set slow_unaligned_access to true.
19313 (optimize_size_tune_info): Set slow_unaligned_access to false.
19314 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19315 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19316 (riscv_option_override): Set riscv_slow_unaligned_access.
19317 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19319 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19321 * config/riscv/riscv.md: Unify indentation.
19323 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19328 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19330 (extendsi<mode>2): Add support for doing sign extension via
19331 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19332 don't have ISA 3.0 instructions.
19333 (extendsi<mode>2 splitter): Likewise.
19334 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19335 generate the normal insns since SImode can now go in vector
19336 registers. Disallow the special UNSPECs needed for previous
19337 machines to hide SImode being used. Add new insns
19338 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19339 (fix_trunc<mode>si2_stfiwx): Likewise.
19340 (fix_trunc<mode>si2_internal): Likewise.
19341 (fixuns_trunc<mode>si2): Likewise.
19342 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19343 (fctiw<u>z_<mode>_smallint): Likewise.
19344 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19345 of floating point to 32-bit integer from doing a direct move to
19346 the GPR registers to do a store.
19347 (fctiwz_<mode>): Break long line.
19349 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19351 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19352 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19353 (addr_list, addr_offset_valid_p): New.
19354 (split_address_groups): Check offset validity with above function.
19355 (gt-tree-ssa-loop-ivopts.h): Include header file.
19357 2017-05-05 Nathan Sidwell <nathan@acm.org>
19359 * config.gcc (arm*-*-*): Add missing 'fi'.
19361 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19363 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19364 included in -fopt-info does not matter.
19365 * doc/optinfo.texi (-fopt-info): Fix description of default
19366 behavour. Explicitly say order of options included in -fopt-info
19369 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19371 * config.gcc: Allow combinations of aprofile and rmprofile values for
19372 --with-multilib-list.
19373 * config/arm/t-multilib: New file.
19374 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19375 variables. Remove setting of ISA and floating-point ABI in
19376 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19377 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19378 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19379 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19381 * config/arm/t-rmprofile: Likewise except for the matches changes.
19382 * doc/install.texi (--with-multilib-list): Document the combination of
19383 aprofile and rmprofile values and warn about pitfalls in doing that.
19385 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19387 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19388 (movdi_aarch64): Likewise.
19390 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19392 PR tree-optimization/80632
19393 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19395 (build_arrays): Initialize it for virtual phis.
19396 (fix_phi_nodes): Use it for virtual phis.
19398 PR tree-optimization/80558
19399 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19400 [x, y] op z into [x op, y op z] for op & or | if conditions
19403 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19404 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19407 * config/arm/arm.md (use_literal_pool): Remove.
19408 (64-bit immediate split): No longer takes cost into consideration
19409 if arm_disable_literal_pool is enabled.
19410 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19411 used when arm_disable_literal_pool is enabled.
19412 (arm_max_const_double_inline_cost): Remove use of
19413 arm_disable_literal_pool.
19414 (push_minipool_fix): Add assert.
19415 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19416 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19417 (no_literal_pool_sf_immediate): New.
19419 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19421 PR tree-optimization/80613
19422 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19423 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19425 2017-05-05 Richard Biener <rguenther@suse.de>
19427 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19429 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19431 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19432 of this flag from insn conditions due to removal from r247495.
19434 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19436 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19438 (arm_early_store_addr_dep_ptr): Likewise.
19439 * config/arm/aarch-common-protos.h
19440 (arm_early_load_addr_dep_ptr): Add prototype.
19441 (arm_early_store_addr_dep_ptr): Likewise.
19442 * config/arm/cortex-a53.md: Add new bypasses.
19444 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19446 * tree.c (next_type_uid): Change type to unsigned.
19447 (type_hash_canon): Decrement back next_type_uid if
19448 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19449 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19452 2017-05-04 Martin Sebor <msebor@redhat.com>
19454 * builtins.c: Fix a trivial typo in a comment.
19456 PR middle-end/79234
19457 * builtins.c (check_sizes): Adjust to handle reading past the end.
19458 Avoid printing excessive upper bound of ranges. Use %E to print
19459 tree nodes instead of converting them to %wu.
19460 (expand_builtin_memchr): New function.
19461 (compute_dest_size): Rename...
19462 (compute_objsize): ...to this.
19463 (expand_builtin_memcpy): Adjust.
19464 (expand_builtin_mempcpy): Adjust.
19465 (expand_builtin_strcat): Adjust.
19466 (expand_builtin_strcpy): Adjust.
19467 (check_strncat_sizes): Adjust.
19468 (expand_builtin_strncat): Adjust.
19469 (expand_builtin_strncpy): Adjust and simplify.
19470 (expand_builtin_memset): Adjust.
19471 (expand_builtin_bzero): Adjust.
19472 (expand_builtin_memcmp): Adjust.
19473 (expand_builtin): Handle memcmp.
19474 (maybe_emit_chk_warning): Check strncat just once.
19476 2017-05-04 Martin Sebor <msebor@redhat.com>
19478 PR preprocessor/79214
19479 PR middle-end/79222
19480 PR middle-end/79223
19481 * builtins.c (check_sizes): Add inlining context and issue
19482 warnings even when -Wno-system-headers is set.
19483 (check_strncat_sizes): Same.
19484 (expand_builtin_strncat): Same.
19485 (expand_builtin_memmove): New function.
19486 (expand_builtin_stpncpy): Same.
19487 (expand_builtin): Handle memmove and stpncpy.
19489 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19491 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19492 which is not used any more.
19494 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19496 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19498 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19500 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19501 (cortexa53_tunings): Likewise.
19502 (cortexa57_tunings): Likewise.
19503 (cortexa72_tunings): Likewise.
19504 (cortexa73_tunings): Likewise.
19506 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19508 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19509 Set loop alignment to 8.
19511 2017-05-04 Martin Sebor <msebor@redhat.com>
19513 PR translation/80280
19514 * builtins.c (expand_builtin_object_size): Add missing quoting to
19515 %D and like directives.
19516 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19517 (hsa_type_for_tree_type): Same.
19518 (verify_function_arguments): Same.
19519 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19520 * varasm.c (get_section): Same.
19523 2017-05-04 Martin Sebor <msebor@redhat.com>
19525 PR translation/80280
19526 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19528 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19530 * config/aarch64/aarch64.c (generic_addrcost_table):
19531 Change HI/TI mode setting.
19533 2017-05-04 Martin Jambor <mjambor@suse.cz>
19535 PR tree-optimization/80622
19536 * tree-sra.c (comes_initialized_p): New function.
19537 (build_accesses_from_assign): Only set write lazily when
19538 comes_initialized_p is false.
19539 (analyze_access_subtree): Use comes_initialized_p.
19540 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19541 instead of testing for PARM_DECL.
19543 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19545 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19546 constraint on operand 0 to allow more general addressing modes.
19547 Adjust output template.
19548 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19550 * config/aarch64/aarch64-protos.h
19551 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19552 * config/aarch64/constraints.md (Dp): New address constraint.
19553 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19556 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19558 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19559 update use of estimate_ipcp_clone_size_and_time.
19560 (estimate_local_effects): Update use of
19561 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19562 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19563 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19564 Return nonspecialized time.
19566 2017-05-04 Richard Biener <rguenther@suse.de>
19568 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19569 for the last VUSE which def dominates the PHI. Directly call
19571 (get_continuation_for_phi_1): Remove.
19573 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19575 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19576 to explain the use of truncating division. Cap the number of
19577 iterations to the maximum given by nb_iterations_upper_bound,
19580 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19582 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19583 * configure: Regenerate.
19584 * config.in: Regenerate.
19585 * config/i386/driver-mingw32.c: new file.
19586 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19587 * config.host: Link driver-mingw32.o on MinGW host.
19588 * doc/install.texi: Document new --enable-mingw-wildcard configure
19591 2017-05-04 Marek Polacek <polacek@redhat.com>
19593 PR tree-optimization/80612
19594 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19596 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19597 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19599 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19600 (movt splitter): Likewise.
19601 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19602 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19603 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19604 block for Thumb-1 with MOVT.
19605 (thumb2_legitimate_address_p): Move code block ...
19606 (can_avoid_literal_pool_for_label_p): ... into this new function.
19607 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19609 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19610 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19611 "M-profile targets with the MOVT instruction".
19613 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19615 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19616 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19617 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19619 2017-05-04 Martin Liska <mliska@suse.cz>
19621 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19622 variable cond_code.
19624 2017-05-04 Richard Biener <rguenther@suse.de>
19626 * tree.c (array_at_struct_end_p): Handle arrays at struct
19627 end with flexarrays more conservatively. Refactor and treat
19628 arrays of arrays or aggregates more strict. Fix
19629 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19630 * tree.c (array_at_struct_end_p): Adjust prototype.
19631 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19632 * gimple-fold.c (get_range_strlen): Likewise.
19633 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19635 2017-05-04 Richard Biener <rguenther@suse.de>
19637 PR tree-optimization/31130
19638 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19640 (supports_overflow_infinity): Likewise.
19641 (is_negative_overflow_infinity): Likewise.
19642 (is_positive_overflow_infinity): Likewise.
19643 (is_overflow_infinity): Likewise.
19644 (stmt_overflow_infinity): Likewise.
19645 (overflow_infinity_range_p): Likewise.
19646 (usable_range_p): Remove as always returning true.
19647 (make_overflow_infinity): Remove.
19648 (negative_overflow_infinity): Likewise.
19649 (positive_overflow_infinity): Likewise.
19650 (avoid_overflow_infinity): Likewise.
19651 (set_value_range): Adjust accordingly.
19652 (set_value_range_to_nonnegative): Likewise, remove now unused
19653 overflow_infinity arg.
19654 (vrp_operand_equal_p): Adjust.
19655 (update_value_range): Likewise.
19656 (range_int_cst_singleton_p): Likewise.
19657 (operand_less_p): Likewise.
19658 (compare_values_warnv): Likewise.
19659 (extract_range_for_var_from_comparison_expr): Likewise.
19660 (vrp_int_const_binop): Likewise.
19661 (zero_nonzero_bits_from_vr): Likewise.
19662 (extract_range_from_multiplicative_op_1): Likewise.
19663 (extract_range_from_binary_expr_1): Likewise.
19664 (extract_range_from_unary_expr): Likewise.
19665 (extract_range_from_comparison): Likewise.
19666 (extract_range_basic): Likewise.
19667 (adjust_range_with_scev): Likewise.
19668 (compare_ranges): Likewise.
19669 (compare_range_with_value): Likewise.
19670 (dump_value_range): Likewise.
19671 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19673 (simplify_cond_using_ranges): Adjust.
19675 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19677 * brig-builtins.def: Added a builtin for class_f64.
19678 * builtin-types.def: Added a builtin type needed by class_f64.
19680 2017-05-03 Jason Merrill <jason@redhat.com>
19682 * timevar.def: Add TV_CONSTEXPR.
19684 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19686 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19688 2017-05-03 Martin Jambor <mjambor@suse.cz>
19690 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19691 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19692 * ipa-cp.c (ipcp_propagate_stage): Do not call
19693 ipa_update_after_lto_read.
19694 * ipa-inline.c (ipa_inline): Likewise.
19696 2017-05-03 Martin Jambor <mjambor@suse.cz>
19698 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19699 tag. Added a default constructor and a destructor.
19700 (ipa_edge_args_sum_t): New class;
19701 (ipa_edge_args_sum): Declare.
19702 (ipa_edge_args_vector): Remove declaration.
19703 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19704 (ipa_free_edge_args_substructures): Remove declaration.
19705 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19706 (ipa_edge_args_info_available_for_edge_p): Likewise.
19707 * ipa-prop.c (ipa_edge_args_vector): Removed.
19708 (edge_removal_hook_holder): Likewise.
19709 (edge_duplication_hook_holder): Likewise.
19710 (ipa_edge_args_sum): New variable.
19711 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19712 ipa_edge_args_vector.
19713 (ipa_free_edge_args_substructures): Likewise.
19714 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19715 ipa_edge_args_vector.
19716 (ipa_edge_removal_hook): Turned into method
19717 ipa_edge_args_sum_t::remove.
19718 (ipa_edge_duplication_hook): Turned into method
19719 ipa_edge_args_sum_t::duplicate.
19720 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19721 registering edge hooks.
19722 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19723 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19724 ipa_edge_args_sum instead of ipa_edge_args_vector.
19725 * ipa-profile.c (ipa_profile): Likewise.
19727 2017-05-03 Martin Jambor <mjambor@suse.cz>
19729 * symbol-summary.h (function_summary): New method exists.
19730 (function_summary::symtab_removal): Deallocate through release.
19731 (call_summary): New class.
19732 (gt_ggc_mx): New overload.
19733 (gt_pch_nx): Likewise.
19734 (gt_pch_nx): Likewise.
19736 2017-05-03 Jeff Law <law@redhat.com>
19738 PR tree-optimization/78496
19739 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19740 from simplify_cond_using_ranges. Split off code to walk
19741 backwards through casts into ...
19742 (simplify_cond_using_ranges_2): New function.
19743 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19744 (execute_vrp): After identifying jump threads, call
19745 simplify_cond_using_ranges_2.
19747 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19750 * ipa-inline.h (inline_summary): Add ctor.
19751 (create_ggc): Do not use ggc_cleared_alloc.
19753 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19754 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19756 * gcc.c (handle_braces): Support escaping in switch matching
19758 * doc/invoke.texi (Spec Files): Document it.
19759 Remove superfluous @code markup in items.
19761 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19763 * diagnostic-show-locus.c (struct column_range): New struct.
19764 (get_affected_columns): New function.
19765 (get_printed_columns): New function.
19766 (struct correction): New struct.
19767 (correction::ensure_capacity): New function.
19768 (correction::ensure_terminated): New function.
19769 (struct line_corrections): New struct.
19770 (line_corrections::~line_corrections): New dtor.
19771 (line_corrections::add_hint): New function.
19772 (layout::print_trailing_fixits): Reimplement in terms of the new
19774 (selftest::test_overlapped_fixit_printing): New function.
19775 (selftest::diagnostic_show_locus_c_tests): Call it.
19777 2017-05-03 Nathan Sidwell <nathan@acm.org>
19779 Canonicalize canonical type hashing
19780 * tree.h (type_hash_canon_hash): Declare.
19781 * tree.c (type_hash_list, attribute_hash_list): Move into
19782 type_hash_canon_hash.
19783 (build_type_attribute_qual_variant): Break out hash code calc into
19784 type_hash_canon_hash.
19785 (type_hash_canon_hash): New. Generic type hash computation.
19786 (build_range_type_1, build_array_type_1, build_function_type,
19787 build_method_type_directly, build_offset_type, build_complex_type,
19788 make_vector_type): Call it.
19790 2017-05-03 Richard Biener <rguenther@suse.de>
19792 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19793 When all DRs have unknown misaligned do not always peel
19794 when there is a store but apply the same costing model as if
19795 there were only loads.
19797 2017-05-03 Richard Biener <rguenther@suse.de>
19800 PR tree-optimization/80492
19801 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19802 compare_base_decls returning dont-know properly.
19804 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19806 * config/arm/iterators.md (CCSI): New mode iterator.
19807 (arch): New mode attribute.
19808 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19809 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19810 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19811 code iterator for success result mode.
19812 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19813 the corresponding new insn generators.
19815 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19818 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19819 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19821 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19823 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19824 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19825 (DDR_A): Wrap DDR argument in brackets.
19826 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19827 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19828 (DDR_REVERSED_P): Likewise.
19830 2017-05-03 Jakub Jelinek <jakub@redhat.com>
19832 PR tree-optimization/79472
19833 * tree-switch-conversion.c (struct switch_conv_info): Add
19834 contiguous_range and default_case_nonstandard fields.
19835 (collect_switch_conv_info): Compute contiguous_range and
19836 default_case_nonstandard fields, don't clear final_bb if
19837 contiguous_range and only the default case doesn't have the required
19839 (check_all_empty_except_final): Set default_case_nonstandard instead
19840 of failing if contiguous_range and the default case doesn't have empty
19842 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
19843 and only the default case doesn't have the required constants. Skip
19845 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
19846 if default_case_nonstandard.
19847 (build_constructors): Build constant 1 just once. Assert that default
19848 values aren't inserted in between cases if contiguous_range. Skip
19850 (build_arrays): Skip virtual phis.
19851 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
19852 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
19853 Handle virtual phis.
19854 (gen_inbound_check): Handle default_case_nonstandard case.
19855 (process_switch): Adjust check_final_bb caller. Call
19856 gather_default_values with the first non-default case instead of
19857 default case if default_case_nonstandard.
19859 2017-05-02 Nathan Sidwell <nathan@acm.org>
19861 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
19862 check. Fix formatting.
19864 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
19866 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
19867 errors when comparing specialized and unspecialized times.
19869 2017-05-02 David Malcolm <dmalcolm@redhat.com>
19871 * diagnostic-show-locus.c
19872 (layout::should_print_annotation_line_p): Make private.
19873 (layout::print_annotation_line): Make private.
19874 (layout::annotation_line_showed_range_p): Make private.
19875 (layout::show_ruler): Make private.
19876 (layout::print_source_line): Make private. Pass in line and
19877 line_width, rather than calling location_get_source_line. Drop
19879 (layout::print_leading_fixits): New method.
19880 (layout::print_any_fixits): Rename to...
19881 (layout::print_trailing_fixits): ...this, and make private.
19882 Don't print newline fixits.
19883 (diagnostic_show_locus): Move logic for printing one row into...
19884 (layout::print_line): ...this new function. Move the
19885 location_get_source_line call and error-handling from
19886 print_source_line to here. Call print_leading_fixits, and rename
19887 print_any_fixits to print_trailing_fixits.
19888 (selftest::test_fixit_insert_containing_newline): Update now that
19889 newlines are partially supported.
19890 (selftest::test_fixit_insert_containing_newline_2): New test.
19891 (selftest::test_fixit_replace_containing_newline): Update comments.
19892 (selftest::diagnostic_show_locus_c_tests): Call the new test.
19893 * edit-context.c (class added_line): New class.
19894 (class edited_line): Describe newline handling in comment.
19895 (edited_line::actually_edited_p): New method.
19896 (edited_line::print_content): Delete redundant decl.
19897 (edited_line::m_predecessors): New field.
19898 (edited_file::print_content): Call edited_line::print_content.
19899 (edited_file::print_diff): Update to support newlines.
19900 (edited_file::print_diff_hunk): Likewise.
19901 (edited_file::print_run_of_changed_lines): New function.
19902 (edited_file::print_diff_line): Convert to...
19903 (print_diff_line): ...this.
19904 (edited_file::get_effective_line_count): New function.
19905 (edited_line::edited_line): Initialize new field m_predecessors.
19906 (edited_line::~edited_line): Clean up m_predecessors.
19907 (edited_line::apply_fixit): Handle newlines.
19908 (edited_line::get_effective_line_count): New function.
19909 (edited_line::print_content): New function.
19910 (edited_line::print_diff_lines): New function.
19911 (selftest::test_applying_fixits_insert_containing_newline): New
19913 (selftest::test_applying_fixits_replace_containing_newline): New
19915 (selftest::insert_line): New function.
19916 (selftest::test_applying_fixits_multiple_lines): Add example of
19918 (selftest::edit_context_c_tests): Call the new tests.
19920 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19922 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
19923 parameter cand. Update dump information.
19924 (get_computation_cost): Update uses.
19926 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19928 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
19929 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
19930 (get_computation_at, rewrite_use_address): Update use of
19931 get_computation_aff.
19933 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19935 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
19936 (get_computation): Delete.
19937 (get_computation_cost): Implement like get_computation_cost_at.
19938 Use get_computation_at.
19939 (get_computation_cost_at): Delete.
19940 (rewrite_use_nonlinear_expr): Use get_computation_at.
19941 (rewrite_use_compare, remove_unused_ivs): Ditto.
19943 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19945 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
19947 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19949 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
19950 (ivopts_global_cost_for_size): Rename parameter and update uses.
19951 (iv_ca_recount_cost): Update uses.
19952 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
19953 candidates seperately in n_invs and n_cands.
19954 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
19956 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19958 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
19959 (find_inv_vars_cb): New.
19960 (find_depends): Renamed to ...
19961 (find_inv_vars): ... this.
19962 (add_candidate_1, force_var_cost): Call find_inv_vars.
19963 (split_address_cost, determine_group_iv_cost_cond): Ditto.
19965 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19967 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
19968 inv_vars. Add inv_exprs.
19969 (struct iv_cand): Rename depends_on to inv_vars.
19970 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
19971 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
19972 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
19973 (dump_cand): Dump inv_vars.
19974 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
19975 (record_invariant, find_depends, add_candidate_1): Ditto.
19976 (set_group_iv_cost, force_var_cost): Ditto.
19977 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
19978 (get_computation_cost_at, get_computation_cost): Ditto.
19979 (determine_group_iv_cost_generic): Ditto.
19980 (determine_group_iv_cost_address): Ditto.
19981 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
19982 (determine_group_iv_costs): Ditto.
19983 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
19984 (iv_ca_set_remove_invariants): Renamed to ...
19985 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
19986 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
19987 (iv_ca_set_add_invariants): Renamed to ...
19988 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
19989 (iv_ca_set_cp): Use iv_ca_set_add_invs.
19990 (iv_ca_has_deps): Support inv_vars and inv_exprs.
19991 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
19992 (create_new_ivs): Remove useless dump.
19994 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19996 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
19998 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
19999 (iv_ca_set_no_cp, create_new_iv): Ditto.
20001 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20003 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20005 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20007 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20008 function tree_check2.
20010 2017-05-02 Martin Liska <mliska@suse.cz>
20012 * doc/gcov.texi: Add missing preposition.
20013 * gcov.c (function_info::function_info): Properly fill up
20014 all member variables.
20016 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20018 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20020 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20022 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20024 2017-05-02 Martin Liska <mliska@suse.cz>
20027 * lto-streamer-in.c (lto_read_tree_1): Remove
20028 LTO_STREAMER_DEBUG.
20029 * lto-streamer.c (struct tree_hash_entry): Likewise.
20030 (struct tree_entry_hasher): Likewise.
20031 (tree_entry_hasher::hash): Likewise.
20032 (tree_entry_hasher::equal): Likewise.
20033 (lto_streamer_init): Likewise.
20034 (lto_orig_address_map): Likewise.
20035 (lto_orig_address_get): Likewise.
20036 (lto_orig_address_remove): Likewise.
20037 * lto-streamer.h: Likewise.
20038 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20039 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20041 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20043 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20044 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20045 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20046 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20047 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20048 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20049 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20050 (mm_maskz_sub_ss): New intrinsics.
20051 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20052 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20053 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20054 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20055 (__builtin_ia32_subss_mask_round): New builtins.
20056 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20057 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20058 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20060 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20061 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20063 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20066 2017-05-02 Martin Jambor <mjambor@suse.cz>
20068 PR tree-optimization/78687
20069 * tree-sra.c (access): New field parent.
20070 (process_subtree_disqualification): New function.
20071 (disqualify_candidate): Call it.
20072 (build_accesses_from_assign): Reset write flag if creating an
20074 (build_access_subtree): Fill in parent field and also prpagate
20075 down grp_write flag.
20076 (create_artificial_child_access): New parameter set_grp_write, set
20077 grp_write to its value.
20078 (propagate_subaccesses_across_link): Also propagate grp_write flag
20080 (propagate_all_subaccesses): Push the closest parent back to work
20081 queue if add_access_to_work_queue returned true.
20083 2017-05-02 Richard Biener <rguenther@suse.de>
20085 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20086 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20087 -fstrict-overflow documentation.
20088 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20089 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20090 flag_strict_overflow.
20091 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20092 * lto-opts.c (lto_write_options): Do not stream it.
20093 * lto-wrapper.c (merge_and_complain): Do not handle it.
20094 * opts.c (default_options_table): Do not set -fstrict-overflow.
20095 (finish_options): Likewise do not clear it when sanitizing.
20096 * simplify-rtx.c (simplify_const_relational_operation): Do not
20097 test flag_strict_overflow.
20099 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20101 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20102 using enabled attribute.
20103 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20104 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20105 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20106 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20107 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20108 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20109 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20110 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20111 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20112 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20114 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20116 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20118 2017-05-02 Richard Biener <rguenther@suse.de>
20120 PR tree-optimization/80591
20122 2017-04-10 Richard Biener <rguenther@suse.de>
20124 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20127 2017-05-02 Richard Biener <rguenther@suse.de>
20129 PR tree-optimization/80549
20130 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20131 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20132 headers if they do not have a preheader.
20134 2017-05-02 Martin Liska <mliska@suse.cz>
20137 * common.opt: Fix typo.
20138 * doc/invoke.texi: Likewise.
20140 2017-05-01 Jan Beulich <jbeulich@suse.com>
20142 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20143 swapping, add (x,x,m,x,n) alternative.
20145 2017-05-01 Nathan Sidwell <nathan@acm.org>
20147 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20148 unnecessary unadjusted_alignment check.
20150 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20153 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20155 * gimplify.c (gimplify_cilk_detach): New function.
20156 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20157 * tree-core.h: Document EXPR_CILK_SPAWN.
20158 * tree.h (EXPR_CILK_SPAWN): Define.
20160 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20162 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20163 to use new fixit_hint representation, using the "replace" logic.
20164 (get_line_span_for_fixit_hint): Likewise.
20165 (layout::print_any_fixits): Likewise.
20166 (selftest::test_one_liner_many_fixits): Rename to...
20167 (selftest::test_one_liner_many_fixits_1): ...this, and update
20168 comment and expected output to reflect that the multiple fix-it
20169 hints are now consolidated into one insertion.
20170 (selftest::test_one_liner_many_fixits_2): New test.
20171 (selftest::test_diagnostic_show_locus_one_liner): Update for
20173 (selftest::test_fixit_consolidation): Update for fix-it API
20175 * diagnostic.c (print_parseable_fixits): Likewise.
20176 * edit-context.c (edited_line::m_line_events): Convert from
20177 auto_vec <line_event *> to auto_vec <line_event>.
20178 (class line_event): Convert from abstract base class to a concrete
20179 class, taking over the role of replace_event.
20180 (class insert_event): Delete.
20181 (class replace_event): Rename to class line_event. Convert to
20183 (edit_context::add_fixits): Reimplement.
20184 (edit_context::apply_insert): Delete.
20185 (edit_context::apply_replace): Rename to...
20186 (edit_context::apply_fixit): ...this. Convert to half-open range.
20187 (edited_file::apply_insert): Delete.
20188 (edited_file::apply_replace): Rename to...
20189 (edited_file::apply_fixit): ...this.
20190 (edited_line::~edited_line): Drop deletion of events.
20191 (edited_line::apply_insert): Delete.
20192 (edited_line::apply_replace): Rename to...
20193 (edited_line::apply_fixit): ...this. Convert to half-open range.
20194 Update for change to type of m_line_events.
20195 * edit-context.h (edit_context::apply_insert): Delete.
20196 (edit_context::apply_replace): Rename to...
20197 (edit_context::apply_fixit): ...this.
20199 2017-05-01 Martin Sebor <msebor@redhat.com>
20201 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20204 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20207 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20208 __get_cpuid_max returns 0.
20209 (__get_cpuid_count): Ditto.
20211 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20213 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20214 replacement expression is another instance of one of its arguments.
20216 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20219 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20220 check for stack push/pop autoinc.
20221 * config/i386/i386.c (ix86_agi_dependent): Return false
20222 if the only reason why modified_in_p returned true is that
20223 addr is SP based and set_insn is a push or pop.
20225 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20227 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20230 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20233 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20234 (account_size_time): Use two predicates - exec_pred and
20236 (evaluate_conditions_for_known_args): Compute both clause and
20238 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20239 (inline_summary_t::duplicate): Update.
20240 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20242 (compute_inline_parameters): Likewise.
20243 (estimate_edge_size_and_time): Update caluclation of time.
20244 (estimate_node_size_and_time): Compute both time and nonspecialized
20246 (estimate_ipcp_clone_size_and_time): Update.
20247 (inline_merge_summary): Update.
20248 (do_estimate_edge_time): Update.
20249 (do_estimate_edge_size): Update.
20250 (do_estimate_edge_hints): Update.
20251 (inline_read_section, inline_write_summary): Stream both new predicates.
20252 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20254 (compute_inlined_call_time): Cleanup.
20255 (big_speedup_p): Update.
20256 (edge_badness): Update.
20257 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20258 (size_time_entry): Replace predicate by exec_predicate and
20259 nonconst_predicate.
20260 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20261 (estimate_edge_time): Return also nonspec_time.
20262 (reset_edge_growth_cache): Update.
20264 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20266 PR rtl-optimization/80491
20267 * ifcvt.c (noce_process_if_block): When looking for x setter
20268 with missing else_bb, don't check only the insn right before
20269 cond_earliest, but look for the last insn that x is modified in
20270 within the same bb.
20272 PR rtl-optimization/80491
20273 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20275 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20277 PR tree-optimization/80487
20278 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20280 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20282 PR tree-optimization/79697
20283 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20284 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20285 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20287 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20288 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20290 2017-04-28 Martin Sebor <msebor@redhat.com>
20292 PR tree-optimization/80523
20293 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20294 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20296 (maybe_warn, format_directive, parse_directive): Use new functions.
20297 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20299 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20301 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20303 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20305 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20306 target_header_dir): Set correctly.
20307 * configure: Regenerated.
20308 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20309 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20310 instead of SYSTEM_HEADER_DIR.
20312 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20314 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20315 (estimate_local_effects): Likewise.
20316 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20317 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20318 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20319 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20320 do_estimate_edge_time, estimate_edge_time): Likewise.
20321 * ipa-inline-analysis.c (estimate_node_size_and_time,
20322 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20323 (estimate_time_after_inlining): Remove.
20325 2017-04-28 Martin Liska <mliska@suse.cz>
20327 * doc/gcov.texi: Enhance documentation of gcov.
20329 2017-04-28 Martin Liska <mliska@suse.cz>
20331 * doc/gcov.texi: Sort options in alphabetic order.
20332 * doc/gcov-dump.texi: Likewise.
20333 * doc/gcov-tool.texi: Likewise.
20334 * gcov.c (print_usage): Likewise.
20335 * gcov-dump.c (print_usage): Likewise.
20336 * gcov-tool.c (print_merge_usage_message): Likewise.
20337 (print_rewrite_usage_message): Likewise.
20338 (print_overlap_usage_message): Likewise.
20340 2017-04-28 Martin Liska <mliska@suse.cz>
20342 PR gcov-profile/53915
20343 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20345 2017-04-28 Martin Liska <mliska@suse.cz>
20347 PR gcov-profile/79891
20348 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20349 is marked by compiler as living on a line.
20350 (get_cycles_count): Remove usage of the union.
20351 (output_intermediate_file): Likewise.
20352 (find_source): Fix GNU coding style.
20353 (accumulate_line_counts): Remove old non-all block mode.
20354 (output_lines): Remove usage of the union.
20355 * profile.c (output_location): Include all BBs, even if
20356 belonging to a same line (and file) as a previous BB.
20358 2017-04-28 Martin Liska <mliska@suse.cz>
20360 * gcov.c (process_args): Handle new argument 'w'.
20361 (read_graph_file): Assign ID to BBs.
20362 (output_branch_count): Display BB # if verbose flag is set.
20363 (output_lines): Likewise for arcs.
20364 (print_usage): Add '--verbose' option help.
20365 * doc/gcov.texi: Document --verbose (-w) option.
20367 2017-04-28 Martin Liska <mliska@suse.cz>
20369 * gcov.c (struct block_location_info): New struct.
20370 (process_file): Fill up the new structure.
20371 (read_graph_file): Replace usage of encoding by the newly added
20373 (add_line_counts): Likewise.
20374 (accumulate_line_counts): Remove usage of the union.
20375 (function_info::function_info): New function.
20376 (function_info::~function_info): Likewise.
20377 (process_file): Call delete instead of release_function.
20378 (release_function): Release the function.
20379 (release_structures): Call delete instead of release_function.
20380 (solve_flow_graph): Replace usage of num_blocks.
20381 (find_exception_blocks): Likewise.
20382 (output_lines): Fix GNU coding style.
20384 2017-04-28 Martin Liska <mliska@suse.cz>
20387 * coverage.c (coverage_remove_note_file): New function.
20388 * coverage.h: Declare the function.
20389 * toplev.c (finalize): Clean if an error has been seen.
20391 2017-04-28 Martin Liska <mliska@suse.cz>
20393 PR gcov-profile/80031
20394 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20395 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20396 * gcov.c (read_graph_file): Read just number of blocks.
20397 * profile.c (branch_prob): Do not stream 0 flags per a basic
20400 2017-04-28 Martin Liska <mliska@suse.cz>
20402 * gcov-dump.c (tag_*): Add new argument to declarations.
20403 (dump_gcov_file): Likewise.
20404 (tag_blocks): Add and use new argument depth.
20405 (tag_arcs): Likewise.
20406 (tag_lines): Likewise.
20407 (tag_counters): Likewise.
20408 (tag_summary): Likewise.
20409 (dump_working_sets): Use depth to do a proper indentation.
20411 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20414 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20415 * symtab.c (symtab_node::debug_symtab): Move definition here.
20417 2017-04-28 Richard Biener <rguenther@suse.de>
20419 * lto-streamer.h (LTO_major_version): Bump to 7.
20421 2017-04-28 Richard Biener <rguenther@suse.de>
20423 * tree-vrp.c (assert_info): New struct.
20424 (add_assert_info): New helper.
20425 (register_edge_assert_for_2): Refactor to add asserts to a vector
20427 (register_edge_assert_for_1): Likewise.
20428 (register_edge_assert_for): Likewise.
20429 (finish_register_edge_assert_for): New helper actually registering
20430 asserts where live on edge.
20431 (find_conditional_asserts): Adjust.
20432 (find_switch_asserts): Likewise.
20433 (evrp_dom_walker::try_find_new_range): Generalize.
20434 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20436 2017-04-27 Marek Polacek <polacek@redhat.com>
20439 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20440 arg10 and arg11 to itype.
20442 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20444 * doc/extend.texi (Object Size Checking): Improve grammar.
20446 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20449 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20450 that the logic for permitting reciprocal estimates matches that
20453 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20456 * tree.c (type_cache_hasher::equal): Only compare
20457 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20458 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20459 non-aggregate element types.
20460 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20461 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20463 2017-04-27 Richard Biener <rguenther@suse.de>
20465 PR middle-end/80533
20466 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20467 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20468 keeping a reference to a trailing array.
20470 2017-04-27 Richard Biener <rguenther@suse.de>
20472 PR middle-end/80539
20473 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20474 being in loop-closed SSA form conservatively.
20475 (chrec_fold_multiply_poly_poly): Likewise.
20477 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20479 PR middle-end/79665
20480 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20481 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20483 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20486 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20487 (aarch64_function_arg_alignment): Return unsigned int again, but still
20488 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20489 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20490 Don't emit -Wpsabi note.
20491 (aarch64_function_arg_boundary): Likewise.
20492 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20495 2017-04-26 Nathan Sidwell <nathan@acm.org>
20497 * tree.h (crc32_unsigned_n): Declare.
20498 (crc32_unsigned, crc32_unsigned): Make inline.
20499 * tree.c (crc32_unsigned_bits): Replace with ...
20500 (crc32_unsigned_n): ... this.
20501 (crc32_unsigned, crc32_byte): Remove.
20502 (crc32_string): Remove unnecessary braces.
20504 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20506 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20507 * ipa-inline-analysis.c (MAX_TIME): Remove.
20508 (account_size_time): Use sreal for time.
20509 (dump_inline_summary): Update.
20510 (estimate_function_body_sizes): Update.
20511 (estimate_edge_size_and_time): Update.
20512 (estimate_calls_size_and_time): Update.
20513 (estimate_node_size_and_time): Update.
20514 (inline_merge_summary): Update.
20515 (inline_update_overall_summary): Update.
20516 (estimate_time_after_inlining): Update.
20517 (inline_read_section): Update.
20518 (inline_write_summary): Update.
20519 * ipa-inline.c (compute_uninlined_call_time): Update.
20520 (compute_inlined_call_time): Update.
20521 (recursive_inlining): Update.
20522 (inline_small_functions): Update.
20523 (dump_overall_stats): Update.
20524 * ipa-inline.h: Include sreal.h.
20525 (size_time_entry): Turn time to sreal.
20526 (inline_summary): Turn self_time nad time to sreal.
20528 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20530 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20532 (sreal::stream_out, sreal::stream_in): New.
20533 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20535 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20537 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20540 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20543 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20545 (dimode_scalar_chain::compute_convert_gain): Ditto.
20546 (dimode_scalar_chain::make_vector_copies): Ditto.
20547 (dimode_scalar_chain::convert_reg): Ditto.
20548 (dimode_scalar_chain::convert_insn): Ditto.
20549 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20550 (VI248_AVX512BW_1): New mode iterator.
20551 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20552 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20555 2017-04-25 Martin Sebor <msebor@redhat.com>
20557 PR tree-optimization/80497
20558 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20559 constants are representable in HOST_WIDE_INT.
20560 (parse_directive): Ditto.
20562 2017-04-25 Martin Sebor <msebor@redhat.com>
20565 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20566 (new_zero_array): Adjust signature.
20567 (dom_info::dom_init): Used unsigned rather that size_t.
20568 (dom_info::dom_info): Same.
20570 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20571 Jakub Jelinek <jakub@redhat.com>
20574 * config/arm/arm.c: Include gimple.h.
20575 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20576 returns negative, increment ncrn only if it returned positive.
20577 (arm_needs_doubleword_align): Return int instead of bool,
20578 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20579 members, but if there is any such non-FIELD_DECL
20580 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20581 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20582 returns negative, increment nregs only if it returned positive.
20583 (arm_setup_incoming_varargs): Likewise.
20584 (arm_function_arg_boundary): Emit -Wpsabi note if
20585 arm_needs_doubleword_align returns negative, return
20586 DOUBLEWORD_ALIGNMENT only if it returned positive.
20588 2017-04-25 Marek Polacek <polacek@redhat.com>
20591 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20592 first argument to type.
20594 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20597 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20598 type checks to test for compatibility instead of equality.
20600 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20601 Jakub Jelinek <jakub@redhat.com>
20604 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20606 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20607 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20608 the alignment computation, but return their maximum in warn_alignment.
20609 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20610 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20612 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20613 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20616 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20618 * config/arc/simdext.md (dmpyh): Fix typo.
20620 2017-04-25 Richard Biener <rguenther@suse.de>
20622 PR tree-optimization/80492
20623 * alias.c (compare_base_decls): Handle registers with asm
20624 specification conservatively.
20625 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20626 compare_base_decls returning dont-know properly.
20628 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20630 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20631 (legitimate_offset_address_p): New function.
20632 (arc_legitimate_address_p): Use above function.
20634 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20636 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20638 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20640 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20641 ACCH registers whenever they are available.
20643 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20645 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20646 double regs fix when not used.
20648 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20650 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20652 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20653 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20655 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20657 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20658 when emitting short ADD instructions.
20660 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20662 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20664 (cmpsi_cc_c_insn): Likewise.
20665 (cbranchsi4_scratch): Compute proper instruction length using
20666 compact_hreg_operand.
20667 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20669 2017-04-25 Richard Biener <rguenther@suse.de>
20671 PR middle-end/80509
20672 * passes.c (pass_manager::pass_manager): Initialize
20673 m_name_to_pass_map.
20675 2017-04-25 Richard Biener <rguenther@suse.de>
20677 PR tree-optimization/79201
20678 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20680 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20683 * config/s390/vector.md: Split MEM->GPR vector moves for
20684 non-s_operand addresses.
20686 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20689 * config/s390/predicates.md (reload_const_wide_int_operand): New
20691 * config/s390/s390.md ("movti"): Remove d/P alternative.
20692 ("movti_bigconst"): New pattern definition.
20694 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20697 * s390-protos.h (s390_expand_cs_hqi): Removed.
20698 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20699 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20700 modes as well as CCZ1mode and CCZmode.
20701 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20702 signature of s390_emit_compare_and_swap.
20703 (s390_expand_cs_hqi): Likewise, make static.
20704 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20705 compare-and-swap, in some cases.
20706 (s390_expand_cs): Wrapper function.
20707 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20709 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20710 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20711 patterns for small and large integers. Forbid symref memory operands.
20712 Move expander to s390.c. Require cc register.
20713 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20714 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20715 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20716 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20717 symref memory operands. Remove CC mode and call s390_match_ccmode
20719 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20721 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20723 * config/s390/s390.md (define_peephole2): New peephole to help
20724 combining the load-and-test pattern with volatile memory.
20726 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20728 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20729 with CCZmode for TARGET_Z196.
20731 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20733 PR rtl-optimization/80501
20734 * combine.c (make_compound_operation_int): Set subreg_code to SET
20735 even for AND with mask of the sign bit of mode.
20737 PR rtl-optimization/80500
20738 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20739 sum's initial value.
20741 2017-04-25 Julian Brown <julian@codesourcery.com>
20742 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20744 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20746 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20748 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20750 2017-04-25 Julian Brown <julian@codesourcery.com>
20751 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20753 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20754 (thunderx2t99_sha): New Reservation.
20756 2017-04-25 Julian Brown <julian@codesourcery.com>
20757 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20759 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20760 type for 1-element load.
20762 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20764 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20766 2017-04-24 Martin Jambor <mjambor@suse.cz>
20768 PR tree-optimization/80293
20769 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20770 char arrays not totally scalarizable if it is false.
20771 (analyze_all_variable_accesses): Pass correct value in the new
20772 parameter. Add a statistics counter.
20774 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20776 PR middle-end/79931
20777 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20779 2017-04-24 Richard Biener <rguenther@suse.de>
20781 PR tree-optimization/80494
20782 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20783 out for complex types.
20785 2017-04-24 Richard Biener <rguenther@suse.de>
20787 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20788 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20789 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20790 (DFS): Adjust and never fail.
20791 (sccvn_dom_walker::fail): Remove.
20792 (sccvn_dom_walker::before_dom_children): Adjust.
20793 (run_scc_vn): Likewise and never fail.
20794 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20795 (pass_fre::execute): Likewise.
20797 2017-04-24 Richard Biener <rguenther@suse.de>
20799 PR tree-optimization/79725
20800 * tree-ssa-sink.c (statement_sink_location): Return whether
20801 failure reason was zero uses. Move that check later.
20802 (sink_code_in_bb): Deal with zero uses by removing the stmt
20805 2017-04-24 Richard Biener <rguenther@suse.de>
20808 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20809 pointer-based references.
20811 2017-04-24 Richard Biener <rguenther@suse.de>
20814 * pass_manager.h (pass_manager::operator new): Remove.
20815 (pass_manager::operator delete): Likewise.
20816 * passes.c (pass_manager::operator new): Remove.
20817 (pass_manager::operator delete): Likewise.
20818 (pass_manager::pass_manager): Zero individual pass members.
20820 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20823 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20824 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20825 Check "XEXP (src, 1)" operand here.
20826 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20827 Check "XEXP (src, 1)" operand here.
20828 (dimode_scalar_chain::make_vector_copies): Detect count register
20829 of a shift instruction. Zero extend count register from QImode
20830 to DImode to satisfy vector shift pattern count operand predicate.
20831 Substitute vector shift count operand with a DImode copy.
20832 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
20835 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
20837 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
20838 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20839 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
20840 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
20841 (UNSPEC_NOREX_MEM): Remove definition.
20843 2017-04-21 Richard Biener <rguenther@suse.de>
20845 PR tree-optimization/79547
20846 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
20847 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
20848 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
20849 without any constraints.
20851 2017-04-21 Richard Biener <rguenther@suse.de>
20853 PR tree-optimization/78847
20854 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
20856 2017-04-21 Richard Biener <rguenther@suse.de>
20858 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
20859 (build_distinct_type_copy): Likewise.
20860 (build_variant_type_copy): Likewise.
20861 * tree.c (build_qualified_type): Pass down mem-stat info.
20862 (build_distinct_type_copy): Likewise.
20863 (build_variant_type_copy): Likewise.
20865 2017-04-21 Richard Biener <rguenther@suse.de>
20867 PR tree-optimization/80237
20868 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
20870 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
20871 for a simplified result.
20873 2016-04-21 Richard Biener <rguenther@suse.de>
20875 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
20876 sth as strict as a simple_iv but a chrec without symbols and an
20877 operand defined in the loop we are peeling (and not some subloop).
20878 (propagate_constants_for_unrolling): Propagate all constants.
20880 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20883 * config/i386/i386.c (print_reg): Remove assert for disalowed
20884 regno values, call output_operand_lossage instead.
20886 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
20889 * config/i386/constraints.md (Yc): New register constraint.
20890 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
20891 Use Yc constraint for alternative 2 of operand 0. Remove
20892 preferred_for_speed attribute.
20894 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20896 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
20897 lastprivate clauses in SIMT case.
20899 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
20901 * doc/invoke.texi (-Wextra-semi): Document new warning option.
20903 2017-04-20 Richard Biener <rguenther@suse.de>
20905 PR tree-optimization/57796
20906 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
20907 as N scalar stores.
20908 (vect_model_load_cost): Cost gathers as N scalar loads.
20910 2017-04-20 Richard Biener <rguenther@suse.de>
20912 * ggc-page.c (ggc_allocated_p): Rename to ...
20913 (safe_lookup_page_table_entry): ... this and return the lookup
20915 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
20917 2017-04-20 Richard Biener <rguenther@suse.de>
20919 PR tree-optimization/80453
20920 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
20921 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
20922 from the conditions.
20923 (vn_phi_eq): Pass them down.
20924 (vn_phi_lookup): Record them.
20925 (vn_phi_insert): Likewise.
20927 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
20929 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
20930 uninitialized variable warning to avoid buffer overrun.
20932 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
20935 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
20936 is suppressed for '{ 0 }' in C.
20938 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20940 * BASE-VER: Set to 8.0.0.
20942 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
20944 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
20945 priority .init_array and .fini_array section with SECTION_NOTYPE
20948 2017-04-20 Jakub Jelinek <jakub@redhat.com>
20950 PR middle-end/80423
20951 * tree.h (build_array_type): Add typeless_storage default argument.
20952 * tree.c (type_cache_hasher::equal): Also compare
20953 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
20954 (build_array_type): Add typeless_storage argument, set
20955 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
20957 (build_nonshared_array_type): Adjust build_array_type_1 caller.
20958 (build_array_type): Likewise. Add typeless_storage argument.
20960 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
20961 Jakub Jelinek <jakub@redhat.com>
20963 PR tree-optimization/80426
20964 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
20965 operation on symbolic operands, also compute the overflow for the
20966 invariant part when the operation degenerates into a negation.
20968 2017-04-19 Jakub Jelinek <jakub@redhat.com>
20971 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
20972 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
20975 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
20977 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
20980 * config/avr/avr.c (tree.h): Include it.
20981 (cgraph.h): Include it.
20982 (avr_encode_section_info): Don't warn for uninitialized progmem
20983 variable if it's just an alias.
20985 2017-04-19 Richard Biener <rguenther@suse.de>
20988 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
20989 when needed by AutoPGO.
20991 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
20994 * doc/lto.texi: Remove an extra 'that'.
20996 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
20998 PR rtl-optimization/80429
20999 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21000 are only used in debug insns.
21002 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21003 Vladimir Makarov <vmakarov@redhat.com>
21005 * config/sparc/predicates.md (input_operand): Add comment. Return
21006 true for any memory operand when LRA is in progress.
21007 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21009 2017-04-18 Jeff Law <law@redhat.com>
21012 * mips.md ({return,simple_return}_internal): Do not overwrite
21015 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21017 PR tree-optimization/80443
21018 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21019 instead of adding 1, subtract -1 and similarly instead of subtracting
21022 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21024 PR rtl-optimization/80357
21025 * haifa-sched.c (tmp_bitmap): New variable.
21026 (model_recompute): Handle duplicate use records.
21027 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21028 (free_global_sched_pressure_data): Free it.
21030 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21033 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21034 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21035 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21036 instead of SYSTEM_HEADER_DIR.
21038 2017-04-18 Jeff Law <law@redhat.com>
21040 PR middle-end/80422
21041 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21042 predecessors after walking up the insn chain.
21044 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21047 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21048 sizetype type into debug info.
21050 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21053 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21054 unneeded test for TARGET_UPPER_REGS_SF.
21055 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21057 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21060 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21061 instead of gsi_after_labels.
21063 2017-04-18 Jeff Law <law@redhat.com>
21065 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21069 2017-04-13 Jeff Law <law@redhat.com>
21070 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21071 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21073 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21076 * config/avr/avr.c (intl.h): Include it.
21077 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21079 2017-04-18 Martin Liska <mliska@suse.cz>
21081 PR gcov-profile/78783
21082 * gcov-tool.c (gcov_output_files): Validate that destination
21083 file is either removed by the tool or by a user.
21085 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21086 Guy Benyei <guybe@mellanox.com>
21088 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21089 block, and do not negate it, the stored id is already negative.
21091 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21093 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21095 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21098 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21099 masks of options that should be turned off if the VSX vector
21100 options are turned off.
21101 (OTHER_P8_VECTOR_MASKS): Likewise.
21102 (OTHER_VSX_VECTOR_MASKS): Likewise.
21103 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21104 rs6000_disable_incompatible_switches to validate no type switches
21106 (rs6000_incompatible_switch): New function to disallow turning on
21107 other vector options if -mno-vsx, -mno-power8-vector, or
21108 -mno-power9-vector are specified.
21110 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21112 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21114 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21116 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21117 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21118 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21119 (ARG_POINTER_CFA_OFFSET): Likewise.
21121 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21123 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21124 conditions to take advantage of various optimizations.
21126 2017-04-13 Jeff Law <law@redhat.com>
21128 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21129 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21130 (zero_extendsidi2_dext): Likewise.
21132 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21135 * fold-const.c (fold_ternary_loc): Revert
21136 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21139 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21141 PR rtl-optimization/80343
21142 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21143 new scratch pseudo.
21145 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21148 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21149 to ubsan_encode_value.
21151 2017-04-13 Jeff Law <law@redhat.com>
21153 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21154 appearing in DEBUG_INSNs.
21156 2017-04-13 Martin Liska <mliska@suse.cz>
21158 PR gcov-profile/80413
21159 * gcov-io.c (gcov_write_string): Copy to buffer just when
21160 allocated size is greater than zero.
21162 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21165 * dwarf2out.c (decls_for_scope): Ignore declarations of
21166 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21168 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21171 * ipa-visibility.c (non_local_p): Fix typos.
21172 (localize_node): When localizing symbol in same comdat group,
21173 dissolve the group only when we know external symbols are going
21175 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21177 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21179 PR tree-optimization/79390
21180 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21181 order does not result in usable sequence, retry with reversed operand
21187 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21188 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21189 op0 instead of fold_convert_loc (loc, type, arg0).
21191 2017-04-12 Jeff Law <law@redhat.com>
21193 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21194 has a delay slot in the generated code.
21196 * config/cris/cris.md (cris_preferred_reload_class): Return
21197 GENNONACR_REGS rather than GENERAL_REGS.
21199 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21202 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21203 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21204 signedness of the result type.
21206 2017-04-12 Richard Biener <rguenther@suse.de>
21207 Jeff Law <law@redhat.com>
21209 PR tree-optimization/80359
21210 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21211 trim stores to TARGET_MEM_REFs.
21213 2017-04-12 Richard Biener <rguenther@suse.de>
21215 PR tree-optimization/79390
21216 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21217 threading case even more.
21219 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21222 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21223 for quad_address_p for TImode, instead of just not indexed_address.
21225 2017-04-12 Richard Biener <rguenther@suse.de>
21226 Bernd Edlinger <bernd.edlinger@hotmail.de>
21228 PR middle-end/79671
21229 * alias.c (component_uses_parent_alias_set_from): Handle
21230 TYPE_TYPELESS_STORAGE.
21231 (get_alias_set): Likewise.
21232 * tree-core.h (tree_type_common): Add typeless_storage flag.
21233 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21234 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21235 for types containing members with TYPE_TYPELESS_STORAGE.
21236 (place_field): Likewise.
21237 (layout_type): Likewise for ARRAY_TYPE.
21238 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21239 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21240 TYPE_TYPELESS_STORAGE.
21241 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21243 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21246 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21247 first argument to type.
21249 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21253 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21254 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21255 (rs6000_expand_binop_builtin): Likewise.
21256 (rs6000_expand_ternop_builtin): Likewise; also add missing
21257 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21258 vshasigma built-ins.
21259 * doc/extend.texi: Document that vec_xxpermdi's third argument
21260 must be a constant.
21262 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21264 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21265 Use shift_const cost parameter when calculating gain of STV shifts.
21267 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21269 PR rtl-optimization/70478
21270 * lra-constraints.c (process_alt_operands): Check memory for
21271 disfavoring memory insn operand.
21273 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21275 PR middle-end/80100
21276 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21277 left shift in unsigned HOST_WIDE_INT type.
21279 PR rtl-optimization/80385
21280 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21281 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21284 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21285 if they have any depend clauses.
21287 2017-04-11 Martin Liska <mliska@suse.cz>
21290 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21291 * ipa-split.c (split_function): Create a local comdat symbol
21292 if caller is in a comdat group.
21294 2017-04-11 Martin Liska <mliska@suse.cz>
21297 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21300 2017-04-11 Martin Sebor <msebor@redhat.com>
21302 PR middle-end/80364
21303 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21304 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21306 (directive::set_width, directive::set_precision, format_character):
21308 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21311 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21314 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21315 conflict, set target->arch_name instead of target->cpu_name.
21317 2017-04-11 Richard Biener <rguenther@suse.de>
21319 PR tree-optimization/80374
21320 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21321 build_zero_cst, remove fold_convertible_p check again.
21323 2017-04-11 Martin Liska <mliska@suse.cz>
21326 * ubsan.c (instrument_object_size): Do not instrument register
21329 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21332 * config/i386/i386-builtin-types.def
21333 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21334 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21335 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21336 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21337 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21338 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21339 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21340 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21341 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21342 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21343 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21344 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21345 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21346 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21347 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21348 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21349 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21350 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21351 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21352 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21353 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21354 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21355 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21356 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21357 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21358 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21359 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21360 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21361 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21362 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21363 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21364 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21365 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21366 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21367 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21368 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21369 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21370 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21371 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21372 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21373 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21374 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21375 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21376 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21377 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21378 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21380 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21381 flag to second_arg_count, handle 4 argument function type _COUNT
21382 aliases, handle second_arg_count on second argument rather than last.
21384 2017-04-10 Jeff Law <law@redhat.com>
21386 PR tree-optimization/80374
21387 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21388 record anything if we can not convert integer_zero_node to the
21391 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21394 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21395 Enhance special handling given to the TARGET_P9_MINMAX option in
21396 relation to certain other options.
21398 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21400 PR tree-optimization/80153
21401 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21402 remove POINTER_PLUS_EXPR's base part directly, rather than through
21405 2017-04-10 Richard Biener <rguenther@suse.de>
21406 Bin Cheng <bin.cheng@arm.com>
21408 PR tree-optimization/80153
21409 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21410 the first element of pointer type aff_tree. Build result expr in
21412 (add_elt_to_tree): Convert to type unconditionally. Remove other
21413 fold_convert calls.
21414 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21415 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21417 2017-04-10 Richard Biener <rguenther@suse.de>
21419 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21422 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21424 PR rtl-optimization/70478
21425 * lra-constraints.c (curr_small_class_check): New.
21426 (update_and_check_small_class_inputs): New.
21427 (process_alt_operands): Update curr_small_class_check. Disfavor
21428 alternative insn memory operands. Check available regs for small
21431 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21434 * config/mips/mips.opt (-mvirt): Update description.
21435 * doc/invoke.texi (-mvirt): Likewise.
21437 2017-04-10 Richard Biener <rguenther@suse.de>
21439 PR middle-end/80362
21440 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21441 looking for NEGATE_EXPR in -A / -B to A / B folding.
21443 2017-04-10 Martin Liska <mliska@suse.cz>
21445 PR gcov-profile/80224
21446 * gcov.c (print_usage): Fix usage string.
21447 (get_gcov_intermediate_filename): Remove.
21448 (output_gcov_file): Use both for normal and intermediate format.
21449 (generate_results): Do not initialize special file for
21450 intermediate format.
21452 2017-04-10 Richard Biener <rguenther@suse.de>
21454 PR tree-optimization/80304
21455 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21458 2017-04-10 Nathan Sidwell <nathan@acm.org>
21461 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21462 (rs6000_init_builtins): Use it.
21464 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21466 * config/arm/arm.md (<mrc>): Add mode to SET source.
21467 (<mrrc>): Likewise.
21469 2017-04-10 Richard Biener <rguenther@suse.de>
21471 PR middle-end/80344
21472 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21474 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21477 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21478 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21479 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21480 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21481 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21482 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21483 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21484 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21485 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21486 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21487 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21488 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21489 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21490 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21491 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21492 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21493 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21494 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21495 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21496 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21497 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21498 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21499 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21501 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21503 PR rtl-optimization/70478
21504 * lra-constraints.c: Reverse the last patch.
21506 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21508 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21509 Add comment for WCHAR_T.
21511 2017-04-08 Martin Liska <mliska@suse.cz>
21514 2017-04-07 Martin Liska <mliska@suse.cz>
21517 * ipa-split.c (split_function): Add function part to a same comdat
21520 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21523 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21525 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21527 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21528 (vec_pairing): ...to this.
21529 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21530 (rs6000_sched_init): Adjust for name change.
21531 (struct rs6000_sched_context): Likewise.
21532 (rs6000_init_sched_context): Likewise.
21533 (rs6000_set_sched_context): Likewise.
21535 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21541 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21543 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21544 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21545 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21547 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21549 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21551 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21553 PR rtl-optimization/70703
21554 * ira-color.c (update_conflict_hard_regno_costs): Use
21555 int64_t instead of HOST_WIDE_INT.
21557 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21559 PR rtl-optimization/70478
21560 * lra-constraints.c (process_alt_operands): Disfavor alternative
21561 insn memory operands.
21563 2017-04-07 Jeff Law <law@redhat.com>
21565 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21566 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21568 2017-04-07 Martin Liska <mliska@suse.cz>
21571 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21572 Show error message instead of an ICE.
21574 2017-04-07 Martin Liska <mliska@suse.cz>
21577 * ipa-split.c (split_function): Add function part to a same comdat
21580 2017-04-07 Richard Biener <rguenther@suse.de>
21582 PR middle-end/80341
21583 * tree.c (get_unwidened): Also handle ! for_type case for
21585 * convert.c (do_narrow): Split out from ...
21586 (convert_to_integer_1): ... here. Do not pass final truncation
21587 type to get_unwidened for TRUNC_DIV_EXPR.
21589 2017-04-07 Richard Biener <rguenther@suse.de>
21591 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21593 (aff_combination_const): Adjust.
21594 (aff_combination_scale): Likewise.
21595 (aff_combination_add_elt): Likewise.
21596 (aff_combination_add_cst): Likewise.
21597 (aff_combination_convert): Likewise.
21598 (add_elt_to_tree): Likewise. Remove unused argument.
21599 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21601 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21603 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21605 * config/arm/arm.c (arm_default_short_enums): Use
21606 ARM_DEFAULT_SHORT_ENUMS.
21607 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21609 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21612 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21613 members with redundant out-of-class redeclaration.
21615 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21618 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21619 * config/i386/i386.md (*zero_extendsidi2):
21620 Add (?*x,*x) and (?*v,*v) alternatives.
21622 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21625 * config/i386/i386.c (ix86_expand_builtin)
21626 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21627 mode from insn data. Convert operands to insn operand mode.
21628 Copy operands that don't satisfy insn predicate to a register.
21630 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21632 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21635 2017-04-06 Richard Biener <rguenther@suse.de>
21637 PR tree-optimization/80334
21638 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21639 preserve alignment of accesses.
21641 2017-04-06 Richard Biener <rguenther@suse.de>
21643 PR tree-optimization/80262
21644 * tree-sra.c (build_ref_for_offset): Preserve address-space
21646 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21647 Drop useless address-space information on MEM_REF offsets.
21649 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21651 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21653 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21655 PR rtl-optimization/70703
21656 * ira-color.c (update_conflict_hard_regno_costs): Use
21657 HOST_WIDE_INT instead of long.
21659 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21662 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21663 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21665 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21666 for x86_64 target. Handle -m3dnowa option.
21668 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21670 PR rtl-optimization/70703
21671 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21672 (update_conflict_hard_regno_costs): Use long instead of unsigned
21673 arithmetic for cost calculation.
21675 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21676 Bernd Edlinger <bernd.edlinger@hotmail.de>
21679 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21682 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21685 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21686 ptr_mode with Pmode throughout.
21687 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21688 into probe_stack_range and use DImode.
21690 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21693 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21694 call_eh_return is true.
21696 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21698 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21699 Initialize last_match_fntype_index.
21701 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21704 * tree-nvr.c: Include internal-fn.h.
21705 (pass_return_slot::execute): Ignore internal calls without
21708 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21709 Richard Biener <rguenther@suse.de>
21712 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21713 captures used multiple times, except for the last use.
21714 * generic-match-head.c: Include gimplify.h.
21716 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21718 PR tree-optimization/79390
21719 * target.h (struct noce_if_info): Declare.
21720 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21721 * target.def (noce_conversion_profitable_p): New target hook.
21722 * ifcvt.h (struct noce_if_info): New type, moved from ...
21723 * ifcvt.c (struct noce_if_info): ... here.
21724 (noce_conversion_profitable_p): Renamed to ...
21725 (default_noce_conversion_profitable_p): ... this. No longer
21727 (noce_try_store_flag_constants, noce_try_addcc,
21728 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21729 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21730 instead of noce_conversion_profitable_p.
21731 * config/i386/i386.c: Include ifcvt.h.
21732 (ix86_option_override_internal): Don't override
21733 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21734 (ix86_noce_conversion_profitable_p): New function.
21735 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21736 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21737 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21738 * doc/tm.texi: Regenerated.
21740 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21742 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21745 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21748 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21749 instructions for small multiply cores.
21751 2017-04-04 Jeff Law <law@redhat.com>
21753 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21755 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21756 that are not set by the loop over the elements.
21758 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21761 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21762 int mode, convert_modes it to mode as unsigned, otherwise use
21763 lowpart_subreg to mode rather than SImode.
21764 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21765 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21766 Use DImode instead of SImode for the shift count operand.
21767 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21770 2017-04-04 Richard Biener <rguenther@suse.de>
21772 PR middle-end/80281
21773 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21774 arithmetic done for the negate or the plus. Simplify.
21775 (A - (-B) -> A + B): Likewise.
21776 * fold-const.c (split_tree): Make sure to not negate pointers.
21778 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21780 PR rtl-optimization/60818
21781 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21782 a compare of comparisons with the thing compared if this results
21783 in a different machine mode.
21785 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21787 * alias.c (base_alias_check): Fix typo in comment.
21788 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21789 * cgraphunit.c (symbol_table::compile): Likewise.
21790 * collect2.c (maybe_run_lto_and_relink): Likewise.
21791 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21792 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21793 * config/avr/avr.c (avr_map_op_t): Likewise.
21794 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21795 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21796 * config/epiphany/epiphany.md (movcc): Likewise.
21797 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21798 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21800 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21801 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21802 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21803 * config/sh/sh.c (sh_rtx_costs): Likewise.
21804 * fold-const.c (fold_truth_andor): Likewise.
21805 * genautomata.c (collapse_flag): Likewise.
21806 * gengtype.h (struct type::u::s): Likewise.
21807 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21808 * input.c (FORMAT_AMOUNT): Likewise.
21809 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21810 (known_aggs_to_agg_replacement_list): Likewise.
21811 * ipa-inline-analysis.c: Likewise.
21812 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21813 * ipa-polymorphic-call.c
21814 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21815 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21816 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21818 * modulo-sched.c (apply_reg_moves): Likewise.
21819 * omp-expand.c (build_omp_regions_1): Likewise.
21820 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21821 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21822 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21823 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21824 * value-prof.c: Likewise.
21825 * var-tracking.c (val_reset): Likewise.
21827 2017-04-03 Richard Biener <rguenther@suse.de>
21829 PR tree-optimization/80275
21830 * fold-const.c (split_address_to_core_and_offset): Handle
21833 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
21835 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
21836 descriptors is at least equal to that of functions.
21838 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21840 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
21842 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
21845 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
21846 (mov<IMOD4:mode>): New expander.
21847 (*mov<IMOD4:mode>_internal): New insn and split pattern.
21849 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21851 PR rtl-optimization/79405
21852 * fwprop.c (propagations_left): New variable.
21853 (forward_propagate_into): Decrement it.
21854 (fwprop_init): Initialize it.
21855 (fw_prop): If the variable has reached zero, stop propagating.
21856 (fwprop_addr): Ditto.
21858 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21861 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
21862 a FUNCTION_DECL, pass it as decl instead of origin to
21865 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
21867 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
21870 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
21873 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
21874 TARGET_VSX_SMALL_INTEGER.
21876 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21878 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21879 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
21881 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21883 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
21884 extraction from odd-numbered MSA register.
21886 2017-03-31 Jakub Jelinek <jakub@redhat.com>
21888 PR middle-end/80173
21889 * expmed.c (store_bit_field_1): Don't attempt to create
21890 a word subreg out of hard registers wider than word if they
21891 have HARD_REGNO_NREGS of 1 for their mode.
21893 PR middle-end/80163
21894 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
21895 conversions to integer types wider than word and pointer.
21898 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
21899 (rtx_equal_for_cselib_p): Pass 0 to it.
21900 * cselib.c (cselib_hasher::equal): Likewise.
21901 (rtx_equal_for_cselib_1): Add depth argument. If depth
21902 is 128, don't look up VALUE locs and punt. Increment
21903 depth in recursive calls when walking VALUE locs.
21905 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
21907 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
21908 (make_gcov_file_name): Use the canonical path name for generating
21910 (read_line): Fix handling of files with ascii null bytes.
21912 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
21914 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
21915 to initialise a vector register instead
21916 of using a const_int.
21918 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21920 PR translation/80189
21921 * gimplify.c (omp_default_clause): Use %qs instead of %s in
21922 diagnostic messages.
21924 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
21927 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
21928 (dfp_diex_<mode>): Update mode of operand 1.
21929 * doc/extend.texi (dxex, dxexq): Document change to return type.
21930 (diex, diexq): Document change to argument type.
21932 2017-03-30 Martin Jambor <mjambor@suse.cz>
21935 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
21936 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
21937 it reflects the signature changes performed at the callee side.
21938 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
21939 to cgraph_build_function_type_skip_args.
21940 (build_function_decl_skip_args): Adjust call to the above function.
21942 2017-03-30 Jakub Jelinek <jakub@redhat.com>
21945 * config/i386/sse.md
21946 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
21947 register as dest whenever it is a MEM not rtx_equal_p to the
21948 corresponding dup operand, and when forcing into reg move the
21949 reg into the memory afterwards.
21950 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
21951 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
21952 for the force_reg mode.
21953 (avx512vl_vextractf128<mode>): Use register as dest either
21954 always when a MEM, or when it is a MEM not rtx_equal_p to the
21955 corresponding dup operand, or even not when it is a CONST_VECTOR
21956 depending on the mode and lo vs. hi.
21957 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
21959 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
21960 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
21961 Likewise. Require that operands[2] is even.
21962 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
21963 Remove extraneous parens. Require that operands[2] is a multiple
21965 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
21966 operands[0] is a MEM if <mask_applied>, the predicates/constraints
21967 disallow memory then.
21969 2017-03-30 Richard Biener <rguenther@suse.de>
21971 PR tree-optimization/77498
21972 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
21973 to non-constants over backedges.
21975 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
21977 PR rtl-optimization/80233
21978 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
21979 as last_combined_insn. Do not test for BARRIER_P separately.
21981 2017-03-29 Andreas Schwab <schwab@suse.de>
21984 * calls.c (prepare_call_address): Convert funexp to Pmode before
21985 copying to temp reg.
21987 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21989 PR tree-optimization/80158
21990 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
21991 Handle possible future case of more than one alternate
21993 (replace_rhs_if_not_dup): Likewise.
21994 (replace_one_candidate): Likewise.
21996 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
21998 PR rtl-optimization/80193
21999 * ira.c (ira): Do not check allocation for LRA.
22001 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22003 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22004 (nvptx_output_simt_exit): Declare.
22005 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22006 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22007 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22008 (nvptx_declare_function_name): ...here. Emit declaration of local
22009 memory space buffer for omp_simt_enter insn.
22010 (nvptx_output_unisimt_switch): New.
22011 (nvptx_output_softstack_switch): New.
22012 (nvptx_output_simt_enter): New.
22013 (nvptx_output_simt_exit): New.
22014 * config/nvptx/nvptx.h (struct machine_function): New fields
22015 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22016 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22017 (UNSPECV_SIMT_EXIT): Ditto.
22018 (omp_simt_enter_insn): New insn.
22019 (omp_simt_enter): New expansion.
22020 (omp_simt_exit): New insn.
22021 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22023 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22024 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22025 (expand_GOMP_SIMT_EXIT): New.
22026 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22027 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22028 (GOMP_SIMT_EXIT): Ditto.
22029 * target-insns.def (omp_simt_enter): New insn.
22030 (omp_simt_exit): Ditto.
22031 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22033 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22034 (lower_rec_input_clauses): Likewise.
22035 (lower_lastprivate_clauses): Handle SIMT privatization.
22037 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22038 (ompdevlow_adjust_simt_enter): New.
22039 (find_simtpriv_var_op): New.
22040 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22041 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22043 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22044 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22045 (copy_decl_for_dup_finish): Ditto.
22047 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22049 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22052 * config/i386/i386.c (ix86_option_override_internal): Always
22053 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22055 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22057 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22059 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22061 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22062 mark new edge's irreducible flag accordign to it.
22063 (vect_do_peeling): Check loop preheader edge's irreducible flag
22064 and pass it to function slpeel_add_loop_guard.
22066 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22068 PR tree-optimization/80218
22069 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22070 Update block frequencies and counts.
22072 2017-03-28 Richard Biener <rguenther@suse.de>
22074 PR tree-optimization/78644
22075 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22076 of a simplification result we may not use it at all.
22078 2017-03-28 Richard Biener <rguenther@suse.de>
22081 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22082 without arguments, generate default definition of a SSA name.
22084 2017-03-28 Richard Biener <rguenther@suse.de>
22086 PR middle-end/80222
22087 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22088 TYPE_REF_CAN_ALIAS_ALL references.
22089 * fold-const.c (fold_indirect_ref_1): Likewise.
22091 2017-03-28 Martin Liska <mliska@suse.cz>
22094 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22095 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22097 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22098 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22100 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22101 (EXTRA_SPECS): Define.
22102 (SUBTARGET_EXTRA_SPECS): Likewise.
22103 (SUBTARGET_CPP_SPEC): Likewise.
22104 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22105 SUBTARGET_EXTRA_SPECS.
22106 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22108 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22110 * config/arc/simdext.md (vst64_insn): Update pattern.
22111 (vld32wh_insn): Likewise.
22112 (vld32wl_insn): Likewise.
22113 (vld64_insn): Likewise.
22114 (vld32_insn): Likewise.
22116 2017-03-28 Marek Polacek <polacek@redhat.com>
22119 * fold-const.c (fold_comparison): Use protected_set_expr_location
22120 instead of SET_EXPR_LOCATION.
22122 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22124 * tree.c (add_expr): Avoid name lookup warning.
22126 2017-03-27 Jeff Law <law@redhat.com>
22128 PR tree-optimization/80216
22129 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22130 function name. Limit recursion depth.
22131 (record_temporary_equivalences): Corresponding changes.
22133 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22135 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22138 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22141 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22143 * cfgcleanup.c (reg_note_cfa_p): New array.
22144 (insns_have_identical_cfa_notes): New function.
22145 (old_insns_match_p): Don't cross-jump in between /f
22146 and non-/f instructions. If both i1 and i2 are frame related,
22147 verify all CFA notes, their order and content.
22149 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22152 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22153 HImode and SImode with zero extend to DImode to one insn.
22154 (bswap<mode>2_extenddi): Likewise.
22155 (bswapsi2_extenddi): Likewise.
22156 (bswaphi2_extendsi): Likewise.
22157 (bswaphi2): Combine bswap HImode and SImode into one insn.
22158 Separate memory insns from swapping register.
22159 (bswapsi2): Likewise.
22160 (bswap<mode>2): Likewise.
22161 (bswaphi2_internal): Delete, no longer used.
22162 (bswapsi2_internal): Likewise.
22163 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22164 store, and gpr<-gpr swap insns.
22165 (bswap<mode>2_store): Likewise.
22166 (bswaphi2_reg): Register only splitter, combine with the splitter.
22167 (bswaphi2 splitter): Likewise.
22168 (bswapsi2_reg): Likewise.
22169 (bswapsi2 splitter): Likewise.
22170 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22171 the insns into load, store, and register/register insns.
22172 (bswapdi2_ldbrx): Likewise.
22173 (bswapdi2_load): Likewise.
22174 (bswapdi2_store): Likewise.
22175 (bswapdi2_reg): Likewise.
22177 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22179 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22180 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22182 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22185 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22187 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22188 special handling for target option conflicts between dform
22189 options (-mpower9-dform, -mpower9-dform-vector,
22190 -mpower9-dform-scalar) and -mno-direct-move.
22192 2017-03-27 Richard Biener <rguenther@suse.de>
22194 PR tree-optimization/80181
22195 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22197 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22199 * config/arc/predicates.md (move_double_src_operand): Replace the
22200 call to move_double_src_operand with a call to address_operand.
22202 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22204 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22205 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22206 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22208 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22210 * config/arc/predicates.md (long_immediate_loadstore_operand):
22211 Consider scaled addresses cases.
22213 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22215 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22216 restored when in interrupt.
22217 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22218 doesn't have delay slot.
22220 2017-03-27 Richard Biener <rguenther@suse.de>
22223 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22224 inlined thunk clones.
22226 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22229 * asan.c (instrument_derefs): Copy over last operand from
22230 original COMPONENT_REF to the new COMPONENT_REF with
22231 DECL_BIT_FIELD_REPRESENTATIVE.
22232 * ubsan.c (instrument_object_size): Likewise.
22234 2017-03-27 Richard Biener <rguenther@suse.de>
22236 PR tree-optimization/80170
22237 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22238 sure DR/SCEV didnt fold in constants we do not see when looking
22239 at the reference base alignment.
22241 2017-03-27 Richard Biener <rguenther@suse.de>
22243 PR middle-end/80171
22244 * gimple-fold.c (fold_ctor_reference): Properly guard against
22245 NULL return value from canonicalize_constructor_val.
22247 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22250 * config/i386/i386.c (ix86_expand_builtin)
22251 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22252 flags reg setting and flags reg using instructions.
22253 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22254 clobbering instructions to zero extend op2.
22256 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22258 * doc/install.texi (Configuration) <--with-aix-soname>:
22259 Update link to AIX ld.
22261 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22263 PR rtl-optimization/80160
22264 PR rtl-optimization/80159
22265 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22266 reg_alternate_class into account.
22268 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22271 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22272 to consider in curr_insn_transform.
22274 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22276 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22277 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22278 and emit_mode_inner.
22280 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22282 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22283 argument to the overloaded builtin variants. Use the new flag to
22284 deprecate certain builtin variants.
22285 * config/s390/s390-builtin-types.def: Add new builtin types.
22286 * config/s390/s390-builtins.h: Support new flags field for
22287 overloaded builtins.
22288 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22289 (s390_macro_to_expand): Enable vector float data type.
22290 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22291 builtins by incrementing the __VEC__ version number.
22292 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22294 (s390_resolve_overloaded_builtin): Emit error messages depending
22295 on the builtin flags.
22296 * config/s390/s390.c (s390_expand_builtin): Support additional
22297 flags argument. Change error message to match the messages
22298 emitted in s390-c.c.
22299 * config/s390/s390.md: New UNSPEC_* constants.
22300 (op_type): Add new instruction types.
22301 * config/s390/vecintrin.h: Add new builtins and test data class
22303 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22304 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22305 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22306 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22307 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22308 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22310 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22311 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22312 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22313 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22315 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22316 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22317 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22318 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22320 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22321 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22322 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22323 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22324 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22325 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22326 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22328 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22329 ("vec_scatter_element<V_HW_4:mode>_DI")
22330 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22331 ("vec_fpint<mode>", "vflls")
22332 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22333 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22334 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22335 ("*vec_cmphe<mode>_cc"): ... these.
22337 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22338 mode constant instead of magic value.
22340 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22342 * config/s390/s390.c (s390_expand_vec_compare): Support other
22343 vector floating point modes than just V2DF.
22344 (s390_expand_vcond): Likewise.
22345 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22346 (s390_cannot_change_mode_class): Prevent mode changes between TF
22347 and V1TF in vector registers.
22348 * config/s390/s390.md (DF, SF): New mode attributes.
22349 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22350 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22351 SFmode support for VRs.
22352 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22354 (VFT, VF_HW): New mode iterators.
22355 (vw, sdx): New mode attributes.
22356 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22357 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22358 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22359 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22360 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22361 also the new vector floating point modes. Renaming to ...
22363 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22364 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22365 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22366 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22367 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22368 ("vec_unordered<mode>"): ... these.
22370 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22371 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22372 ("*vec_extendv2df"): New insn definitions.
22374 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22376 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22377 ("mulditi3_2", "*muldi3_sign"): New patterns.
22378 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22379 rename the pattern definition.
22381 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22383 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22385 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22389 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22390 instruction if possible.
22391 * config/s390/vector.md (vec_halfnumelts): New mode
22393 ("*vec_vllezlf<mode>"): New pattern.
22395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22397 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22398 ("popcountv4si2", "popcountv2di2"): Rename to ...
22399 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22400 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22402 ("popcount<mode>2_vxe"): New pattern.
22404 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22406 * common/config/s390/s390-common.c (processor_flags_table): Add
22408 * config.gcc: Add arch12.
22409 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22410 Default to arch12 for unknown CPU model numbers.
22411 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22412 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22413 PROCESSOR_max sanity check.
22414 * config/s390/s390-opts.h (enum processor_type): Add
22416 * config/s390/s390.c (processor_table): Add arch12.
22417 (s390_expand_builtin): Add check for B_VXE flag.
22418 (s390_issue_rate): Add PROCESSOR_ARCH12.
22419 (s390_get_sched_attrmask): Likewise.
22420 (s390_get_unit_mask): Likewise.
22421 (s390_sched_score): Enable z13 scheduling for arch12.
22422 (s390_sched_reorder): Likewise.
22423 (s390_sched_variable_issue): Likewise.
22424 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22426 (s390_tune_attr): Use z13 scheduling also for arch12.
22427 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22428 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22429 (TARGET_VXE_P): New macros.
22430 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22431 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22432 * config/s390/s390.opt: Add arch12 as processor_type.
22434 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22436 * config/s390/s390.md
22437 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22438 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22439 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22441 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22442 Rename expanders to ...
22444 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22445 ("fixuns_truncdddi2_emu"): ... these.
22447 ("fixuns_trunc<mode>si2_emu"): New expander.
22449 ("*fixuns_truncdfdi2_z13"): Rename to ...
22450 ("*fixuns_truncdfdi2_vx"): ... this.
22452 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22454 * config/s390/2964.md: Remove the single element vector compare
22455 instructions which are no longer used.
22456 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22458 (s390_canonicalize_comparison): Remove handling of DFmode
22460 (s390_expand_vec_compare_scalar): Remove function.
22461 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22462 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22464 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22468 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22470 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22471 will anyway by matched by mov<mode>_64dfp.
22473 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22475 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22476 vlef/vstef. Add missing operand to vleif.
22478 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22480 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22481 pair for all vector types with 64 bit elements.
22482 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22483 * config/s390/vector.md (V_HW_64): ... here.
22484 (V_128_NOSINGLE): New mode iterator.
22485 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22486 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22487 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22488 ("*vec_load_pairv2di"): Change to ...
22489 ("*vec_load_pair<mode>"): ... this one.
22491 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22493 * config/s390/constraints.md: Add comments.
22494 (jKK): Reject element sizes > 8 bytes.
22495 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22497 * config/s390/s390.md: Add the s_operand checks formerly in
22498 s390_split_ok_p to various splitters where they are still
22500 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22501 for 128 bit vectors. Plus two splitters.
22503 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22505 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22508 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22511 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22512 error if the boundary argument is not constant.
22514 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22516 PR rtl-optimization/80112
22517 * loop-doloop.c (doloop_condition_get): Don't check condition
22518 if cmp isn't SET with IF_THEN_ELSE src.
22520 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22522 PR tree-optimization/80158
22523 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22524 replacing a candidate statement, also replace it for the
22525 candidate's alternate interpretation.
22526 (replace_rhs_if_not_dup): Likewise.
22527 (replace_one_candidate): Likewise.
22529 2017-03-24 Richard Biener <rguenther@suse.de>
22531 PR tree-optimization/80167
22532 * graphite-isl-ast-to-gimple.c
22533 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22535 (translate_isl_ast_to_gimple::get_rename): Likewise.
22537 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22539 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22540 handling of certain combinations of target options, including the
22541 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22542 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22544 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22547 * config/arm/arm.md (*load_multiple): Add reload_completed to
22548 matching condition.
22550 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22551 Richard Biener <rguenth@suse.de>
22553 PR tree-optimization/79908
22554 PR tree-optimization/80136
22555 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22556 been cast away, gimplify_and_add suffices.
22558 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22560 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22562 2017-03-23 Richard Biener <rguenther@suse.de>
22564 PR tree-optimization/80032
22565 * gimplify.c (gimple_push_cleanup): Forced unconditional
22566 cleanups still have to go to the conditional_cleanups
22569 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22571 PR tree-optimization/80072
22572 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22574 (next_operand_entry_id): Change type to unsigned int.
22575 (sort_by_operand_rank): Make sure to return the right return value
22576 even if unsigned fields are bigger than INT_MAX.
22577 (struct oecount): Change cnt and id type to unsigned int.
22578 (oecount_hasher::equal): Formatting fix.
22579 (oecount_cmp): Make sure to return the right return value
22580 even if unsigned fields are bigger than INT_MAX.
22581 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22584 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22585 TREE_READONLY on result if writing it more than once.
22588 * doc/invoke.texi (-fsanitize=thread): Document that with
22589 -fnon-call-exceptions atomics are not able to throw
22593 * tsan.c: Include tree-eh.h.
22594 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22595 maybe_clean_or_replace_eh_stmt where needed.
22596 (instrument_memory_accesses): Add cfg_changed argument.
22597 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22598 if it returned true.
22599 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22601 PR rtl-optimization/63191
22602 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22603 wrapper function, moved the whole old content into ...
22604 (ix86_delegitimize_address_1): ... this. New inline function.
22605 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22606 true as last argument instead of ix86_delegitimize_address.
22608 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22610 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22611 cortexa57_branch_cost.
22613 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22615 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22617 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22620 * doc/md.texi (Constraints): Document wA constraint.
22621 * config/rs6000/constraints.md (wA): New.
22622 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22623 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22624 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22625 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22627 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22630 * gimplify.c (is_oacc_declared): New function.
22631 (oacc_default_clause): Use it to set default flags for acc declared
22632 variables inside parallel regions.
22633 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22634 declared variables.
22635 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22636 declare attribute to any decl as necessary.
22638 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22641 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22642 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22643 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22644 (arm_arch_lpae): This.
22645 * config/arm/arm.c (arm_arch7ve): Rename into ...
22646 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22647 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22650 2017-03-22 Martin Liska <mliska@suse.cz>
22653 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22654 error message instead of an ICE.
22656 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22658 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22660 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22662 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22664 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22667 2017-03-21 Martin Sebor <msebor@redhat.com>
22669 * doc/extend.texi: Use "cannot" instead of "can't."
22670 * doc/hostconfig.texi: Same.
22671 * doc/install.texi: Same.
22672 * doc/invoke.texi: Same.
22673 * doc/loop.texi: Same.
22674 * doc/md.texi: Same.
22675 * doc/objc.texi: Same.
22676 * doc/rtl.texi: Same.
22677 * doc/tm.texi: Same.
22678 * doc/tm.texi.in: Same.
22679 * doc/trouble.texi: Same.
22681 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22684 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22685 (collect_checksum_attributes): Set it.
22686 (die_checksum_ordered): Use it.
22688 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22690 PR tree-optimization/79908
22691 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22692 change: For a VA_ARG whose LHS has been cast away, use
22693 force_gimple_operand to construct the side effects.
22695 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22697 PR translation/80001
22698 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22699 more amenable to translation.
22700 (oacc_loop_auto_partitions): Likewise.
22702 2017-03-21 Marek Polacek <polacek@redhat.com>
22703 Martin Sebor <msebor@redhat.com>
22705 PR tree-optimization/80109
22706 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22707 on INTEGRAL_TYPE_P.
22709 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22710 Segher Boessenkool <segher@kernel.crashing.org>
22713 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22714 check reg_used_between_p between insn and one of succ or succ2
22715 depending on if succ is artificial insn not inserted into insn
22718 2017-03-21 Martin Liska <mliska@suse.cz>
22720 PR gcov-profile/80081
22721 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22722 * doc/gcc.texi: Include gcov-dump stuff.
22723 * doc/gcov-dump.texi: New file.
22725 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22727 PR rtl-optimization/79150
22728 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22729 conditional jump, if the jump is the last insn of the loop.
22731 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22732 Richard Biener <rguenth@suse.de>
22734 PR tree-optimization/79908
22735 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22736 been cast away, use force_gimple_operand to construct the side
22739 2017-03-21 Martin Liska <mliska@suse.cz>
22741 PR libfortran/79956
22742 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22745 2017-03-21 Brad Spengler <spender@grsecurity.net>
22748 * plugin.c (htab_hash_plugin): New function.
22749 (add_new_plugin): Use it and adjust.
22750 (parse_plugin_arg_opt): Adjust.
22751 (init_one_plugin): Likewise.
22753 2017-03-21 Richard Biener <rguenther@suse.de>
22755 PR tree-optimization/80032
22756 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22757 if set force the cleanup to happen unconditionally.
22758 (gimplify_target_expr): Push inserted clobbers with force_uncond
22759 to avoid them being removed by control-dependent DCE.
22761 2017-03-21 Richard Biener <rguenther@suse.de>
22763 PR tree-optimization/80122
22764 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22765 va_arg_pack_len when the inlined call stmt requires pack
22767 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22769 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22772 * tsan.c (instrument_builtin_call): If the memory model argument
22773 is not a constant, assume it is valid.
22776 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22779 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22781 PR rtl-optimization/79910
22782 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22783 if its dest is used by an insn before I2 (other than the combined
22784 insns themselves, which are properly handled already).
22786 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22789 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22791 * combine.c (record_used_regs): New static function.
22792 (try_combine): Handle situations where there is an additional
22793 instruction between I2 and I3 which needs to have a LOG_LINK
22797 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22799 * combine.c (try_combine): Delete redundant i1 test. Call
22800 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22802 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22805 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22806 alternatives 13/14.
22808 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22810 PR tree-optimization/80054
22811 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22812 the optimization if a PHI or any of its arguments is not dominated
22813 by the candidate's basis. Use gphi* rather than gimple* as
22815 (replace_profitable_candidates): Clean up a gimple* variable that
22816 should be a gphi* variable.
22818 2017-03-20 Martin Sebor <msebor@redhat.com>
22821 * doc/extend.texi (attribute constructor): Document present limitation.
22823 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22826 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22827 __POWER9_VECTOR__ #ifdef control, change template definition to
22828 use Power9-specific built-in function.
22829 (vec_any_eq): Likewise.
22830 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
22831 to control outcomes from this test.
22832 (vector_ae_<mode>p): For VEC_F modes, likewise.
22834 2017-03-20 Ian Lance Taylor <iant@google.com>
22836 * config/i386/i386.c (ix86_function_regparm): Save an extra
22837 register for -fsplit-stack with DECL_STATIC_CHAIN.
22839 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22842 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
22843 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
22845 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
22847 * config/riscv/riscv.c (riscv_print_operand): Use "fence
22849 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
22852 2017-03-20 Marek Polacek <polacek@redhat.com>
22855 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
22857 2017-03-20 Richard Biener <rguenther@suse.de>
22859 PR tree-optimization/80113
22860 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
22861 allocate extra SSA name for PHI def.
22862 (add_close_phis_to_outer_loops): Likewise.
22863 (add_close_phis_to_merge_points): Likewise.
22864 (copy_loop_close_phi_args): Likewise.
22865 (copy_cond_phi_nodes): Likewise.
22867 2017-03-20 Martin Liska <mliska@suse.cz>
22869 PR middle-end/79753
22870 * tree-chkp.c (chkp_build_returned_bound): Do not build
22871 returned bounds for a LHS that's not a BOUNDED_P type.
22873 2017-03-20 Martin Liska <mliska@suse.cz>
22877 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
22878 COMPLEX_CST and VECTOR_CST.
22880 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22883 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
22884 target operand. A new splitter adds the clobber statement in case
22885 the target operand is dead anyway.
22887 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
22889 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
22890 to age-old versions of binutils and glibc.
22892 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
22894 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
22896 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22898 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
22900 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
22902 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
22903 requirement for binutils 2.13.
22905 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22907 * combine.c (try_combine): Delete redundant i1 test. Call
22908 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22910 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
22912 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
22913 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
22915 <riscv64-*-elf>: Re-arrange section
22916 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
22917 <riscv32-*-linux>: Likewise.
22918 <riscv64-*-elf>: Likewise
22919 <riscv64-*-linux>: Likewise.
22921 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
22924 * aarch64.opt(verbose-cost-dump): Fix typo.
22926 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
22929 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
22930 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
22932 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22934 * reload.c (find_reloads): When reloading a nonoffsettable address,
22935 use RELOAD_OTHER for it and its address reloads.
22937 PR rtl-optimization/79910
22938 * combine.c (record_used_regs): New static function.
22939 (try_combine): Handle situations where there is an additional
22940 instruction between I2 and I3 which needs to have a LOG_LINK
22943 2017-03-17 Jeff Law <law@redhat.com>
22945 PR tree-optimization/71437
22946 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
22947 conditional in the hash table first.
22948 (vrp_dom_walker::before_dom_children): Extract condition from
22949 ASSERT_EXPR. Record condition, its inverion and any implied
22950 conditions as well.
22952 2017-03-17 Marek Polacek <polacek@redhat.com>
22953 Markus Trippelsdorf <markus@trippelsdorf.de>
22955 PR tree-optimization/80079
22956 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
22959 2017-03-17 Richard Biener <rguenther@suse.de>
22961 PR middle-end/80075
22962 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
22963 Properly verify the LHS before the RHS possibly claims to be
22965 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
22968 2017-03-17 Martin Jambor <mjambor@suse.cz>
22970 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
22971 (List of -O2 options): Likewise.
22972 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
22975 2017-03-17 Tom de Vries <tom@codesourcery.com>
22977 * gcov-dump.c (print_usage): Print bug_report_url.
22979 2017-03-17 Richard Biener <rguenther@suse.de>
22981 PR middle-end/80050
22982 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
22983 (parser::peek): Likewise.
22985 2017-03-17 Richard Biener <rguenther@suse.de>
22987 PR tree-optimization/80048
22988 * sese.c (free_sese_info): Properly release rename_map and
22989 copied_bb_map elements.
22991 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
22993 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
22994 Add linked-list forward and backlinks. Insert on
22995 construction, remove on destruction.
22996 (class pass_store_merging): Add m_stores_head field.
22997 (pass_store_merging::terminate_and_process_all_chains):
22998 Iterate over m_stores_head list.
22999 (pass_store_merging::terminate_all_aliasing_chains):
23001 (pass_store_merging::execute): Check for debug stmts first.
23002 Push new chains onto the m_stores_head stack.
23004 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23007 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23008 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23009 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23011 2017-03-16 Jeff Law <law@redhat.com>
23013 PR tree-optimization/71437
23014 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23015 member function. Implementation moved into after_dom_children
23016 member function and into the threader's thread_outgoing_edges
23018 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23019 some code into new thread_outgoing_edges.
23020 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23021 definition. Simplify marker handling (do it here). Assume we always
23022 have the available expression and the const/copies tables.
23023 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23025 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23026 * tree-vrp.c (equiv_stack): No longer file scoped.
23027 (vrp_dom_walker): New class.
23028 (vrp_dom_walker::before_dom_children): New member function.
23029 (vrp_dom_walker::after_dom_children): Likewise.
23030 (identify_jump_threads): Setup domwalker. Use it rather than
23031 walking edges in a random order by hand. Simplify setup/finalization.
23032 (finalize_jump_threads): Remove.
23033 (vrp_finalize): Do not call identify_jump_threads here.
23034 (execute_vrp): Do it here instead and call thread_through_all_blocks
23037 PR tree-optimization/71437
23038 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23040 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23042 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23043 (dom_opt_dom_walker::thread_across_edge): Remove
23044 handle_dominating_asserts argument. All callers changed.
23045 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23046 changes. Remove calls to lhs_of_dominating_assert. Other
23047 uses of handle_dominating_asserts turn into unconditional code
23048 (simplify_control_stmt_condition_1): Likewise.
23049 (simplify_control_stmt_condition): Likewise.
23050 (thread_through_normal_block, thread_across_edge): Likewise.
23051 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23052 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23053 object if it is not an SSA_NAME.
23054 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23055 before calling into the VRP specific simplifiers.
23056 (identify_jump_threads): Remove handle_dominating_asserts
23059 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23062 * tree-diagnostic.c (default_tree_printer): No longer static.
23063 * tree-diagnostic.h (default_tree_printer): New prototype.
23065 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23067 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23068 Change ins into fmov.
23070 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23072 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23073 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23074 Use h_con constraint for operand 1.
23075 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23076 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23078 2017-03-15 Jeff Law <law@redhat.com>
23080 PR tree-optimization/71437
23081 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23082 (record_temporary_equivalences): Use it.
23084 PR tree-optimization/71437
23085 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23086 tree-ssa-scopedtables.
23087 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23088 (record_conditions, record_cond, vuse_eq): Likewise.
23089 (record_edge_info): Adjust to API tweak of record_conditions.
23090 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23091 (record_temporary_equivalences, optimize_stmt): Likewise.
23092 (eliminate_redundant_computations): Likewise.
23093 (record_equivalences_from_stmt): Likewise.
23094 * tree-ssa-scopedtables.c: Include options.h and params.h.
23095 (vuse_eq): New function, moved from tree-ssa-dom.c
23096 (build_and_record_new_cond): Likewise.
23097 (record_conditions): Likewise. Accept vector of conditions rather
23098 than edge_equivalence structure for first argument.
23099 for the first argument.
23100 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23101 from tree-ssa-dom.c.
23102 (avail_exprs_stack::record_cond): Likewise.
23103 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23104 from tree-ssa-dom.c.
23105 (avail_exprs_stack): Add new member functions lookup_avail_expr
23107 (record_conditions): Declare.
23109 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23112 * lra-constraints.c (process_alt_operands): Increase reject for
23113 reloading an input/output operand.
23115 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23118 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23119 insns to convert from signed/unsigned char/short to IEEE 128-bit
23121 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23123 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23126 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23127 subreg of inner mode for values already in registers.
23129 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23131 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23132 iteration reg is used after the loop.
23134 2017-03-14 Martin Sebor <msebor@redhat.com>
23136 PR tree-optimization/79800
23137 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23138 precision in negative-positive range.
23139 (format_floating): Call non-const overload with adjusted precision.
23141 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23144 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23147 2017-03-14 Martin Sebor <msebor@redhat.com>
23149 PR middle-end/80020
23150 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23151 * builtins.def (aligned_alloc): Use it.
23154 * Makefile.in (GTFILES): Add calls.c.
23155 * calls.c: Include "gt-calls.h".
23157 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23159 PR rtl-optimization/79728
23160 * regs.h (struct target_regs): New field
23161 x_contains_allocatable_regs_of_mode.
23162 (contains_allocatable_regs_of_mode): New macro.
23163 * reginfo.c (init_reg_sets_1): Initialize it, and change
23164 contains_reg_of_mode so it includes global regs as well.
23165 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23166 rather than contains_regs_of_mode.
23168 2017-03-14 Martin Liska <mliska@suse.cz>
23170 * doc/invoke.texi: Document options that can't be combined with
23171 -fcheck-pointer-bounds.
23173 2017-03-14 Martin Liska <mliska@suse.cz>
23175 PR middle-end/79831
23176 * doc/invoke.texi (-Wchkp): Document the option.
23178 2017-03-14 Martin Liska <mliska@suse.cz>
23180 * Makefile.in: Install gcov-dump.
23182 2017-03-14 Martin Liska <mliska@suse.cz>
23184 * multiple_target.c (expand_target_clones): Bail out for
23185 an invalid attribute.
23187 2017-03-14 Richard Biener <rguenther@suse.de>
23189 * alias.c (struct alias_set_entry): Pack properly.
23190 * cfgloop.h (struct loop): Likewise.
23191 * cse.c (struct set): Likewise.
23192 * ipa-utils.c (struct searchc_env): Likewise.
23193 * loop-invariant.c (struct invariant): Likewise.
23194 * lra-remat.c (struct cand): Likewise.
23195 * recog.c (struct change_t): Likewise.
23196 * rtl.h (struct address_info): Likewise.
23197 * symbol-summary.h (function_summary): Likewise.
23198 * tree-loop-distribution.c (struct partition): Likewise.
23199 * tree-object-size.c (struct object_size_info): Likewise.
23200 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23201 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23202 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23203 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23204 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23205 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23206 (struct _stmt_vec_info): Likewise.
23208 2017-03-14 Martin Liska <mliska@suse.cz>
23211 * multiple_target.c (create_dispatcher_calls): Check that
23212 a target can create a function dispatcher.
23214 2017-03-14 Martin Liska <mliska@suse.cz>
23217 * multiple_target.c (expand_target_clones): Drop local.local
23218 flag for default implementation.
23220 2017-03-14 Richard Biener <rguenther@suse.de>
23222 PR tree-optimization/80030
23223 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23225 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23227 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23228 gcc_fallthrough() instead of __attribute__((fallthrough));
23230 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23232 * doc/gcc.texi: Remove "up" link to (DIR).
23233 * doc/gccint.texi: Ditto.
23235 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23237 * doc/install.texi (Specific) <avr>: Remove reference to
23240 2017-03-13 Jeff Law <law@redhat.com>
23242 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23243 attribute rather than comments.
23245 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23246 match_scratch operand is highest.
23248 2017-03-13 Martin Liska <mliska@suse.cz>
23250 PR middle-end/78339
23251 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23252 is a CHKP clone, use original declaration.
23254 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23256 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23257 (arc_conditional_register_usage): Use a different allocation order
23258 when optimizing for size.
23259 * common/config/arc/arc-common.c (arc_option_optimization_table):
23260 Section anchors default on when optimizing for size.
23262 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23264 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23266 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23268 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23269 * config/arc/arc.md (cpu_facility): Add cd variant.
23270 (*movqi_insn): Add code density variant.
23271 (*movhi_insn): Likewise.
23272 (*movqi_insn): Likewise.
23273 (*addsi3_mixed): Likewise.
23274 (subsi3_insn): Likewise.
23276 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23278 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23280 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23282 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23283 expressions with MINUS and UNARY ops.
23285 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23288 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23290 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23291 between vec_select and vector argument.
23292 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23293 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23294 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23295 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23296 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23297 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23299 2017-03-13 Richard Biener <rguenther@suse.de>
23302 * params.def (vect-max-peeling-for-alignment): Fix typo.
23304 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23306 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23307 issue that only occurred with binutils below 2.18.
23309 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23311 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23312 refer to binutils 2.11/2.12 minimum.
23314 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23316 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23317 ftp.kernel.org and simplify binutils requirement.
23319 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23321 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23323 (Optimize Options): Ditto. Also remove redundancy.
23325 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23327 PR translation/79848
23328 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23330 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23331 to G_ to avoid double translation.
23333 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23335 PR translation/79923
23336 * auto-profile.c (get_combined_location): Convert leading
23337 character of diagnostics to lower case and remove trailing period.
23338 (read_profile): Likewise for various diagnostics.
23339 * config/arm/arm.c (arm_option_override): Remove trailing period
23340 from various diagnostics.
23341 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23342 (msp430_expand_delay_cycles): Likewise.
23344 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23347 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23348 full command-line argument, rather than just "str".
23349 (aarch64_validate_march): Likewise.
23350 (aarch64_validate_mtune): Likewise.
23352 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23354 PR rtl-optimization/78911
23355 * lra-assigns.c (must_not_spill_p): New function.
23356 (spill_for): Use it.
23358 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23360 PR tree-optimization/79981
23361 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23362 ATOMIC_COMPARE_EXCHANGE ifn result.
23363 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23364 IFN_ATOMIC_COMPARE_EXCHANGE.
23366 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23369 * opts.c (parse_sanitizer_options): Add missing question mark to
23370 "did you mean" message.
23372 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23374 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23376 (VMULEUH_UNS): Likewise.
23377 (VMULOUB_UNS): Likewise.
23378 (VMULOUH_UNS): Likewise.
23379 * config/rs6000/rs6000.c (builtin_function_type): Remove
23380 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23382 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23385 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23386 x with result of extra_parsing_for_operand_code_0.
23387 (function_reader::extra_parsing_for_operand_code_0): Convert
23388 return type from void to rtx, returning x. When reading
23389 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23390 larger size containing struct block_symbol.
23392 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23394 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23395 -mfloat128-hardware without -m64.
23397 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23400 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23401 entries to the case statement that marks unsigned arguments to
23402 overloaded functions.
23404 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23407 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23409 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23412 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23413 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23415 2017-03-10 Martin Liska <mliska@suse.cz>
23419 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23420 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23423 2017-03-10 Olivier Hainque <hainque@adacore.com>
23425 * tree-switch-conversion (array_value_type): Start by resetting
23426 candidate type to it's main variant.
23428 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23430 PR rtl-optimization/79909
23431 * combine.c (try_combine): Use simplify_replace_rtx on individual
23432 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23433 of the whole CALL_INSN_FUNCTION_USAGE.
23435 PR tree-optimization/79972
23436 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23437 get_range_info on SSA_NAMEs. Formatting fixes.
23439 2017-03-10 Richard Biener <rguenther@suse.de>
23440 Jakub Jelinek <jakub@redhat.com>
23442 PR tree-optimization/77975
23443 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23444 edge to be constant.
23445 (get_val_for): For constant x return it. Formatting fix.
23446 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23447 would use the same bases as the current one.
23449 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23451 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23452 instead of vec_select for V1TImode.
23453 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23455 (VSX_LE_128): Add V1TI to this mode iterator.
23456 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23457 (*vsx_le_perm_store_<mode>): Likewise.
23458 (pre-reload splitter for VSX stores): Likewise.
23459 (post-reload splitter for VSX stores): Likewise.
23460 (*vsx_xxpermdi2_le_<mode>): Likewise.
23461 (*vsx_lxvd2x2_le_<mode>): Likewise.
23462 (*vsx_stxvd2x2_le_<mode>): Likewise.
23464 2017-03-09 Michael Eager <eager@eagercon.com>
23466 Correct failures with --enable-checking=yes,rtl.
23468 * config/microblaze/microblaze.c (microblaze_expand_shift):
23469 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23470 test for const0_rtx.
23471 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23472 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23474 2017-03-09 Richard Biener <rguenther@suse.de>
23476 PR tree-optimization/79977
23477 * graphite-scop-detection.c (scop_detection::merge_sese):
23478 Handle the case of extra exits to blocks dominating the entry.
23480 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23482 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23485 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23487 PR rtl-optimization/79949
23488 * lra-constraints.c (process_alt_operands): Check memory when
23489 trying to predict a cycle. Print about the overall increase.
23491 2017-03-09 Richard Biener <rguenther@suse.de>
23493 PR middle-end/79971
23494 * gimple-expr.c (useless_type_conversion_p): Preserve
23495 TYPE_SATURATING for fixed-point types.
23497 2017-03-09 Richard Biener <rguenther@suse.de>
23500 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23501 alignment of BLKmode params.
23503 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23506 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23507 (VALL_NO_V2Q): Likewise.
23509 * config/aarch64/aarch64-simd.md
23510 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23512 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23513 VALL_NO_V2Q mode iterator.
23514 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23516 2017-03-09 Martin Liska <mliska@suse.cz>
23518 PR tree-optimization/79631
23519 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23520 tree_int_cst_sign_bit just for INTEGER constants.
23522 2017-03-09 Martin Liska <mliska@suse.cz>
23526 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23529 2017-03-09 Marek Polacek <polacek@redhat.com>
23532 * tree.c (inchash::add_expr): Handle TREE_VEC.
23534 2017-03-09 Martin Liska <mliska@suse.cz>
23537 (chkp_narrow_size_and_offset): New function.
23538 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23539 (void chkp_parse_bit_field_ref): New function.
23540 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23541 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23543 2017-03-09 Martin Liska <mliska@suse.cz>
23546 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23547 (chkp_find_bounds_1): Remove gcc_unreachable.
23549 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23552 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23553 BUILT_IN_SYNC*, determine the access type from the size suffix and
23554 always build a MEM_REF with that type. Handle forgotten
23555 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23558 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23559 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23560 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23561 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23562 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23563 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23564 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23565 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23566 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23567 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23568 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23569 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23570 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23571 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23572 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23573 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23574 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23575 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23576 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23577 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23578 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23579 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23580 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23581 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23582 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23583 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23584 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23585 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23586 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23587 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23588 definitions outside of __OPTIMIZE__ guarded section.
23591 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23592 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23593 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23594 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23597 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23599 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23600 ("vfenez<mode>"): Add missing constraints.
23602 2017-03-08 Martin Sebor <msebor@redhat.com>
23605 * config/nds32/nds32.c (nds32_option_override):
23606 Fix misspelled diagnostic.
23608 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23611 * gimplify.c (gimplify_omp_for): Replace index var in outer
23612 taskloop statement with an artificial variable and add
23613 OMP_CLAUSE_PRIVATE clause for it.
23615 2017-03-08 Richard Biener <rguenther@suse.de>
23617 PR tree-optimization/79955
23618 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23619 for accesses that are completely outside of the variable.
23621 2017-03-08 Andrew Haley <aph@redhat.com>
23623 PR tree-optimization/79943
23624 * tree-ssa-loop-split.c (compute_new_first_bound): When
23625 calculating the new upper bound, (END-BEG) should be added, not
23628 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23630 * config/avr/avr.md (setmemhi): Make sure match_dup
23631 operand number comes before match_scratch.
23633 2017-03-08 Richard Biener <rguenther@suse.de>
23635 PR tree-optimization/79920
23636 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23637 with ncopies == 1 to ...
23638 (vect_transform_slp_perm_load): ... here. Properly compute
23639 all element loads by iterating VF times over the group. Do
23640 not handle ncopies (computed in a broken way) in
23641 vect_create_mask_and_perm.
23643 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23646 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23647 is a uniform vector, use uniform_vector_p return value instead of
23648 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23650 2017-03-07 Marek Polacek <polacek@redhat.com>
23652 PR middle-end/79809
23653 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23654 (alloca_call_type): Likewise.
23656 2017-03-07 Martin Liska <mliska@suse.cz>
23658 * gcov.c (process_args): Put comment to correct location.
23660 2017-03-07 Martin Liska <mliska@suse.cz>
23662 PR middle-end/68270
23663 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23664 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23665 (chkp_narrow_bounds_for_field): Likewise.
23666 (chkp_parse_array_and_component_ref): Pass one more argument to
23669 2017-03-07 Richard Biener <rguenther@suse.de>
23671 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23674 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23676 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23677 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23679 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23682 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23683 to end of description.
23684 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23686 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23688 PR rtl-optimization/79901
23689 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23691 (*avx512f_<code><mode>3<mask_name>): ... this.
23692 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23693 iterator instead of VI8_AVX2_AVX512BW.
23695 PR rtl-optimization/79901
23696 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23697 min/max expander, expand it using expand_vec_cond_expr.
23700 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23703 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23706 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23707 to void * for PCH reasons.
23708 * dwarf2out.c (output_loc_operands, output_die): Cast
23709 v.val_vec.array to unsigned char *.
23711 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23714 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23717 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23719 PR rtl-optimization/79571
23720 * lra-constraints.c (process_alt_operands): Calculate static
23721 reject and subtract it from overall when only addresses will be
23724 2017-03-06 Julia Koval <julia.koval@intel.com>
23727 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23728 incoming stack boundary to 128 for 64-bit targets.
23730 2017-03-06 Richard Biener <rguenther@suse.de>
23732 PR tree-optimization/79894
23733 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23734 to NULL after folding it.
23736 2017-03-06 Richard Biener <rguenther@suse.de>
23738 PR tree-optimization/79824
23739 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23740 check disabling peeling for gaps.
23742 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23744 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23745 attributes): Document gettimeofday.
23747 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23749 * config/s390/s390.c (s390_option_override_internal): Set
23750 PARAM_MIN_VECT_LOOP_BOUND
23752 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23754 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23755 * config/s390/s390.md: Likewise.
23757 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23760 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23761 (<avx2_avx512>_perm<mode>): Rename to ...
23762 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23764 (<avx512>_perm<mode>_mask): Rename to ...
23765 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23767 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23768 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23769 instead of VI8F_256_512.
23770 (avx512f_perm<mode>): New define_expand.
23771 (avx512f_perm<mode>_mask): Likewise.
23772 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23773 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23775 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23777 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23778 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23780 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23782 2017-03-06 Martin Liska <mliska@suse.cz>
23785 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23786 when having a SSA NAME w/o VAR_DECL assigned to it.
23788 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23790 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23791 msa_dpsub_<su>_d): Fix MODE for vec_select.
23793 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23795 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23797 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23799 2017-03-06 Richard Biener <rguenther@suse.de>
23801 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23802 * plugin.c (register_plugin_info): Likewise.
23803 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23805 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23807 * config/i386/sse.md (sse_storehps, sse_storelps,
23808 avx_<castmode><avxsizesuffix>_<castmode>,
23809 avx512f_<castmode><avxsizesuffix>_<castmode>,
23810 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23811 in condition that at least one operand is not a MEM.
23813 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23815 PR middle-end/79805
23816 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23817 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23819 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23820 gimple_call_nothrow_p flag based on whether original builtin can throw.
23821 If it can, emit following stmts on the fallthrough edge.
23822 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23823 don't create new bb if inserting just debug stmts on the edge, try to
23824 insert them on the fallthru bb or just reset debug stmts.
23826 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23829 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23830 restore recog_data (including the operand rtxes inside it) around
23831 the call to get_insn_template.
23833 2017-03-03 Martin Sebor <msebor@redhat.com>
23835 PR tree-optimization/79699
23836 * context.c (context::~context): Free MPFR caches to avoid
23837 a memory leak on program exit.
23839 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23841 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
23842 Use wide_int::ulow () instead of .elt (0).
23844 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23846 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
23847 (*pushxf): Limit oF constraint to 32bit targets and add oC
23848 constraint for 64bit targets.
23849 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
23850 (*pushdf): Change rmF constraint to rmC.
23852 2017-03-03 Martin Liska <mliska@suse.cz>
23854 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
23855 Remove unused variable.
23857 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23860 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
23861 is a memory operand, increase num_memory.
23862 (ix86_expand_args_builtin): Likewise.
23864 2017-03-03 Jan Hubicka <jh@suse.cz>
23867 * ipa-devirt.c (maybe_record_node): Properly handle
23868 __cxa_pure_virtual visibility.
23870 2017-03-03 Martin Liska <mliska@suse.cz>
23872 PR tree-optimization/79803
23873 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
23875 (pass_loop_prefetch::execute): Disabled optimization if an
23876 assumption about L1 cache size is not met.
23878 2017-03-03 Martin Liska <mliska@suse.cz>
23880 PR rtl-optimization/79574
23881 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
23882 (hash_scan_set): Likewise.
23883 (dump_hash_table): Likewise.
23884 (hoist_code): Likewise.
23886 2017-03-03 Richard Biener <rguenther@suse.de>
23888 * fixed-value.c (fixed_from_string): Restore use of elt (1)
23889 in place of uhigh ().
23890 (fixed_convert_from_real): Likewise.
23892 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
23895 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
23897 2017-03-03 Richard Biener <rguenther@suse.de>
23899 PR middle-end/79818
23900 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
23901 TYPE_OVERFLOW_UNDEFINED check.
23903 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23905 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
23907 (vector_ae_<mode>_p): Likewise.
23908 (vector_nez_<mode>_p): Likewise.
23909 (vector_ne_v2di_p): Likewise.
23910 (vector_ae_v2di_p): Likewise.
23911 (vector_ne_<mode>_p): Likewise.
23912 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
23914 (vsx_tsqrt<mode>2_fe): Likewise.
23916 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
23919 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
23921 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23923 PR rtl-optimization/79780
23924 * cprop.c (one_cprop_pass): When second and further conditional trap
23925 in a single basic block is turned into an unconditional trap, turn it
23926 into a deleted note to avoid RTL verification failures.
23928 2017-03-02 Richard Biener <rguenther@suse.de>
23930 * fold-const.c (const_binop): Use ulow () instead of elt (0).
23932 2017-03-02 Richard Biener <rguenther@suse.de>
23934 PR tree-optimization/79345
23936 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
23937 param and abort the walk, returning -1 if it is hit.
23938 (walk_aliased_vdefs): Take a limit param and pass it on.
23939 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
23940 defaulting to 0 and return a signed int.
23941 * tree-ssa-uninit.c (struct check_defs_data): New struct.
23942 (check_defs): New helper.
23943 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
23944 about uninitialized memory.
23945 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
23946 bogus uninitialized warning.
23947 (fixed_convert_from_real): Likewise.
23949 2017-03-02 Bin Cheng <bin.cheng@arm.com>
23951 PR tree-optimization/66768
23952 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
23953 iv_use if base object can't be determined.
23955 2017-03-02 Jakub Jelinek <jakub@redhat.com>
23957 PR tree-optimization/79345
23958 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
23959 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
23960 (get_pattern_stats): Initialize it.
23961 * genemit.c (gen_expand): Verify match_scratch numbers come after
23962 match_operand/match_dup numbers.
23963 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
23964 match_scratch numbers.
23965 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
23967 * config/s390/s390.md (trunctdsd2): Likewise.
23969 2017-03-02 Richard Biener <rguenther@suse.de>
23971 * wide-int.h (wide_int_storage::operator=): Implement in terms
23974 2017-03-02 Richard Biener <rguenther@suse.de>
23976 PR tree-optimization/79777
23977 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
23978 the to insert expression to sth existing.
23980 2017-03-01 Martin Sebor <msebor@redhat.com>
23982 PR middle-end/79692
23983 * gimple-ssa-sprintf.c
23984 (directive::known_width_and_precision): New function.
23985 (format_integer): Use it.
23986 (get_mpfr_format_length): Consider the full range of precision
23987 when computing %g output with the # flag. Set the likely byte
23988 count to 3 rather than 1 when precision is indeterminate.
23989 (format_floating): Correct the lower bound of precision.
23991 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23993 * doc/invoke.texi: Document default code model for 64-bit Linux.
23995 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23998 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
23999 udiv rather than div since input pattern is unsigned.
24001 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24003 * config/i386/i386.c (print_reg): Warn for values of
24004 unsupported size in integer register.
24006 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24009 * config/rs6000/predicates.md (current_file_function_operand): Do
24010 not allow self calls to be local if the function is replaceable.
24012 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24015 * config/rs6000/altivec.h (vec_ctz and others): Change the
24016 preprocessor macro that controls conditional compilation from
24017 _ARCH_PWR9 to __POWER9_VECTOR__.
24018 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24019 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24020 control (instead of _ARCH_PWR9 control) so that template
24021 definition uses power9-specific function.
24022 (vec_any_eq): Likewise.
24023 (vec_all_ne): Change macro definition to use a power9-specific
24024 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24025 _ARCH_PWR9 control).
24026 (vec_any_eq) Likewise.
24027 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24028 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24029 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24030 support for xvcmpnedp instruction.
24031 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24032 macro expansion so that Power9 implementation of vec_all_ne does
24033 not use the AltiVec predicate framework.
24034 (VCMPNEH_P): Likewise.
24035 (VCMPNEW_P): Likewise.
24036 (VCMPNED_P): Likewise.
24037 (VCMPNEFP_P): Likewise.
24038 (VCMPNEDP_P): Likewise.
24039 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24040 implementation of vec_any_eq to not use AltiVec predicate
24042 (VCMPAEH_P): Likewise.
24043 (VCMPAEW_P): Likewise.
24044 (VCMPAED_P): Likewise.
24045 (VCMPAEFP_P): Likewise.
24046 (VCMPAEDP_P): Likewise.
24047 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24048 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24049 not use the AltiVec predicate framework.
24050 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24051 of vec_any_eq to not use AltiVec predicate framework.
24052 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24053 support for predefined __POWER9_VECTOR__ macro to indicate that
24054 Power9 instruction selection is enabled.
24055 (altivec_overloaded_builtins): Remove extraneous
24056 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24057 function argument types RS6000_BTI_bool_V16QI and
24058 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24059 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24060 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24061 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24062 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24063 Power9 for implementations of vec_cmpne. Change the signature for
24064 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24065 (representing vec_all_ne) to remove the previously described first
24066 argument of type RS6000_BTI_INTSI, as this was an artifact of
24067 reliance on the AltiVec predicate framework, which is no longer
24068 used in the implementation of these functions. Add
24069 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24070 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24071 since, unlike the AltiVec predicate framework implementation, we
24072 do not share function descriptors between vec_alle and vec_anyeq.
24073 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24074 set of modes that receive special treatment even when
24075 TARGET_P9_VECTOR is true. The special treatment emits code that
24076 does not depend on Power9 instructions.
24077 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24078 define_expand to not rely on AltiVec predicate framework.
24079 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24081 (vector_ne_v2di_p): Change this define_expand to not rely on
24082 AltiVec predicate framework.
24083 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24085 (vector_ne_<mode>_p): Change this define_expand to not rely on
24086 AltiVec predicate framework.
24087 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24089 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24090 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24091 define_insn pattern.
24092 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24093 define_insn pattern because the xvcmpne<VSs>. instruction is not
24095 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24096 instruction is not supported.
24098 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24100 * config/nvptx/nvptx.c: Include intl.h.
24102 2017-03-01 Martin Jambor <mjambor@suse.cz>
24105 * ipa-prop.h (ipa_bits): Removed field known.
24106 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24107 to pointers. Adjusted their comments to warn about their sharing.
24108 (ipcp_transformation_summary): Change bits to a vector of pointers.
24109 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24110 (ipa_get_ipa_bits_for_value): Declare.
24111 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24112 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24113 (ipa_bits_hash_table): Likewise.
24114 (ipa_vr_ggc_hash_traits): Likewise.
24115 (ipa_vr_hash_table): Likewise.
24116 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24117 being pointers and vr_known being removed.
24118 (ipa_set_jf_unknown): Likewise.
24119 (ipa_get_ipa_bits_for_value): New function.
24120 (ipa_set_jfunc_bits): Likewise.
24121 (ipa_get_value_range): New overloaded functions.
24122 (ipa_set_jfunc_vr): Likewise.
24123 (ipa_compute_jump_functions_for_edge): Use the above functions to
24124 construct bits and vr parts of jump functions.
24125 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24126 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24128 (ipcp_grow_transformations_if_necessary): Also allocate
24129 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24131 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24132 them. Fix too long lines.
24133 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24134 vr_known being removed.
24135 (ipa_read_jump_function): Use new setter functions to construct bits
24136 and vr parts of jump functions or set them to NULL.
24137 (write_ipcp_transformation_info): Adjust for bits being pointers.
24138 (read_ipcp_transformation_info): Likewise.
24139 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24141 Include gt-ipa-prop.h.
24142 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24144 (ipcp_store_bits_results): Likewise.
24145 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24146 Do not write to existing jump functions but use a temporary instead.
24148 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24151 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24152 attempt to use its first operand as BIT_FIELD_REF base.
24154 2017-03-01 Richard Biener <rguenther@suse.de>
24156 PR middle-end/79721
24157 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24158 interpolating formula in wrapping arithmetic.
24159 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24161 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24163 PR tree-optimization/79734
24164 * tree-vect-generic.c (expand_vector_condition): Optimize
24165 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24166 Handle VEC_COND_EXPR where comparison has different inner width from
24167 type's inner width.
24169 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24171 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24172 markup, and similar issues. Remove @opindex entries for things
24173 that aren't options. Add missing -mmpy-option entries.
24175 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24177 PR tree-optimization/79737
24178 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24179 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24180 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24181 instead of byte_size. Formatting fix.
24182 (shift_bytes_in_array_right): Formatting fix.
24184 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24187 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24188 condition on optimize for the leaf function test.
24190 2017-02-28 Martin Liska <mliska@suse.cz>
24193 * read-rtl-function.c (function_reader::handle_unknown_directive):
24194 Bail out when one uses -flto.
24196 2017-02-28 Martin Liska <mliska@suse.cz>
24198 * common.opt: Replace space with tabular for options of <number>
24200 * config/i386/i386.opt: Show <number> value for
24201 -mlarge-data-threshold.
24202 * opts.c (print_filtered_help): Do not display number in hexadecimal
24205 2017-02-28 Martin Liska <mliska@suse.cz>
24207 * common.opt: Fix --help=option -Q for options which are of
24210 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24212 * config/i386/i386.c (print_reg): Error out for values
24213 of 8-bit size in invalid integer register.
24215 2017-02-28 Martin Sebor <msebor@redhat.com>
24217 PR tree-optimization/79691
24218 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24220 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24223 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24224 gcc_unreachable with output_operand_lossage.
24226 2017-02-28 Richard Biener <rguenther@suse.de>
24228 PR tree-optimization/79740
24229 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24231 (visit_nary_op): Insert the nary into the hashtable if we
24232 pattern-matched sth.
24233 * tree-ssa-pre.c (eliminate_insert): Robustify.
24235 2017-02-28 Richard Biener <rguenther@suse.de>
24237 PR middle-end/79731
24238 * fold-const.c (decode_field_reference): Reject out-of-bound
24241 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24243 * config/i386/i386.c: Include intl.h.
24244 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24245 instead of just cond ? "..." : "...".
24246 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24247 * coverage.c (read_counts_file): Likewise.
24248 * omp-offload.c: Include intl.h.
24249 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24250 of just cond ? "..." : "...".
24251 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24252 of just cond ? "..." : "...".
24254 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24257 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24259 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24260 'tune for' CPU name.
24261 * config/arm/arm-cpu-data.h: Regenerated.
24263 2017-02-28 Richard Biener <rguenther@suse.de>
24265 PR tree-optimization/79732
24266 * tree-inline.c (expand_call_inline): Do not shadow var.
24268 2017-02-28 Richard Biener <rguenther@suse.de>
24270 PR tree-optimization/79723
24271 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24272 address-space properly.
24274 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24276 * doc/optinfo.texi (Optimization groups): Fix option used for
24278 * doc/invoke.texi (-fopt-info): Document "omp".
24279 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24280 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24281 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24282 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24283 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24285 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24287 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24288 OPTGROUP_OMP with "omp".
24290 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24293 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24294 for arithmetic shift of unsigned V2DI.
24296 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24298 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24299 arc/linux.h headers.
24300 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24301 (LINK_SPEC): Likewise.
24302 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24303 (EXTRA_SPECS): Likewise.
24304 (STARTFILE_SPEC): Likewise.
24305 (ENDFILE_SPEC): Likewise.
24306 (LIB_SPEC): Likewise.
24307 (TARGET_SDATA_DEFAULT): Likewise.
24308 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24309 (MULTILIB_DEFAULTS): Likewise.
24310 (DWARF2_UNWIND_INFO): Likewise.
24311 * config/arc/big.h: New file.
24312 * config/arc/elf.h: Likewise.
24313 * config/arc/linux.h: Likewise.
24314 * config/arc/t-uClibc: Remove.
24316 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24318 PR tree-optimization/77536
24319 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24320 (tree_transform_and_unroll_loop): Use above function to compute the
24321 estimated niter of unrolled loop and use it when scaling profile.
24322 Also use count info rather than frequency if it's non-zero.
24323 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24324 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24325 (vect_transform_loop): Call above function.
24327 2017-02-27 Richard Biener <rguenther@suse.de>
24329 PR tree-optimization/45397
24330 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24331 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24332 (visit_nary_op): Add pattern matching for CSEing sign-changed
24333 or truncated operations with wider ones.
24335 2017-02-27 Richard Biener <rguenther@suse.de>
24337 PR tree-optimization/79690
24338 * tree-vect-stmts.c (vectorizable_store): Use vector type
24339 built from the DR with address-space.
24341 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24343 * doc/invoke.texi (Optimize Options): Refine the description
24344 of asan-use-after-return.
24346 2017-02-25 Alan Modra <amodra@gmail.com>
24348 PR rtl-optimization/79584
24349 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24350 base, not ad->base_term, the reg within base. Remove assertion
24351 that ad->base == ad->base_term. Replace gen_int_mode using
24352 bogus mode with const0_rtx.
24354 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24356 PR middle-end/79396
24357 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24358 FMA_EXPR like tcc_binary or tcc_unary.
24360 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24363 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24365 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24366 (output_loc_operands): Handle DW_OP_call_ref and
24367 DW_OP_GNU_variable_value.
24368 (struct variable_value_struct): New type.
24369 (struct variable_value_hasher): Likewise.
24370 (variable_value_hash): New variable.
24371 (string_types): Remove.
24372 (copy_loc_descr): New function.
24373 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24374 (prepend_loc_descr_to_each): New function.
24375 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24376 instead of add_loc_descr_to_each if the first argument is single
24377 location list and the second has multiple.
24378 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24379 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24380 when looking for variable value which doesn't have other location info.
24381 (loc_list_from_tree): Formatting fix.
24382 (gen_array_type_die): Simplify DW_AT_string_length handling.
24383 (adjust_string_types): Remove.
24384 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24385 string_types. Call resolve_variable_values.
24386 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24387 (resolve_addr_in_expr): Likewise. Add A argument.
24388 (copy_deref_exprloc): Remove deref argument. Adjust for the
24389 original expression being DW_OP_GNU_variable_value with optionally
24390 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24391 optionally after it.
24392 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24393 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24394 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24395 (variable_value_hasher::hash, variable_value_hasher::equal): New
24397 (resolve_variable_value_in_expr, resolve_variable_value,
24398 resolve_variable_values, note_variable_value_in_expr,
24399 note_variable_value): New functions.
24400 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24403 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24406 * opts.h (handle_generated_option): Add GENERATED_P argument.
24407 * opts-common.c (handle_option): Adjust function comment.
24408 (handle_generated_option): Add GENERATED_P argument, pass it to
24410 (control_warning_option): Pass false to handle_generated_option
24412 * opts.c (maybe_default_option): Pass true to handle_generated_option
24414 * optc-gen.awk: Likewise.
24416 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24418 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24419 a REG, look at the REG it is a SUBREG of.
24420 (splitter for cmpeqsi_t): Ditto.
24422 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24424 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24425 the special USEs with the pattern of the insn, not the insn itself.
24427 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24430 * doc/invoke.texi: Document -mload-store-pairs.
24432 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24433 Sandra Loosemore <sandra@codesourcery.com>
24435 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24436 argument isn't a CONST_INT.
24437 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24438 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24439 (nios2_validate_compare): Bypass alternate compare logic if *op2
24440 is not a CONST_INT.
24441 (ldstwm_operation_p): Return false if first_base is not a REG or
24442 if first_offset is not a CONST_INT.
24444 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24446 * config/cris/cris.md: Use correct operand in a define_peephole2.
24448 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24450 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24452 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24454 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24455 this_insn if it is an INSN or JUMP_INSN.
24456 (force_offsettable): Look at base, not at addr.
24457 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24458 on things that aren't necessarily CONST_INTs.
24460 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24462 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24463 -mfpmath=sse is the default also for x86-32 targets with SSE2
24464 instruction set when @option{-ffast-math} is enabled
24466 2017-02-24 Jeff Law <law@redhat.com>
24468 PR rtl-optimizatoin/79286
24469 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24472 2017-02-24 Richard Biener <rguenther@suse.de>
24474 PR tree-optimization/79389
24475 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24478 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24480 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24481 function comment to reflect reality.
24482 (loop_exits_before_overflow): Fix typo in function description.
24484 2017-02-24 Richard Biener <rguenther@suse.de>
24486 PR tree-optimization/79389
24487 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24488 properly that a threading opportunity exists. Detect conditional
24489 copy/constant propagation opportunities.
24491 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24493 * config/visium/visium.md (type): Add trap.
24494 (b): New mode attribute.
24495 (*btst): Rename into...
24496 (*btst<mode>): ...this and adjust.
24497 (*cbranchsi4_btst_insn): Rename into...
24498 (*cbranch<mode>4_btst_insn): ...this and adjust.
24499 (trap): New define_insn.
24501 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24503 PR tree-optimization/79389
24504 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24505 (noce_reversed_cond_code): New function.
24506 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24507 reversed_comparison_code. Formatting fix.
24508 (noce_try_store_flag): Test rev_cond != NULL in addition to
24509 reversed_comparison_code.
24510 (noce_try_store_flag_constants): Likewise.
24511 (noce_try_store_flag_mask): Likewise.
24512 (noce_try_addcc): Use rev_cond if non-NULL instead of
24513 reversed_comparison_code.
24514 (noce_try_cmove_arith): Likewise. Formatting fixes.
24515 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24516 (noce_find_if_block): Initialize rev_cond.
24517 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24518 instead of false as last argument never attempt to reverse it
24521 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24523 PR tree-optimization/79663
24524 * tree-predcom.c (combine_chains): Process refs in reverse order
24525 only for ZERO length chains, and add explaining comment.
24527 2017-02-23 Jeff Law <law@redhat.com>
24529 PR tree-optimization/79578
24530 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24531 in call to operand_equal_p.
24533 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24536 * config/i386/cpuid.h: Fix another undefined behavior.
24538 2017-02-23 Richard Biener <rguenther@suse.de>
24540 PR tree-optimization/79683
24541 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24542 vector types for data-refs.
24544 2017-02-23 Martin Liska <mliska@suse.cz>
24546 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24548 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24550 PR middle-end/79665
24551 * internal-fn.c (get_range_pos_neg): Moved to ...
24552 * tree.c (get_range_pos_neg): ... here. No longer static.
24553 * tree.h (get_range_pos_neg): New prototype.
24554 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24555 are known to be in between 0 and signed maximum inclusive, try to
24556 expand both unsigned and signed divmod and use the cheaper one from
24559 2017-02-22 Jeff Law <law@redhat.com>
24561 PR tree-optimization/79578
24562 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24563 to compare base operands.
24565 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24568 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24569 gpc_reg_operand instead of fpr_reg_operand.
24571 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24573 * config/mips/mips.c (mips_return_in_memory): Force FP
24574 vector types to be returned in memory for o32 ABI.
24576 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24578 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24579 instead of DW_TAG_member for static data member declarations and don't
24580 set no_linkage_name for static inline data members.
24581 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24584 2017-02-22 Martin Liska <mliska@suse.cz>
24586 * doc/invoke.texi: Replace inequality signs with square brackets
24589 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24591 PR tree-optimization/68644
24592 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24594 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24597 * lra-constraints.c (simplify_operand_subreg): Handle
24598 WORD_REGISTER_OPERATIONS targets.
24600 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24603 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24604 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24605 elimination by swapping fld*.
24607 2017-02-22 Richard Biener <rguenther@suse.de>
24609 PR tree-optimization/79673
24610 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24611 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24612 irrelevant address-space qualifiers and avoiding a
24613 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24615 2017-02-22 Richard Biener <rguenther@suse.de>
24617 PR tree-optimization/79666
24618 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24619 to not symbolically negate if that may introduce undefined
24622 2017-02-22 Martin Liska <mliska@suse.cz>
24625 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24626 * data-streamer-out.c (streamer_write_gcov_count_stream):
24628 * value-prof.c (stream_out_histogram_value): Make assert more
24629 precise based on type of counter.
24631 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24634 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24635 Use nonimmediate_operand instead of memory_operand for operand 1.
24636 (float-extend standard_x87sse_constant_load splitter): Ditto.
24638 2017-02-21 Jeff Law <law@redhat.com>
24640 PR tree-optimization/79621
24641 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24642 blocks with edges to themselves.
24644 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24647 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24648 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24649 Use gimple_call_builtin_p.
24652 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24653 on temporarily removed DEBUG_INSNs.
24655 PR tree-optimization/79649
24656 * tree-loop-distribution.c (classify_partition): Give up on
24657 non-generic address space loads/stores.
24659 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24661 * doc/loop.texi (Loop manipulation): Remove nonexistent
24662 tree_ssa_loop_version from the documentation.
24663 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24665 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24668 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24669 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24670 * config/rs6000/rs6000.c: Include except.h.
24671 (rs6000_expand_split_stack_prologue): Call
24672 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24674 2017-02-21 Martin Jambor <mjambor@suse.cz>
24677 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24678 have been analyzed.
24680 2017-02-21 Martin Jambor <mjambor@suse.cz>
24682 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24683 for backward compatibility only.
24684 * doc/invoke.texi (Option Summary): Remove all references to
24685 -fipa-cp-alignment.
24687 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24691 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24693 * lra-constraints.c (curr_insn_transform): Handle
24694 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24696 2017-02-21 Martin Liska <mliska@suse.cz>
24698 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24700 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24702 PR translation/79638
24703 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24705 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24708 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24709 (arm_function_ok_for_sibcall): Return false for an indirect call by
24710 descriptor if all the argument registers are used.
24711 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24712 alignment of the function.
24714 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24716 PR tree-optimization/61441
24717 * simplify-rtx.c (simplify_const_unary_operation): For
24718 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24719 the sNaN unmodified.
24721 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24723 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24724 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24725 instead of SYSTEM_HEADER_DIR.
24727 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24728 Martin Liška <mliska@suse.cz>
24730 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24731 Fix typos and grammar, use active voice, and clarify.
24733 2017-02-20 Marek Polacek <polacek@redhat.com>
24735 PR middle-end/79537
24736 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24739 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24741 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24744 * config/i386/i386.c (ix86_expand_builtin): Handle
24745 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24746 ix86_builtins_isa[fcode].isa as a requirement of those
24747 flags and any other flag in the bitmask.
24748 (ix86_init_mmx_sse_builtins): Use 0 instead of
24749 ~OPTION_MASK_ISA_64BIT as mask.
24750 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24751 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24752 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24753 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24755 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24758 * lra-constraints.c (split_reg): Check requested split mode
24759 is supported by the register.
24761 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24763 * lra-constraints.c (simplify_operand_subreg): Remove early
24766 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24769 * lra-constraints.c (curr_insn_transform): Tighten condition
24770 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24772 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24775 * lra-constraints.c (curr_insn_transform): Handle
24776 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24778 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24781 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24783 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24785 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24788 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24791 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24793 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24796 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24798 * final.c (last_columnnum, override_columnnum): New variables.
24799 (final_start_function): Set last_columnnum, pass it to begin_prologue
24800 hook and pass 0 to dwarf2out_begin_prologue.
24801 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24802 to source_line debug hook.
24803 (notice_source_line): Compute last_columnnum and for debug_column_info
24804 return true on column changes.
24805 * debug.h (struct gcc_debug_hooks): Add column argument to
24806 source_line and begin_prologue hooks.
24807 (debug_nothing_int_charstar_int_bool): Remove prototype.
24808 (debug_nothing_int_int_charstar,
24809 debug_nothing_int_int_charstar_int_bool): New prototypes.
24810 (dwarf2out_begin_prologue): Add column argument.
24811 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24812 begin_prologue hooks.
24813 (debug_nothing_int_charstar_int_bool): Remove.
24814 (debug_nothing_int_int_charstar,
24815 debug_nothing_int_int_charstar_int_bool): New functions.
24816 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24817 through to dwarf2out_source_line.
24818 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24819 (dwarf2out_source_line): Add column argument, emit it if requested.
24820 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24822 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24823 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24824 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24825 through to dwarf2out_begin_prologue.
24826 (vmsdbgout_source_line): Add column argument, pass it through to
24827 dwarf2out_source_line.
24828 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24829 dbxout_source_line caller.
24830 (dbxout_source_line): Add column argument.
24832 * common.opt (gno-column-info, gcolumn-info): New options.
24833 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
24834 (check_die): Also test for multiple DW_AT_decl_column attributes.
24835 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
24836 DW_AT_decl_column if requested.
24837 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
24839 (gen_variable_die): Likewise.
24840 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
24841 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
24844 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
24845 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
24846 (ix86_handle_option): Handle OPT_m3dnowa.
24847 * doc/invoke.texi (-m3dnowa): Document.
24848 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
24849 -m3dnowa instead of -m3dnow -march=athlon.
24852 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
24853 instead of gcc_assert for K, r and R code checks. Formatting fixes.
24855 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24858 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
24859 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
24860 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
24861 generator for vsx_xxpermdi_<mode>_be.
24862 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
24863 force big-endian semantics.
24864 (vsx_xxpermdi_<mode>_be): New define_expand with same
24865 implementation as previous version of vsx_xxpermdi_<mode>.
24867 2017-02-17 Jakub Jelinek <jakub@redhat.com>
24869 PR tree-optimization/79327
24870 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
24871 variable, its initialization and use.
24873 2017-02-17 Julia Koval <julia.koval@intel.com>
24875 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
24876 (OPTION_MASK_ISA_PKU_UNSET): New.
24877 (ix86_handle_option): Handle -mrdpid.
24878 * config/i386/cpuid.h (bit_RDPID): New.
24879 * config/i386/driver-i386.c (host_detect_local_cpu):
24880 Detect RDPID feature.
24881 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
24882 * config/i386/i386-c.c (ix86_target_macros_internal):
24884 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
24885 (ix86_valid_target_attribute_inner_p): Add "rdpid".
24886 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24887 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
24888 * config/i386/i386.md (define_insn "rdpid"): New.
24889 * config/i386/i386.opt Add -mrdpid.
24890 * config/i386/immintrin.h (_rdpid_u32): New.
24892 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
24894 PR rtl-optimization/79541
24895 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
24896 instead of transforming it into USE.
24898 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
24900 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
24901 If HONOR_SNANS (SFmode) force the input to a register.
24902 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
24903 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
24904 an frsp or similar insn.
24906 2017-02-17 Martin Liska <mliska@suse.cz>
24908 PR rtl-optimization/79577
24909 * params.def (selsched-max-sched-times): Increase minimum to 1.
24911 2017-02-17 Martin Liska <mliska@suse.cz>
24913 PR rtl-optimization/79574
24914 * gcse.c (want_to_gcse_p): Prevent integer overflow.
24916 2017-02-17 Martin Liska <mliska@suse.cz>
24918 PR tree-optimization/79529
24919 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
24920 ssa_defined_default_def_p to handle cases which are implicitly
24922 * tree-ssa.c (ssa_defined_default_def_p): New function.
24923 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
24924 which are implicitly defined.
24925 * tree-ssa.h (ssa_defined_default_def_p): Declare.
24927 2017-02-17 Richard Biener <rguenther@suse.de>
24929 PR middle-end/79576
24930 * params.def (max-ssa-name-query-depth): Limit to 10.
24932 2017-02-17 Richard Biener <rguenther@suse.de>
24934 PR tree-optimization/79552
24935 * tree-ssa-structalias.c (visit_loadstore): Properly verify
24938 2017-02-17 Richard Biener <rguenther@suse.de>
24941 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
24943 2017-02-17 Marek Polacek <polacek@redhat.com>
24945 PR middle-end/79536
24946 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
24947 (fold_negate_expr): New wrapper.
24949 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
24951 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
24952 Correct terminology and de-emphasize pre-standard behavior.
24954 2017-02-16 Alan Modra <amodra@gmail.com>
24956 PR rtl-optimization/79286
24957 * ira.c (def_dominates_uses): New function.
24958 (update_equiv_regs): Don't create an equivalence for insns that
24959 may trap where the register def does not dominate the use.
24961 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
24963 PR rtl-optimization/78127
24964 * lra.c (lra): Call lra_eliminate before finish the loop after
24967 2017-02-16 Richard Biener <rguenther@suse.de>
24969 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
24971 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
24972 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
24973 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
24974 (isl_val_int_from_wi): New function.
24975 (extract_affine_gmp): Rename to ...
24976 (extract_affine_wi): ... this, take a widest_int.
24977 (extract_affine_int): Just wrap extract_affine_wi.
24978 (add_param_constraints): Use isl_val_int_from_wi.
24979 (add_loop_constraints): Likewise, and extract_affine_wi.
24981 2017-02-15 Jeff Law <law@redhat.com>
24983 PR middle-end/79521
24984 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
24985 ira_init_register_move_cost_if_necessary.
24987 2017-02-15 Martin Sebor <msebor@redhat.com>
24989 PR middle-end/32003
24990 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
24991 removed in a prior commit.
24993 2017-02-15 Bin Cheng <bin.cheng@arm.com>
24995 PR tree-optimization/79347
24996 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
24997 counters during peeling.
24999 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25001 * Makefile.in (site.exp): Remove "set ISLVER".
25003 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25006 * real.c (real_from_integer): Call real_convert even for decimal.
25008 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25011 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25013 2017-02-14 Andrew Pinski <apinski@cavium.com>
25015 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25016 cores and change the partno/implementer to be correct.
25017 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25018 the 'B" as the implementer.
25019 * config/aarch64/aarch64-tune.md: Regenerate.
25021 2017-02-14 Carl Love <cel@us.ibm.com>
25023 * config/rs6000/rs6000.c: Add case statement entry to make the
25024 xvcvuxdsp built-in argument unsigned.
25025 * config/rs6000/vsx.md: Fix the source and return operand types so they
25026 match the instruction definitions from the ISA document. Fix typo
25027 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25030 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25033 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25034 member early_clobber_alts.
25035 * lra-lives.c (reg_early_clobber_p): New.
25036 (process_bb_lives): Use it.
25037 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25038 (debug_operand_data): Initialize early_clobber_alts.
25039 (setup_operand_alternative): Set up early_clobber_alts.
25040 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25041 alternatives to new_insn_reg.
25042 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25044 (lra_update_insn_regno_info): Pass the new arg.
25046 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25048 PR middle-end/79505
25049 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25050 (new_oacc_loop_raw): Don't clear already cleared fields.
25053 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25054 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25055 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25057 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25060 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25062 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25065 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25066 the extra instruction to the right place to store 128-bit constant
25069 2017-02-14 Martin Sebor <msebor@redhat.com>
25071 PR middle-end/79448
25072 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25073 warning for strings of unknown length.
25075 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25077 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25079 2017-02-14 Jeff Law <law@redhat.com>
25082 * ira-costs.c (scan_one_insn): Initialize register move costs
25083 for pseudos seen in USE/CLOBBER insns.
25085 PR tree-optimization/79095
25086 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25087 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25088 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25089 if the operands are known to be not equal, then the resulting range
25091 (intersect_ranges): If the new range is ~[0,0] and the old range is
25092 wide, then prefer ~[0,0].
25093 * tree-vrp.c (overflow_comparison_p_1): New function.
25094 (overflow_comparison_p): New function.
25095 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25096 if NAME is used in an overflow test.
25097 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25098 overflow check that can be expressed as an equality test, then adjust
25099 ops to be that equality test.
25101 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25103 * config/s390/s390-builtin-types.def: Remove flags argument.
25104 * config/s390/s390.c (s390_init_builtins): Likewise.
25106 2017-02-14 Martin Liska <mliska@suse.cz>
25108 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25109 vector. Fix trailing white spaces.
25111 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25113 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25116 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25118 PR rtl-optimization/68664
25119 * config/arm/arm.c (arm_sched_can_speculate_insn):
25120 New function. Declare prototype.
25121 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25123 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25125 PR rtl-optimization/68664
25126 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25128 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25130 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25132 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25133 max skip bytes for function, loop and jump.
25135 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25137 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25138 ABS_EXPR for gimple dump.
25140 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25143 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25145 PR tree-optimization/79408
25146 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25147 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25148 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25149 also if rhs1 is INTEGER_CST.
25151 2017-02-14 Richard Biener <rguenther@suse.de>
25153 PR middle-end/79432
25154 * tree-into-ssa.c (insert_phi_nodes): When the function can
25155 have abnormal edges rewrite SSA names with broken use-def
25156 dominance out of SSA and register them for PHI insertion.
25158 2017-02-13 Martin Sebor <msebor@redhat.com>
25160 PR middle-end/79496
25161 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25162 clearing info.nowrite flag when snprintf size argument is a range.
25164 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25166 * cprop.c (cprop_jump): Add missing space in string literal.
25167 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25168 (get_constraint_for_component_ref): Likewise.
25169 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25170 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25171 * lra-constraints.c (process_alt_operands): Likewise.
25172 * ipa-inline.c (inline_small_functions): Likewise.
25173 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25174 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25175 * trans-mem.c (diagnose_tm_1_op): Likewise.
25176 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25177 (grid_parallel_clauses_gridifiable): Likewise.
25179 * config/nvptx/mkoffload.c (process): Add space in between
25182 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25183 "MOD4_SSE_REGS" and "ALL_REGS".
25185 * spellcheck.c (test_data): Add , in between "foo" and "food".
25187 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25190 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25191 boundary crossing check and subsequent code generation agree.
25193 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25195 * config/aarch64/aarch64.c (has_memory_op): Delete.
25196 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25199 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25201 PR rtl-optimization/79388
25202 PR rtl-optimization/79450
25203 * combine.c (distribute_notes): When removing TEM_INSN for which
25204 corresponding dest has last value recorded, invalidate that last
25207 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25209 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25210 of explicit '@'. Add missing assembly comment marker on branch costs
25213 2017-02-13 Nathan Sidwell <nathan@acm.org>
25215 * gengtype-lex.l (<in_struct>): Add '/'.
25217 2017-02-13 Martin Liska <mliska@suse.cz>
25220 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25222 2017-02-13 Richard Biener <rguenther@suse.de>
25224 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25226 * configure: Re-generate.
25227 * config.in: Likewise.
25228 * graphite-dependences.c: Simplify as if
25229 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25230 * graphite-isl-ast-to-gimple.c: Likewise.
25231 * graphite-optimize-isl.c: Likewise.
25232 * graphite-poly.c: Likewise.
25233 * graphite-sese-to-poly.c: Likewise.
25234 * graphite.h: Likewise.
25235 * toplev.c: Include isl/version.h and use isl_version () for
25236 printing the ISL version.
25237 * doc/install.texi: Update ISL requirement.
25239 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25241 * doc/standards.texi (Standards): Update reference to
25244 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25246 * doc/extend.texi (Named Address Spaces): sourceware.org now
25248 * doc/install.texi (Binaries): Ditto.
25251 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25253 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25254 standard terminology "stringize"/"stringizing" throughout.
25255 * doc/cppinternals.texi: Likewise.
25257 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25259 * doc/extend.texi: Fix some spelling mistakes and typos.
25260 * doc/invoke.texi: Likewise.
25262 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25265 * params.def (inline-min-speedup) Change from 10 to 8.
25267 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25269 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25272 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25275 * ipa-inline-analysis.c (get_minimal_bb): New function.
25276 (record_modified): Use it.
25277 (remap_edge_change_prob): Handle also ancestor functions.
25279 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25281 * doc/contrib.texi (Contributors): Remove broken link into
25282 the Mauve CVS repository.
25284 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25286 PR middle-end/79454
25287 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25288 result computation whenever lhs doesn't have vector mode, not
25289 just when it has BLKmode.
25291 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25293 * doc/makefile.texi (profiledbootstrap): Refer to the
25294 installation instructions only in textual form.
25296 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25299 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25301 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25303 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25304 (Specific): Update mingw-w64 reference.
25306 (Specific): Remove broken link to Renesas RX processor.
25308 2017-02-10 Richard Biener <rguenther@suse.de>
25310 * toplev.c (process_options): Do not mention obsolete graphite
25311 options when printing sorry message about missing graphite support.
25312 Mention -floop-nest-optimize.
25314 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25316 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25317 (vtst_p16): Likewise.
25318 (vtstq_p8): Likewise.
25319 (vtstq_p16): Likewise.
25321 (vtstq_p64): Likewise.
25322 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25323 (vset_lane_p64): New.
25324 (vsetq_lane_p64): New.
25326 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25328 PR tree-optimization/79411
25329 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25330 stmt operands are SSA_NAMEs used in abnormal phis.
25331 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25334 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25337 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25340 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25342 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25344 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25346 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25350 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25351 not arbitrary TREE_CONSTANT.
25354 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25355 "omp declare target link" attribute unless is_global_var.
25356 * omp-offload.c (find_link_var_op): Likewise.
25358 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25359 Chung-Lin Tang <cltang@codesourcery.com>
25361 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25363 (gimplify_adjust_omp_clauses): Don't delete TILE.
25364 (gimplify_omp_for): Deal with TILE.
25365 * internal-fn.c (expand_GOACC_TILE): New function.
25366 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25368 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25369 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25371 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25372 avoid DIV for outermost collapse var.
25373 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25374 Remove out of date comments, fix whitespace.
25375 * omp-general.c (omp_extract_for_data): Deal with tiling.
25376 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25377 adjust OLF_DIM_BASE value.
25378 (struct omp_for_data): Add tiling field.
25379 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25380 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25381 for auto loops. Remove default auto determining, moved to
25382 oacc_loop_fixed_partitions.
25383 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25384 stmts, add e_mask field.
25385 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25386 (oacc_thread_numbers): Use oacc_dim_call.
25387 (oacc_xform_tile): New.
25388 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25389 (finish_oacc_loop): Adjust for ifns vector.
25390 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25391 add case for GOACC_TILE fns.
25392 (oacc_loop_xform_loop): Delete.
25393 (oacc_loop_process): Iterate over call list directly, and add
25394 handling for GOACC_TILE fns.
25395 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25397 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25398 vector partitioning to outer loops. Assign 2 partitions to loops
25399 when available. Add TILE handling.
25400 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25401 (execite_oacc_device_lower): Process GOACC_TILE fns,
25402 ignore unknown specs.
25403 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25404 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25405 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25407 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25409 * configure.ac (ACX_BUGURL): Update.
25410 * configure: Regenerate.
25412 2017-02-09 Richard Biener <rguenther@suse.de>
25414 PR tree-optimization/69823
25415 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25416 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25418 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25420 * config/arc/arc-c.def: Add __NPS400__ definition.
25421 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25422 (TARGET_NPS400): Define.
25424 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25426 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25428 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25429 pointer, arch_info.
25430 (arc_cpu_types): Fill the arch_info field with a pointer into the
25431 arc_arch_types table.
25432 (arc_selected_cpu): Declare.
25433 * config/arc/arc.c (arc_selected_cpu): Make global.
25434 (arc_selected_arch): Delete.
25435 (arc_base_cpu): Delete.
25436 (arc_override_options): Remove references to deleted variables,
25437 update access to arch information.
25438 (ARC_OPT): Update access to arch information.
25439 (ARC_OPTX): Likewise.
25440 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25441 (TARGET_ARC600): Update access to arch information.
25442 (TARGET_ARC601): Likewise.
25443 (TARGET_ARC700): Likewise.
25444 (TARGET_EM): Likewise.
25445 (TARGET_HS): Likewise.
25446 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25449 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25452 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25453 condition/operands for integer GE/LE/GEU/LEU operations.
25455 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25457 PR translation/79397
25458 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25461 2017-02-08 Martin Jambor <mjambor@suse.cz>
25464 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25465 whether allocation happened.
25466 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25467 nothing was allocated.
25469 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25471 PR tree-optimization/79408
25472 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25473 constant, but SSA_NAME with a known integer range, use the minimum
25474 of that range instead of op1 to determine if modulo can be replaced
25475 with its first operand.
25477 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25479 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25481 2017-02-08 Richard Biener <rguenther@suse.de>
25483 PR tree-optimization/71824
25484 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25485 Check all loops contained in the merged region.
25487 2017-02-07 Andrew Pinski <apinski@cavium.com>
25489 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25491 2017-02-07 Andrew Pinski <apinski@cavium.com>
25493 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25494 (thunderxt88): Likewise.
25495 (thunderxt81): Disable LSE and change v8.1 to v8.
25496 (thunderxt83): Likewise.
25498 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25499 Richard Biener <rguenther@suse.de>
25501 PR middle-end/79399
25502 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25503 type from int to size_t.
25504 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25506 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25508 PR rtl-optimization/79386
25509 * cprop.c (bypass_conditional_jumps): Initialize
25510 bypass_last_basic_block already before splitting bbs after
25511 unconditional traps...
25512 (bypass_conditional_jumps): ... rather than here.
25515 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25516 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25517 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25518 fix -masm=intel patterns.
25520 2017-02-07 Richard Biener <rguenther@suse.de>
25522 PR tree-optimization/79256
25523 PR middle-end/79278
25524 * builtins.c (get_object_alignment_2): Use min_align_of_type
25525 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25526 and ADJUST_FIELD_ALIGN.
25528 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25530 * doc/tm.texi: Regenerate.
25531 * stor-layout.c (layout_decl): Adjust.
25532 (update_alignment_for_field): Likewise.
25533 (place_field): Likewise.
25534 (min_align_of_type): Likewise.
25535 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25536 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25537 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25538 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25539 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25540 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25541 * config/i386/i386.c (x86_field_alignment): Likewise.
25542 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25543 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25544 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25545 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25546 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25547 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25551 2017-01-30 Richard Biener <rguenther@suse.de>
25553 PR tree-optimization/79256
25554 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25555 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25558 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25560 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25561 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25562 builtins to SImode and emit a zero-extend, if necessary.
25564 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25566 * docs/invoke.texi (RISC-V Options): Alphabetize.
25568 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25570 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25573 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25575 * config/riscv/riscv.c: New file.
25576 * common/config/riscv/riscv-common.c: Likewise.
25577 * config.gcc: Likewise.
25578 * config/riscv/constraints.md: Likewise.
25579 * config/riscv/elf.h: Likewise.
25580 * config/riscv/generic.md: Likewise.
25581 * config/riscv/linux.h: Likewise.
25582 * config/riscv/multilib-generator: Likewise.
25583 * config/riscv/peephole.md: Likewise.
25584 * config/riscv/pic.md: Likewise.
25585 * config/riscv/predicates.md: Likewise.
25586 * config/riscv/riscv-builtins.c: Likewise.
25587 * config/riscv/riscv-c.c: Likewise.
25588 * config/riscv/riscv-ftypes.def: Likewise.
25589 * config/riscv/riscv-modes.def: Likewise.
25590 * config/riscv/riscv-opts.h: Likewise.
25591 * config/riscv/riscv-protos.h: Likewise.
25592 * config/riscv/riscv.h: Likewise.
25593 * config/riscv/riscv.md: Likewise.
25594 * config/riscv/riscv.opt: Likewise.
25595 * config/riscv/sync.md: Likewise.
25596 * config/riscv/t-elf-multilib: Likewise.
25597 * config/riscv/t-linux: Likewise.
25598 * config/riscv/t-linux-multilib: Likewise.
25599 * config/riscv/t-riscv: Likewise.
25600 * configure.ac: Likewise.
25601 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25602 Waterman as RISC-V maintainers.
25603 * doc/install.texi: Add RISC-V entries.
25604 * doc/invoke.texi: Add RISC-V options section.
25605 * doc/md.texi: Add RISC-V constraints section.
25606 * configure: Regenerated.
25608 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25611 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25612 false values to be constant vectors with all 0 or all 1 bits set.
25613 (vcondu<mode><mode>): Likewise.
25614 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25616 (fpmask_comparison_operator): Update comment.
25617 (vecint_comparison_operator): New predicate.
25618 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25619 vector conditionals when the true and false values are constant
25620 vectors with all 0 bits or all 1 bits set.
25622 2017-02-06 Martin Sebor <msebor@redhat.com>
25624 PR tree-optimization/79376
25625 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25627 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25629 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25630 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25631 to simplify split condition.
25633 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25635 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25636 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25639 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25641 PR rtl-optimization/68664
25642 * target.def (can_speculate_insn): New hook.
25643 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25644 * doc/tm.texi: Regenerate.
25645 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25646 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25647 (rs6000_sched_can_speculate_insn): New function.
25649 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25651 PR tree-optimization/79284
25652 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25653 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25654 vectorizable_mask_load_store, vectorizable_operation,
25655 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25656 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25657 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25658 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25659 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25660 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25661 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25662 is_gimple_assign (stmt). Replace another such test with
25663 is_gimple_assign (stmt).
25665 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25668 * config/avr/avr.c (rtl-iter.h): Include it.
25669 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25670 (avr_legitimate_combined_insn): ...and implementation.
25672 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25674 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25675 * config/s390/s390.c (s390_const_operand_ok)
25676 (s390_canonicalize_comparison, s390_extract_part)
25677 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25678 (s390_contiguous_bitmask_p, s390_rtx_costs)
25679 (legitimize_pic_address): Likewise.
25680 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25681 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25682 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25683 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25684 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25686 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25688 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25689 REGNO($0) == REGNO($1).
25691 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25693 * config/s390/linux.h(SIZE_TYPE): Add comment.
25695 2017-02-06 Julian Brown <julian@codesourcery.com>
25696 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25697 Virendra Pathak <virendra.pathak@broadcom.com>
25699 * config/aarch64/aarch64-cores.def: Change the scheduler
25701 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25702 * config/aarch64/thunderx2t99.md: New file.
25704 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25706 * doc/standards.texi (Go Language): Update link to language
25709 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25711 * tree-eh.c (lower_resx): Sanitize profile.
25712 (cleanup_empty_eh_move_lp): Likewise.
25714 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25717 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25719 * cfgloopmanip.h (loop_version): Update prototype.
25720 * modulo-sched.c (sms_schedule): Update call of loop_version.
25721 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25722 * tree-parloops.c (gen_parallel_loop): Likewise.
25723 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25724 * tree-ssa-loop-split.c (split_loop): Likewise.
25725 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25726 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25728 2017-02-05 Martin Liska <mliska@suse.cz>
25731 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25733 (print_operand_address): Initialize a struct to zero.
25735 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25737 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25738 garbage collector only in textual form.
25740 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25742 * doc/extend.texi (x86 specific memory model extensions for
25743 transactional memory): Simplify a phrase.
25745 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25748 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25749 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25750 (atomic_storedi_1): Likewise.
25752 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25754 PR tree-optimization/79338
25755 * tree-parloops.c (gather_scalar_reductions): Don't call
25756 vect_analyze_loop_form for loop->inner before destroying loop's
25759 2017-02-03 Martin Sebor <msebor@redhat.com>
25761 PR tree-optimization/79327
25762 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25763 when precision has resulted in leading zeros.
25764 (format_integer): Adjust the likely counter to assume an unknown
25765 argument that may be zero is non-zero.
25767 2017-02-03 Jason Merrill <jason@redhat.com>
25770 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25771 avoid copying non-taken branch.
25773 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25775 PR tree-optimization/79340
25776 * tree-vect-loop.c (vectorizable_reduction): Release
25777 vec_defs elements after safe_splicing them into other vectors.
25780 PR tree-optimization/79327
25781 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25782 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25784 (format_integer): Use wide_int_to_tree instead of build_int_cst
25785 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25786 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25787 of shortest and longest sequence.
25789 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25791 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25792 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25794 2017-02-03 Walter Lee <walt@tilera.com>
25797 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25798 after initial stackframe link reg save.
25799 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25801 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25804 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25805 wu for stxssp alternative.
25807 2017-02-03 Martin Sebor <msebor@redhat.com>
25809 PR tree-optimization/79352
25810 * gimple-fold.c (get_range_strlen): Add argument.
25811 (get_range_strlen): Change return type to bool.
25812 (get_maxval_strlen): Pass in a dummy argument.
25813 * gimple-fold.h (get_range_strlen): Change return type to bool.
25814 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25815 * tree.h (array_at_struct_end_p): Add argument.
25816 * tree.c (array_at_struct_end_p): Handle it.
25818 2017-02-03 Martin Liska <mliska@suse.cz>
25821 * multiple_target.c (create_dispatcher_calls): Redirect edge
25822 from a caller of a dispatcher.
25823 (expand_target_clones): Make the clones local.
25824 (ipa_target_clone): Do both target clones and resolvers.
25825 (ipa_dispatcher_calls): Remove the pass.
25826 (pass_dispatcher_calls::gate): Likewise.
25827 (make_pass_dispatcher_calls): Likewise.
25828 * passes.def (pass_target_clone): Put as very first IPA early
25831 2017-02-03 Martin Liska <mliska@suse.cz>
25833 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
25834 in case of a function with ifunc attribute.
25836 2017-02-03 Martin Liska <mliska@suse.cz>
25838 * cgraph.c (cgraph_node::dump): Dump function version info.
25839 * symtab.c (symtab_node::dump_base): Add missing new line.
25841 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25843 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
25844 (ifcombine_ifandif): Use it.
25846 2017-02-03 Martin Liska <mliska@suse.cz>
25848 * doc/invoke.texi: Document default value for
25849 use-after-scope-direct-emission-threshold.
25851 2017-02-03 Martin Liska <mliska@suse.cz>
25853 PR tree-optimization/79339
25854 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
25855 (format_floating): Likewise.
25857 2017-02-03 Martin Liska <mliska@suse.cz>
25860 * ipa-prop.c (ipa_node_params_t::insert): Remove current
25862 (ipa_node_params_t::remove): Likewise.
25863 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
25864 initialization from removed ipa_node_params_t::insert.
25865 (ipa_node_params::~ipa_node_params): Move from removed
25866 ipa_node_params_t::release.
25867 * symbol-summary.h (symbol_summary::m_released): New member.
25868 Do not release a summary twice. Do not allow to call finalizer
25869 for types of a summary that live in GGC memory.
25871 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25873 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
25876 2017-02-02 Martin Sebor <msebor@redhat.com>
25878 PR middle-end/79275
25879 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
25880 (format_string): Tighten up the range of output for non-constant
25881 strings and correct the expected range for wide non-constant strings.
25883 2017-02-02 Martin Sebor <msebor@redhat.com>
25885 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
25887 PR middle-end/32003
25888 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
25890 (-fdump-tree-@var): Add to index and document how to come up
25891 with pass-specific option and dump file names.
25892 (-fdump-passes): Clarify where to look for output.
25894 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
25896 PR middle-end/77445
25897 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
25898 statistics of the analyzed path; allow threading for speed when
25899 any of BBs along the path are optimized for speed.
25901 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
25903 PR middle-end/78468
25904 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
25905 settings of the virtual registers.
25908 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
25910 * explow.c (get_dynamic_stack_size): Take known alignment of stack
25911 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
25914 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25916 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
25917 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
25919 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25921 * config/s390/s390.md: Add missing comments with the expanded
25923 * config/s390/vector.md: Likewise.
25924 * config/s390/vx-builtins.md: Likewise.
25926 2017-02-02 Jakub Jelinek <jakub@redhat.com>
25929 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
25930 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
25931 conditions on a single line.
25933 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25935 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
25936 __S390_VX__ to __VX__.
25938 2017-02-01 Andrew Pinski <apinski@cavium.com>
25940 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
25941 stmt_info to record_stmt_cost.
25942 (vect_get_known_peeling_cost): Pass stmt_info if known to
25944 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
25945 cpu_vector_cost field into
25946 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
25947 field into vec_int_stmt_cost and vec_fp_stmt_cost.
25948 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
25949 splitting of scalar_stmt_cost and vec_stmt_cost.
25950 (thunderx_vector_cost): Likewise.
25951 (cortexa57_vector_cost): LIkewise.
25952 (exynosm1_vector_cost): Likewise.
25953 (xgene1_vector_cost): Likewise.
25954 (thunderx2t99_vector_cost): Improve after the splitting of the two
25956 (aarch64_builtin_vectorization_cost): Update for the splitting of
25957 scalar_stmt_cost and vec_stmt_cost.
25959 2017-02-01 Torvald Riegel <triegel@redhat.com>
25960 Richard Henderson <rth@redhat.com>
25962 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
25963 conditional on existance of a fast atomic load.
25964 * optabs-query.c (can_atomic_load_p): New function.
25965 * optabs-query.h (can_atomic_load_p): Declare it.
25966 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
25967 no fast atomic load is available for the particular size of access.
25968 (expand_atomic_compare_and_swap): Likewise.
25969 (expand_atomic_load): Likewise.
25970 (expand_atomic_store): Likewise.
25971 (expand_atomic_fetch_op): Likewise.
25972 * testsuite/lib/target-supports.exp
25973 (check_effective_target_sync_int_128): Remove x86 because it provides
25974 no fast atomic load.
25975 (check_effective_target_sync_int_128_runtime): Likewise.
25977 2017-02-01 Richard Biener <rguenther@suse.de>
25979 * graphite.c: Include tree-vectorizer.h for find_loop_location.
25980 (graphite_transform_loops): Provide opt-info for optimized nests.
25981 * tree-parloop.c (parallelize_loops): Provide opt-info for
25982 parallelized loops.
25984 2017-02-01 Richard Biener <rguenther@suse.de>
25986 PR middle-end/79315
25987 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
25988 was not set before.
25990 2017-02-01 Richard Biener <rguenther@suse.de>
25992 PR tree-optimization/71824
25993 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25994 Verify the loops are valid in the merged SESE region.
25995 (scop_detection::can_represent_loop_1): Check analyzing the
25996 evolution of the number of iterations in the region succeeds.
25998 2017-01-31 Ian Lance Taylor <iant@golang.org>
26000 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26001 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26003 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26005 PR preprocessor/79210
26006 * input.c (get_substring_ranges_for_loc): Replace line_width
26007 assertion with error-handling.
26009 2017-01-31 Richard Biener <rguenther@suse.de>
26011 PR tree-optimization/77318
26012 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26013 (create_pw_aff_from_tree): Take loop parameter.
26014 (add_condition_to_pbb): Pass loop of the condition to
26015 create_pw_aff_from_tree.
26017 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26019 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26020 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26022 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26026 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26028 (convert_int_to_float128): Likewise.
26029 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26030 (convert_int_to_float128): Likewise.
26031 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26032 (UNSPEC_IEEE128_CONVERT): Likewise.
26033 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26034 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26035 Use local variables for IBM extended format.
26036 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26037 (fix_trunc<mode>si2_fprs): Likewise.
26038 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26039 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26040 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26041 to know that we can now have integers of all sizes in vector
26043 (fix<uns>_<mode>di2_hw): Likewise.
26044 (float<uns>_<mode>si2_hw): Likewise.
26045 (fix_<mode>si2_hw): Likewise.
26046 (fixuns_<mode>si2_hw): Likewise.
26047 (float<uns>_<mode>di2_hw): Likewise.
26048 (float_<mode>di2_hw): Likewise.
26049 (float_<mode>si2_hw): Likewise.
26050 (floatuns_<mode>di2_hw): Likewise.
26051 (floatuns_<mode>si2_hw): Likewise.
26052 (xscvqp<su>wz_<mode>): Delete, no longer used.
26053 (xscvqp<su>dz_<mode>): Likewise.
26054 (xscv<su>dqp_<mode>): Likewise.
26055 (ieee128_mfvsrd_64bit): Likewise.
26056 (ieee128_mfvsrd_32bit): Likewise.
26057 (ieee128_mfvsrwz): Likewise.
26058 (ieee128_mtvsrw): Likewise.
26059 (ieee128_mtvsrd_64bit): Likewise.
26060 (ieee128_mtvsrd_32bit): Likewise.
26062 2017-01-31 Martin Liska <mliska@suse.cz>
26065 * ipa-prop.c (ipa_free_all_node_params): Call release method
26066 instead of ~sumbol_summary to not to trigger double times
26069 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26071 PR tree-optimization/71691
26072 * bitmap.h (class auto_bitmap): New.
26073 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26074 is_maybe_undefined instead of ssa_undefined_value_p.
26076 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26078 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26079 __S390_ARCH_LEVEL__ to __ARCH__.
26081 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26083 PR tree-optimization/79267
26084 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26085 if should_remove_lhs_p is true.
26087 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26090 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26091 (add_alignment_attribute): New.
26092 (base_type_die): Add alignment attribute.
26093 (subrange_type_die): Likewise.
26094 (modified_type_die): Likewise.
26095 (gen_array_type_die): Likewise.
26096 (gen_descr_array_type_die: Likewise.
26097 (gen_enumeration_type_die): Likewise.
26098 (gen_subprogram_die): Likewise.
26099 (gen_variable_die): Likewise.
26100 (gen_field_die): Likewise.
26101 (gen_ptr_to_mbr_type_die): Likewise.
26102 (gen_struct_or_union_type_die): Likewise.
26103 (gen_subroutine_type_die): Likewise.
26104 (gen_typedef_die): Likewise.
26105 (base_type_cmp): Compare alignment attribute.
26107 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26110 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26111 (setb_unsigned) New pattern for setb with CCUNS.
26112 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26113 subfc./subfe sequence to avoid overflow problems. Generate a
26114 shorter sequence with cmpld/setb for power9.
26115 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26116 for generating subfc. instruction.
26117 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26118 now uses this instruction.
26120 2017-01-30 Ian Lance Taylor <iant@google.com>
26123 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26124 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26126 2017-01-30 Martin Sebor <msebor@redhat.com>
26128 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26129 Move constant to the right of a relational operator.
26130 (get_mpfr_format_length, format_character, format_string): Ditto.
26131 (should_warn_p, maybe_warn): Same.
26133 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26135 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26138 * asan.c (get_translation_unit_decl): Remove function.
26139 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26141 2017-01-30 Martin Liska <mliska@suse.cz>
26143 PR gcov-profile/79259
26144 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26145 -fprofile-generate.
26147 2017-01-30 Martin Liska <mliska@suse.cz>
26150 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26151 Initialize variables with NULL value.
26153 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26156 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26158 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26160 2017-01-30 Richard Biener <rguenther@suse.de>
26162 PR tree-optimization/79276
26163 * tree-vrp.c (process_assert_insertions): Properly adjust common
26164 when removing a duplicate.
26166 * gcc.dg/torture/pr79276.c: New testcase.
26168 2017-01-30 Richard Biener <rguenther@suse.de>
26170 PR tree-optimization/79256
26171 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26172 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26174 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26176 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26179 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26180 ("*r<noxa>sbg_<mode>_sll_bitmask")
26181 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26182 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26183 Use contiguous_bitmask_nowrap_operand.
26185 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26188 * config/rs6000/altivec.h (vec_xl): Revise #define.
26189 (vec_xst): Likewise.
26191 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26193 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26195 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26197 PR rtl-optimization/79194
26198 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26199 traps before call to bypass_conditional_jumps.
26201 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26203 PR tree-optimization/71374
26204 * lra-constraints.c (check_conflict_input_operands): New.
26205 (match_reload): Use it.
26207 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26210 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26211 account to calculate conflict_set.
26213 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26215 PR rtl-optimization/78559
26216 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26217 other_insn in combine.
26219 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26221 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26222 uint16_type_node for BT_UINT16.
26224 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26226 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26227 "RTL Tests" to menu.
26228 (GIMPLE Tests): New node.
26229 (RTL Tests): New node.
26231 2017-01-27 Richard Biener <rguenther@suse.de>
26233 PR tree-optimization/79245
26234 * tree-loop-distribution.c (distribute_loop): Apply cost
26235 modeling also to detected patterns.
26237 2017-01-27 Richard Biener <rguenther@suse.de>
26239 PR tree-optimization/71433
26240 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26241 (compare_assert_loc): New function.
26242 (process_assert_insertions): Sort and optimize assert locations
26243 to remove duplicates and push down identical assertions on
26244 edges to their destination block.
26246 2017-01-27 Richard Biener <rguenther@suse.de>
26248 PR tree-optimization/79244
26249 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26250 out SSA names even if abnormal.
26252 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26254 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26255 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26256 instead of MPFR_RNDN.
26258 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26261 * arm.c (arm_option_override): Don't call build_target_option_node
26262 until after doing all option overrides.
26263 (arm_valid_target_attribute_tree): Likewise.
26265 2017-01-27 Martin Liska <mliska@suse.cz>
26267 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26268 for {cd}tors and C++ {cd}tors.
26270 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26272 * config/s390/s390.md ("*setmem_long_and")
26273 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26275 2017-01-26 Martin Sebor <msebor@redhat.com>
26277 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26280 2017-01-26 Martin Sebor <msebor@redhat.com>
26282 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26283 HAVE_DFmode before using XFmode or DFmode.
26284 (parse_directive): Avoid using the z length modifier to avoid
26285 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26287 PR middle-end/78703
26288 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26289 to accept adjustment as an array.
26290 (get_int_range): New function.
26291 (struct directive): Make width and prec arrays.
26292 (directive::set_width, directive::set_precision): Call get_int_range.
26293 (format_integer, format_floating): Handle width and precision ranges.
26294 (format_string, parse_directive): Same.
26296 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26299 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26300 comdat_type_p set, just clone them, but keep the children in the
26304 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26305 which have direct callers with -fvar-tracking-assignments enabled
26307 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26308 inside of type units.
26310 2017-01-26 Martin Sebor <msebor@redhat.com>
26312 PR middle-end/78703
26313 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26315 (struct format_result): Replace number_chars, number_chars_min,
26316 and number_chars_max with a single member of struct result_range.
26318 (format_result::operator+=): Adjust.
26319 (struct fmtresult): Remove bounded. Handle likely and unlikely
26321 (fmtresult::adjust_for_width_or_precision): New function.
26322 (fmtresult:type_max_digits): New function.
26323 (bytes_remaining): Handle likely and unlikely counters.
26324 (min_bytes_remaining): Remove.
26325 (format_percent): Simplify.
26326 (format_integer, format_floating): Set likely and unlikely counters.
26327 (get_string_length, format_character, format_string): Same.
26328 (format_plain, should_warn_p): New function.
26329 (maybe_warn): Call should_warn_p. Update diagnostic messages
26330 and handle those for all directives, including plain strings.
26331 (format_directive): Handle likely and unlikely counters.
26332 Remove unnecessary quoting from diagnostics. Add an informational
26334 (add_bytes): Remove.
26335 (pass_sprintf_length::compute_format_length): Simplify.
26336 (try_substitute_return_value): Handle likely and unlikely counters.
26338 2017-01-26 Carl Love <cel@us.ibm.com>
26340 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26341 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26343 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26346 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26347 endianess for subregs into account.
26348 * lra-constraints.c (lra_constraints): Do risky transformations
26349 always on the first iteration.
26350 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26351 last_call_used_reg_set.
26352 (process_bb_lives): Define and use last_call_used_reg_set.
26353 * lra.c (lra): Always continue after lra_constraints on the first
26356 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26358 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26360 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26362 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26364 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26365 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26366 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26367 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26368 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26369 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26370 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26371 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26372 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26374 2017-01-26 Marek Polacek <polacek@redhat.com>
26377 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26378 for the third operand.
26380 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26382 PR middle-end/79236
26383 * omp-low.c (struct omp_context): Add simt_stmt field.
26384 (scan_omp_for): Return omp_context *.
26385 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26386 context to the _simt_ SIMD stmt.
26387 (lower_omp_for): For combined SIMD with sibling _simt_
26388 SIMD, make sure to use the same decls in _looptemp_
26389 clauses as in the sibling.
26391 2017-01-26 David Sherwood <david.sherwood@arm.com>
26393 PR middle-end/79212
26394 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26397 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26400 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26401 emit fld b; fld a; if possible.
26403 * brig-builtins.def: Update copyright years.
26404 * config/arm/arm_acle_builtins.def: Update copyright years.
26406 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26409 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26410 constraint instead of o for the stxsd instruction.
26412 2017-01-25 Carl Love <cel@us.ibm.com>
26414 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26415 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26417 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26419 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26421 2017-01-25 Richard Biener <rguenther@suse.de>
26423 PR tree-optimization/69264
26424 * target.def (vector_alignment_reachable): Improve documentation.
26425 * doc/tm.texi: Regenerate.
26426 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26428 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26429 earlier changes with respect to TYPE_USER_ALIGN.
26430 (vector_alignment_reachable_p): Likewise. Improve dumping.
26432 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26435 * config/arm/arm.md (xordi3): Force constant operand into a register
26438 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26440 * doc/invoke.texi (-fstore-merging): Correct default optimization
26441 levels at which it is enabled.
26442 (-O): Move -fstore-merging from list to...
26445 2017-01-25 Richard Biener <rguenther@suse.de>
26448 * omp-expand.c: Include debug.h.
26449 (expand_omp_taskreg): Make sure to generate early debug before
26450 outlining anything from a function.
26451 (expand_omp_target): Likewise.
26452 (grid_expand_target_grid_body): Likewise.
26454 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26457 * asan.c (get_translation_unit_decl): New function.
26458 (asan_add_global): Extract modules file name from globals
26459 TRANSLATION_UNIT_DECL name.
26461 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26464 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26465 for long calls with APCS frame and VFP.
26467 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26469 * cfg.c (original_copy_tables_initialized_p): New function.
26470 * cfg.h (original_copy_tables_initialized_p): New decl.
26471 * cfgrtl.c (relink_block_chain): Guard the call to
26472 free_original_copy_tables with a call to
26473 original_copy_tables_initialized_p.
26474 * cgraph.h (symtab_node::native_rtl_p): New decl.
26475 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26476 (symtab_node::needed_p): Don't assert for early assembly output
26477 for __RTL functions.
26478 (cgraph_node::finalize_function): Set "force_output" for __RTL
26480 (cgraph_node::analyze): Bail out early for __RTL functions.
26481 (analyze_functions): Update assertion to support __RTL functions.
26482 (cgraph_node::expand): Bail out early for __RTL functions.
26483 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26485 * function.h (struct function): Update comment for field
26487 * gimple-expr.c: Include "tree-pass.h".
26488 (gimple_has_body_p): Return false for __RTL functions.
26489 * Makefile.in (OBJS): Add run-rtl-passes.o.
26490 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26492 (gcc::pass_manager::get_clean_slate): New accessor.
26493 * passes.c: Include "insn-addr.h".
26494 (should_skip_pass_p): Add logging. Update logic for running
26495 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26496 property-provider override so it is only done for gimple passes.
26498 (skip_pass): New function.
26499 (execute_one_pass): Call skip_pass when skipping passes.
26500 * read-md.c (md_reader::read_char): Support filtering
26501 the input to a subset of line numbers.
26502 (md_reader::md_reader): Initialize fields
26503 m_first_line and m_last_line.
26504 (md_reader::read_file_fragment): New function.
26505 * read-md.h (md_reader::read_file_fragment): New decl.
26506 (md_reader::m_first_line): New field.
26507 (md_reader::m_last_line): New field.
26508 * read-rtl-function.c (function_reader::create_function): Only
26509 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26510 curr_properties. Set DECL_INITIAL to a dummy block.
26511 (read_rtl_function_body_from_file_range): New function.
26512 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26514 * run-rtl-passes.c: New file.
26515 * run-rtl-passes.h: New file.
26517 2017-01-24 Jeff Law <law@redhat.com>
26519 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26522 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26524 PR tree-optimization/79159
26525 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26526 (record_nonwrapping_iv): Improve boundary using above function if no
26527 value range information.
26529 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26530 Martin Jambor <mjambor@suse.cz>
26532 * brig-builtins.def: New file.
26533 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26534 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26535 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26536 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26537 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26538 * builtin-types.def (BT_INT8): New.
26539 (BT_INT16): Likewise.
26540 (BT_UINT8): Likewise.
26541 (BT_UINT16): Likewise.
26542 (BT_FN_ULONG): Likewise.
26543 (BT_FN_UINT_INT): Likewise.
26544 (BT_FN_UINT_ULONG): Likewise.
26545 (BT_FN_UINT_LONG): Likewise.
26546 (BT_FN_UINT_PTR): Likewise.
26547 (BT_FN_ULONG_PTR): Likewise.
26548 (BT_FN_INT8_FLOAT): Likewise.
26549 (BT_FN_INT16_FLOAT): Likewise.
26550 (BT_FN_UINT32_FLOAT): Likewise.
26551 (BT_FN_UINT16_FLOAT): Likewise.
26552 (BT_FN_UINT8_FLOAT): Likewise.
26553 (BT_FN_UINT64_FLOAT): Likewise.
26554 (BT_FN_UINT16_UINT32): Likewise.
26555 (BT_FN_UINT32_UINT16): Likewise.
26556 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26557 (BT_FN_INT_PTR_INT): Likewise.
26558 (BT_FN_UINT_PTR_UINT): Likewise.
26559 (BT_FN_LONG_PTR_LONG): Likewise.
26560 (BT_FN_ULONG_PTR_ULONG): Likewise.
26561 (BT_FN_VOID_UINT64_UINT64): Likewise.
26562 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26563 (BT_FN_INT8_INT8_INT8): Likewise.
26564 (BT_FN_INT16_INT16_INT16): Likewise.
26565 (BT_FN_INT_INT_INT): Likewise.
26566 (BT_FN_UINT_FLOAT_UINT): Likewise.
26567 (BT_FN_FLOAT_UINT_UINT): Likewise.
26568 (BT_FN_ULONG_UINT_UINT): Likewise.
26569 (BT_FN_ULONG_UINT_PTR): Likewise.
26570 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26571 (BT_FN_UINT_UINT_UINT): Likewise.
26572 (BT_FN_VOID_UINT_PTR): Likewise.
26573 (BT_FN_UINT_UINT_PTR: Likewise.
26574 (BT_FN_UINT32_UINT64_PTR): Likewise.
26575 (BT_FN_INT_INT_UINT_UINT): Likewise.
26576 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26577 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26578 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26579 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26580 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26581 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26582 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26583 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26584 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26585 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26586 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26587 * doc/frontends.texi: List BRIG FE.
26588 * doc/install.texi (Testing): Add BRIG tesring requirements.
26589 * doc/invoke.texi (Overall Options): Mention BRIG.
26590 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26592 2017-01-24 Richard Biener <rguenther@suse.de>
26594 PR translation/79208
26595 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26597 2017-01-24 Martin Jambor <mjambor@suse.cz>
26600 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26601 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26602 and known_contexts.
26604 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26606 PR middle-end/79123
26607 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26608 casts from signed to unsigned really don't have a range.
26610 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26612 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26613 GMP_RNDx for compatiblity.
26615 2017-01-24 Martin Liska <mliska@suse.cz>
26618 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26619 that would prevent us to call alloca with -1 as argument.
26621 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26623 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26624 Avoid -Wformat-security warning.
26626 2017-01-23 Andrew Pinski <apinski@cavium.com>
26628 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26631 2017-01-23 Martin Sebor <msebor@redhat.com>
26633 PR middle-end/78703
26634 * gimple-ssa-sprintf.c (warn_level): New global.
26635 (format_integer): Use it here and throughout the rest of the file.
26636 Use the same switch to compute sign as base.
26637 (maybe_warn): New function.
26638 (format_directive): Factor out warnings into maybe_warn.
26639 Add debugging output. Use warn_level.
26640 (add_bytes): Use warn_level.
26641 (pass_sprintf_length::compute_format_length): Add debugging output.
26642 (try_substitute_return_value): Same.
26643 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26645 PR middle-end/78703
26646 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26647 (struct fmtresult, format_integer, format_floating): Adjust.
26648 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26649 (get_string_length, format_string,format_directive): Same.
26650 (pass_sprintf_length::compute_format_length): Same.
26651 (try_substitute_return_value): Simplify slightly.
26653 PR middle-end/78703
26654 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26655 (fmtresult::operator+=): Outlined.
26656 (struct fmtresult): Add ctors.
26657 (struct conversion_spec): Rename...
26658 (struct directive): ...to this. Add and remove data members.
26659 (directive::set_width, directive::set_precision): New functions.
26660 (format_percent): Use fmtresult ctor.
26661 (get_width_and_precision): Remove.
26662 (format_integer): Make naming changes. Avoid computing width and
26664 (format_floating): Same. Adjust indentation.
26665 (format_character, format_none): New functions.
26666 (format_string): Moved character handling to format_character.
26667 (format_directive): Remove arguments, change return type.
26668 (parse_directive): New function.
26669 (pass_sprintf_length::compute_format_length): Move directive
26670 parsing to parse_directive.
26672 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26674 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26675 (assign_assembler_name_if_needed): ... this.
26676 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26677 (assign_assembler_name_if_needed): ... this.
26678 (free_lang_data_in_cgraph): Adjust callers.
26679 * cgraphunit.c (cgraph_node::analyze): Likewise.
26680 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26683 2017-01-23 Richard Biener <rguenther@suse.de>
26685 PR tree-optimization/79088
26686 PR tree-optimization/79188
26687 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26688 resetting loop bounds after last path deletion. Reset loop
26689 bounds of the target loop, make code match the comments.
26690 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26691 Make sure loops need no fixups.
26693 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26695 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26696 exponent support with double type for first argument.
26697 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26698 type returned by __builtin_vec_extract_sig,
26699 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26700 functions from "vector int" to "vector unsigned int" or from
26701 "vector long long int" to "vector unsigned long long int".
26702 Changed type returned by __builtin_vec_extract_exp,
26703 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26704 functions from "vector int" to "vector unsigned int" or from
26705 "vector long long int" to "vector unsigned long long int".
26706 Changed return type of __builtin_vec_test_data_class,
26707 __builtin_vec_test_data_class_sp, and
26708 __builtin_vec_test_data_class_dp from "vector int" to
26709 "vector bool int" or from "vector long long int" to "vector bool
26710 long long int" and changed second argument type from "unsigned
26711 int" to "int". Added new overloaded function forms "vector float
26712 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26713 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26714 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26715 double, vector unsigned long long int)" and "vector double
26716 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26717 long int)". Changed return type of
26718 __builtin_scalar_test_data_class and
26719 __builtin_scalar_test_data_class_sp and
26720 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26721 int" and changed second argument from "unsigned int" to "int".
26722 Changed type returned by __builtin_scalar_test_neg,
26723 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26724 from "int" to "bool int". Added new overloaded function form
26725 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26726 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26727 exponent double-precision with floating point first argument.
26728 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26729 documentation of scalar_test_data_class, scalar_test_neg,
26730 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26731 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26732 vec_test_data_class built-in functions to reflect refinements in
26733 their type signatures.
26735 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26737 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26739 (aarch64_elf_asm_destructor): Likewise.
26741 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26743 PR rtl-optimization/78634
26744 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26745 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26746 * ifcvt.c (noce_try_cmove): Add missing cost check.
26748 PR rtl-optimization/71724
26749 * combine.c (if_then_else_cond): Look for situations where it is
26750 beneficial to undo the work of one of the recursive calls.
26752 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26754 PR tree-optimization/70754
26755 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26756 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26757 combined stmt before it if not NULL.
26758 (combine_chains): Process refs reversely and compute dominance point
26761 2017-01-23 Martin Liska <mliska@suse.cz>
26763 PR tree-optimization/79196
26764 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26765 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26767 (strlen_optimize_stmt): Call the renamed function.
26769 2017-01-23 Michael Matz <matz@suse.de>
26771 PR tree-optimization/78384
26772 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26774 2017-01-23 Richard Biener <rguenther@suse.de>
26776 PR tree-optimization/79186
26777 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26778 both incoming edges before moving an assert.
26780 2017-01-23 Martin Jambor <mjambor@suse.cz>
26782 * ipa-prop.c (load_from_param_1): Removed.
26783 (load_from_unmodified_param): Bits from load_from_param_1 put back
26785 (load_from_param): Removed.
26786 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26787 with stmt. Reverted back to use of load_from_unmodified_param.
26789 2017-01-23 Martin Jambor <mjambor@suse.cz>
26792 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26793 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26794 field a pointer to garbage collected vector, mark lattices and
26795 ipcp_orig_node with GTY((skip)).
26796 (ipa_get_param_count): Adjust to descriptors being a pointer.
26797 (ipa_get_param): Likewise.
26798 (ipa_get_type): Likewise.
26799 (ipa_get_param_move_cost): Likewise.
26800 (ipa_set_param_used): Likewise.
26801 (ipa_get_controlled_uses): Likewise.
26802 (ipa_set_controlled_uses): Likewise.
26803 (ipa_is_param_used): Likewise.
26804 (ipa_node_params_t): Move into garbage collector. New methods insert
26806 (ipa_node_params_sum): Annotate wth GTY(()).
26807 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26809 (ipa_load_from_parm_agg): Adjust declaration.
26810 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26811 * ipa-profile.c (ipa_profile): Likewise.
26812 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26813 (ipa_populate_param_decls): Make descriptors parameter garbage
26815 (ipa_dump_param): Adjust to descriptors being a pointer.
26816 (ipa_alloc_node_params): Likewise.
26817 (ipa_initialize_node_params): Likewise.
26818 (load_from_param_1): Make descriptors parameter garbage collected.
26819 (load_from_unmodified_param): Likewise.
26820 (load_from_param): Likewise.
26821 (ipa_load_from_parm_agg): Likewise.
26822 (ipa_node_params::~ipa_node_params): Removed.
26823 (ipa_free_all_node_params): Remove call to delete operator.
26824 (ipa_node_params_t::insert): New.
26825 (ipa_node_params_t::remove): Likewise.
26826 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26827 copy known_csts and known_contexts vectors.
26828 (ipa_read_node_info): Adjust to descriptors being a pointer.
26829 (ipcp_modif_dom_walker): Make m_descriptors field garbage
26831 (ipcp_transform_function): Make descriptors variable garbage
26834 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
26836 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
26837 * config/i386/avx512dqintrin.h: Ditto.
26838 * config/i386/avx512fintrin.h: Ditto.
26839 * config/i386/i386.c: Handle new builtins.
26840 * config/i386/i386-builtin.def: Add new builtins.
26841 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
26842 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
26844 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26845 Martin Liska <mliska@suse.cz>
26847 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
26848 * asan.c (asan_expand_poison_ifn): Support stores and use
26849 appropriate ASAN report function.
26850 * internal-fn.c (expand_ASAN_POISON_USE): New function.
26851 * internal-fn.def (ASAN_POISON_USE): Declare.
26852 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
26853 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
26854 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
26855 ASAN_POISON calls w/o LHS.
26856 * tree-ssa.c (execute_update_addresses_taken): Create clobber
26857 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
26858 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
26859 * gimplify.c (asan_poison_variables): Add attribute
26860 use_after_scope_memory to variables that really needs to live
26862 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
26863 having the attribute.
26865 2017-01-23 Martin Liska <mliska@suse.cz>
26867 * asan.c (create_asan_shadow_var): New function.
26868 (asan_expand_poison_ifn): Likewise.
26869 * asan.h (asan_expand_poison_ifn): New declaration.
26870 * internal-fn.c (expand_ASAN_POISON): Likewise.
26871 * internal-fn.def (ASAN_POISON): New builtin.
26872 * sanopt.c (pass_sanopt::execute): Expand
26873 asan_expand_poison_ifn.
26874 * tree-inline.c (copy_decl_for_dup_finish): Make function
26876 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
26877 * tree-ssa.c (is_asan_mark_p): New function.
26878 (execute_update_addresses_taken): Rewrite local variables
26879 (identified just by use-after-scope as addressable) into SSA.
26881 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26883 * doc/install.texi (Specific): opensource.apple.com uses https
26884 now. Remove trailing slash.
26886 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
26888 * README.Portability: Remove note on an Irix compatibility issue.
26890 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
26892 * gcov.c (INCLUDE_ALGORITHM): Define.
26893 (INCLUDE_VECTOR): Define.
26894 No longer include <vector> and <algorithm> directly.
26896 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
26898 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
26900 * doc/invoke.texi (Code Gen Options): Ditto.
26902 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
26905 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
26907 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
26909 rtl-optimization/79125
26910 * cprop.c (local_cprop_pass): Handle cases where we make an
26911 unconditional trap.
26913 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
26917 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
26918 read from, for big endian.
26920 2017-01-20 Jiong Wang <jiong.wang@arm.com>
26922 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
26923 register pauth builtins for LP64 only.
26925 2017-01-20 Marek Polacek <polacek@redhat.com>
26928 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
26931 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26933 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
26935 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
26936 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
26937 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
26939 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26942 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
26943 in big-endian mode when they are not a single duplicated value.
26945 2017-01-20 Richard Biener <rguenther@suse.de>
26947 * BASE-VER: Bump to 7.0.1.
26949 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
26951 * omp-low.c (omplow_simd_context): New struct. Use it...
26952 (lower_rec_simd_input_clauses): ...here and...
26953 (lower_rec_input_clauses): ...here to hold common data. Adjust all
26954 references to idx, lane, max_vf, is_simt.
26956 2017-01-20 Graham Markall <graham.markall@embecosm.com>
26958 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
26961 2017-01-20 Martin Jambor <mjambor@suse.cz>
26963 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
26964 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
26966 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
26967 (GTFILES): Rename hsa.c to hsa-common.c.
26968 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
26969 * hsa-dump.c: Likewise.
26970 * hsa-gen.c: Likewise.
26971 * hsa-regalloc.c: Likewise.
26972 * ipa-hsa.c: Likewise.
26973 * omp-expand.c: Likewise.
26974 * omp-low.c: Likewise.
26975 * toplev.c: Likewise.
26977 2017-01-20 Marek Polacek <polacek@redhat.com>
26980 * doc/invoke.texi: Document -Wduplicated-branches.
26981 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
26982 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
26983 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
26984 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
26985 return 0 only when not OEP_LEXICOGRAPHIC.
26986 (fold_build_cleanup_point_expr): Use the expression
26987 location when building CLEANUP_POINT_EXPR.
26988 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
26989 * tree.c (add_expr): Handle error_mark_node.
26991 2017-01-20 Martin Liska <mliska@suse.cz>
26994 * tree-profile.c (init_ic_make_global_vars): Do not call
26996 (gimple_init_gcov_profiler): Likewise.
26998 2017-01-20 Martin Liska <mliska@suse.cz>
27001 * cgraph.h (maybe_create_reference): Remove argument and
27003 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27005 * ipa-cp.c (create_specialized_node): Likewise.
27006 * symtab.c (symtab_node::maybe_create_reference): Handle
27007 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27009 2017-01-20 Martin Liska <mliska@suse.cz>
27011 * read-rtl-function.c (function_reader::create_function): Use
27012 build_decl instread of build_decl_stat.
27014 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27016 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27017 * config/i386/avx512dqintrin.h: Ditto.
27018 * config/i386/avx512fintrin.h: Ditto.
27019 * config/i386/i386-builtin-types.def: Add new types.
27020 * config/i386/i386.c: Handle new types.
27021 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27022 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27023 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27024 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27025 (__builtin_ia32_kshiftridi): New.
27026 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27028 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27032 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27033 define to rs6000_init_stack_protect_guard.
27034 (rs6000_init_stack_protect_guard): New function.
27036 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27037 Yunqiang Su <yunqiang.su@imgtec.com>
27039 * config.gcc (supported_defaults): Add madd4.
27040 (with_madd4): Add validation.
27041 (all_defaults): Add madd4.
27042 * config/mips/mips.opt (mmadd4): New option.
27043 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27045 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27047 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27048 (ISA_HAS_FUSED_MADD4): Likewise.
27049 * doc/invoke.texi (-mmadd4): Document the new option.
27050 * doc/install.texi (--with-madd4): Document the new option.
27052 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27054 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27055 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27056 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27057 (aarch64_init_pauth_hint_builtins): New.
27058 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27059 (aarch64_expand_builtin): Expand new builtins.
27061 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27063 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27064 * combine-stack-adj.c (no_unhandled_cfa): Handle
27065 REG_CFA_TOGGLE_RA_MANGLE.
27066 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27067 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27068 info for return address signing.
27069 (aarch64_expand_epilogue): Likewise.
27071 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27073 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27074 * config/aarch64/aarch64-protos.h
27075 (aarch64_return_address_signing_enabled): New declaration.
27076 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27078 (aarch64_expand_prologue): Sign return address before it's pushed onto
27080 (aarch64_expand_epilogue): Authenticate return address fetched from
27082 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27083 (aarch64_attributes): New function attributes for "sign-return-address".
27084 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27085 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27086 ("*do_return"): Generate combined instructions according to key index.
27087 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27088 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27090 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27091 * config/aarch64/aarch64.opt (msign-return-address=): New.
27092 * doc/extend.texi (AArch64 Function Attributes): Documents
27093 "sign-return-address=".
27094 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27096 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27098 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27099 overall option summary.
27101 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27103 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27104 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27105 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27106 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27108 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27110 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27111 -mpower9-minmax by default for -mcpu=power9.
27112 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27113 128-bit floating point.
27115 2017-01-20 Alan Modra <amodra@gmail.com>
27117 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27118 optimizing for size.
27120 2017-01-20 Alan Modra <amodra@gmail.com>
27123 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27124 for strcmp and strncmp from corresponding builtin decl.
27126 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27128 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27129 instead of i386/rtems-64.h.
27130 * config/i386/rtems-64.h: Remove.
27132 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27136 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27138 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27140 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27142 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27143 Change int to HOST_WIDE_INT.
27144 * config/aarch64/aarch64-protos.h
27145 (aarch64_simd_gen_const_vector_dup): Likewise.
27146 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27148 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27150 * langhooks-def.h (lhd_type_for_size): New decl.
27151 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27152 * langhooks.c (lhd_type_for_size): New function, taken from
27155 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27157 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27158 define_bypass for CR latency.
27159 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27160 (power9-alu2): Add define_bypass for CR latency.
27162 (power9-mul): Update insn latency.
27163 (power9-mul-compare): Update insn latency, bypass latency and remove
27166 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27168 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27170 * config/aarch64/aarch64.md
27171 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27172 aarch64_nopcrelative_literal_loads.
27173 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27175 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27177 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27178 TARGET_LOONGSON_3A.
27179 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27181 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27184 * config.gcc (supported_defaults): Add lxc1-sxc1.
27185 (with_lxc1_sxc1): Add validation.
27186 (all_defaults): Add lxc1-sxc1.
27187 * config/mips/mips.opt (mlxc1-sxc1): New option.
27188 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27190 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27191 __mips_no_lxc1_sxc1.
27192 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27193 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27194 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27196 2017-01-19 Richard Biener <rguenther@suse.de>
27198 PR tree-optimization/72488
27199 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27200 sure to restore SSA info.
27201 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27203 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27205 PR rtl-optimization/79121
27206 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27207 of the inner type when shifting an extended value.
27209 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27212 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27213 interposable aliases.
27215 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27218 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27219 Use the evmergelohi instruction.
27220 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27221 (mov_sitf_e500_subreg8_2_be): Likewise.
27222 (mov_sitf_e500_subreg12_2_le): Likewise.
27223 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27224 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27225 (mov_sitf_e500_subreg8_2_le): Likewise.
27226 (mov_sitf_e500_subreg12_2_be): Likewise.
27228 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27230 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27231 attribute from vecsimple to vecperm.
27232 (altivec_vbpermq2): Likewise.
27234 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27237 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27239 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27240 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27241 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27242 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27243 case where N arg is SIZE_MAX.
27244 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27245 (cmpstrsi): Add pattern.
27247 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27249 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27250 __builtin_vec_revb builtins.
27251 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27252 built-in functions to support generation of the ISA 3.0 XXBR<x>
27253 vector byte reverse instructions.
27254 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27255 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27256 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27257 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27258 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27259 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27260 (P9V_BUILTIN_VEC_REVB): Likewise.
27261 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27262 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27263 (p9_xxbrq_v16qi): Likewise.
27264 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27265 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27266 (p9_xxbrh_v8hi): Likewise.
27267 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27268 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27269 vec_revb built-in functions.
27271 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27273 PR rtl-optimization/78952
27274 * config/i386/i386.md (any_extract): New code iterator.
27275 (*insvqi_2): Use any_extract for source operand.
27276 (*insvqi_3): Use any_shiftrt for source operand.
27278 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27280 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27282 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27284 2017-01-18 Matthias Klose <doko@ubuntu.com>
27286 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27288 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27290 * config/rs6000/altivec.h (vec_bperm): Change #define.
27291 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27292 (altivec_vbpermq2): New define_insn.
27293 (altivec_vbpermd): Likewise.
27294 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27295 function interface.
27296 (VBPERMD): Likewise.
27297 (VBPERM): New polymorphic function interface.
27298 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27299 Add entries for P9V_BUILTIN_VEC_VBPERM.
27300 * doc/extend.texi: Add interfaces for vec_bperm.
27302 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27304 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27305 first letter of error messages.
27306 (s390_resolve_overloaded_builtin): Likewise.
27307 * config/s390/s390.c (s390_expand_builtin): Likewise.
27308 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27309 (s390_valid_target_attribute_inner_p): Likewise.
27310 * config/s390/s390.md ("tabort"): Likewise.
27312 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27314 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27315 (ISA_AVOID_DIV_HILO): New macro.
27316 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27317 (ISA_HAS_DDIV): Likewise.
27319 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27321 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27323 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27325 * doc/invoke.texi (fabi-version): Spelling fix.
27327 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27330 * doc/invoke.texi (fabi-version): Mention mangling fix for
27333 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27336 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27338 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27341 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27342 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27344 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27346 (stack_protect_test): Ditto.
27347 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27348 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27350 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27351 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27352 -mstack-protector-guard-offset=.
27353 (RS/6000 and PowerPC Options): Ditto.
27355 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27357 * config/i386/i386.h (MASK_CLASS_P): New define.
27358 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27359 there are no registers from different register sets also when
27360 mask registers are used. Update function comment.
27361 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27362 to (*k/*r) and (*k/*km) alternatives.
27364 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27366 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27367 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27368 (EH_RETURN_HANDLER_RTX): New define.
27369 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27370 Force frame pointer in EH return functions.
27371 (aarch64_expand_epilogue): Add barrier for eh_return.
27372 (aarch64_final_eh_return_addr): Remove.
27373 (aarch64_eh_return_handler_rtx): New function.
27374 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27376 (aarch64_eh_return_handler_rtx): New prototype.
27378 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27380 * config/rs6000/altivec.h (vec_rlmi): New #define.
27381 (vec_vrlnm): Likewise.
27382 (vec_rlnm): Likewise.
27383 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27384 (UNSPEC_VRLNM): Likewise.
27385 (VIlong): New mode iterator.
27386 (altivec_vrl<VI_char>mi): New define_insn.
27387 (altivec_vrl<VI_char>nm): Likewise.
27388 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27390 (VRLDNM): Likewise.
27391 (RLNM): New polymorphic function entry.
27392 (VRLWMI): New monomorphic function entry.
27393 (VRLDMI): Likewise.
27394 (RLMI): New polymorphic function entry.
27395 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27396 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27397 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27400 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27403 * dwarf2out.c (field_byte_offset): Restore the
27404 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27405 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27406 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27409 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27412 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27414 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27417 * dwarf2out.c (add_data_member_location_attribute): For constant
27418 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27419 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27420 DW_AT_byte_size attributes.
27422 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27424 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27425 after forcing to constant memory when the code model is medium.
27427 2017-01-17 Julia Koval <julia.koval@intel.com>
27430 * config/i386/avx512fintrin.h
27431 (_mm512_i32gather_ps): Change __addr type to void const*.
27432 (_mm512_mask_i32gather_ps): Ditto.
27433 (_mm512_i32gather_pd): Ditto.
27434 (_mm512_mask_i32gather_pd): Ditto.
27435 (_mm512_i64gather_ps): Ditto.
27436 (_mm512_mask_i64gather_ps): Ditto.
27437 (_mm512_i64gather_pd): Ditto.
27438 (_mm512_mask_i64gather_pd): Ditto.
27439 (_mm512_i32gather_epi32): Ditto.
27440 (_mm512_mask_i32gather_epi32): Ditto.
27441 (_mm512_i32gather_epi64): Ditto.
27442 (_mm512_mask_i32gather_epi64): Ditto.
27443 (_mm512_i64gather_epi32): Ditto.
27444 (_mm512_mask_i64gather_epi32): Ditto.
27445 (_mm512_i64gather_epi64): Ditto.
27446 (_mm512_mask_i64gather_epi64): Ditto.
27447 (_mm512_i32scatter_ps): Change __addr type to void*.
27448 (_mm512_mask_i32scatter_ps): Ditto.
27449 (_mm512_i32scatter_pd): Ditto.
27450 (_mm512_mask_i32scatter_pd): Ditto.
27451 (_mm512_i64scatter_ps): Ditto.
27452 (_mm512_mask_i64scatter_ps): Ditto.
27453 (_mm512_i64scatter_pd): Ditto.
27454 (_mm512_mask_i64scatter_pd): Ditto.
27455 (_mm512_i32scatter_epi32): Ditto.
27456 (_mm512_mask_i32scatter_epi32): Ditto.
27457 (_mm512_i32scatter_epi64): Ditto.
27458 (_mm512_mask_i32scatter_epi64): Ditto.
27459 (_mm512_i64scatter_epi32): Ditto.
27460 (_mm512_mask_i64scatter_epi32): Ditto.
27461 (_mm512_i64scatter_epi64): Ditto.
27462 (_mm512_mask_i64scatter_epi64): Ditto.
27463 * config/i386/avx512pfintrin.h
27464 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27465 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27466 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27467 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27468 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27469 (_mm512_prefetch_i32scatter_ps): Ditto.
27470 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27471 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27472 (_mm512_prefetch_i64scatter_pd): Ditto.
27473 (_mm512_prefetch_i64scatter_ps): Ditto.
27474 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27475 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27476 * config/i386/avx512vlintrin.h
27477 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27478 (_mm_mmask_i32gather_ps): Ditto.
27479 (_mm256_mmask_i32gather_pd): Ditto.
27480 (_mm_mmask_i32gather_pd): Ditto.
27481 (_mm256_mmask_i64gather_ps): Ditto.
27482 (_mm_mmask_i64gather_ps): Ditto.
27483 (_mm256_mmask_i64gather_pd): Ditto.
27484 (_mm_mmask_i64gather_pd): Ditto.
27485 (_mm256_mmask_i32gather_epi32): Ditto.
27486 (_mm_mmask_i32gather_epi32): Ditto.
27487 (_mm256_mmask_i32gather_epi64): Ditto.
27488 (_mm_mmask_i32gather_epi64): Ditto.
27489 (_mm256_mmask_i64gather_epi32): Ditto.
27490 (_mm_mmask_i64gather_epi32): Ditto.
27491 (_mm256_mmask_i64gather_epi64): Ditto.
27492 (_mm_mmask_i64gather_epi64): Ditto.
27493 (_mm256_i32scatter_ps): Change __addr type to void*.
27494 (_mm256_mask_i32scatter_ps): Ditto.
27495 (_mm_i32scatter_ps): Ditto.
27496 (_mm_mask_i32scatter_ps): Ditto.
27497 (_mm256_i32scatter_pd): Ditto.
27498 (_mm256_mask_i32scatter_pd): Ditto.
27499 (_mm_i32scatter_pd): Ditto.
27500 (_mm_mask_i32scatter_pd): Ditto.
27501 (_mm256_i64scatter_ps): Ditto.
27502 (_mm256_mask_i64scatter_ps): Ditto.
27503 (_mm_i64scatter_ps): Ditto.
27504 (_mm_mask_i64scatter_ps): Ditto.
27505 (_mm256_i64scatter_pd): Ditto.
27506 (_mm256_mask_i64scatter_pd): Ditto.
27507 (_mm_i64scatter_pd): Ditto.
27508 (_mm_mask_i64scatter_pd): Ditto.
27509 (_mm256_i32scatter_epi32): Ditto.
27510 (_mm256_mask_i32scatter_epi32): Ditto.
27511 (_mm_i32scatter_epi32): Ditto.
27512 (_mm_mask_i32scatter_epi32): Ditto.
27513 (_mm256_i32scatter_epi64): Ditto.
27514 (_mm256_mask_i32scatter_epi64): Ditto.
27515 (_mm_i32scatter_epi64): Ditto.
27516 (_mm_mask_i32scatter_epi64): Ditto.
27517 (_mm256_i64scatter_epi32): Ditto.
27518 (_mm256_mask_i64scatter_epi32): Ditto.
27519 (_mm_i64scatter_epi32): Ditto.
27520 (_mm_mask_i64scatter_epi32): Ditto.
27521 (_mm256_i64scatter_epi64): Ditto.
27522 (_mm256_mask_i64scatter_epi64): Ditto.
27523 (_mm_i64scatter_epi64): Ditto.
27524 (_mm_mask_i64scatter_epi64): Ditto.
27525 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27526 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27527 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27528 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27529 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27530 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27531 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27532 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27533 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27534 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27535 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27536 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27537 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27538 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27539 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27540 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27541 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27542 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27543 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27544 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27545 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27546 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27547 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27548 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27549 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27550 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27551 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27552 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27553 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27554 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27555 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27556 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27557 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27558 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27559 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27560 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27561 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27562 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27563 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27564 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27565 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27566 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27567 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27568 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27569 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27570 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27571 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27572 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27573 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27574 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27575 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27576 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27577 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27578 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27579 definitions accordingly.
27581 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27582 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27585 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27588 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27591 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27592 hard regs for inner regno.
27594 2017-01-17 Martin Liska <mliska@suse.cz>
27597 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27598 assumption and add comment.
27600 2017-01-17 Nathan Sidwell <nathan@acm.org>
27602 * ipa-visibility.c (localize_node): New function, broken out of ...
27603 (function_and_variable_visibility): ... here. Call it.
27605 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27607 PR middle-end/77445
27608 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27609 correctly set frequency of oudgoing edge.
27610 (duplicate_thread_path): Fix profile updating.
27612 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27615 * configure.ac: Add GCC_BASE_VER.
27616 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27617 version from BASE-VER file.
27618 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27619 (gcc.o): Depend on $(BASEVER).
27620 * common.opt (dumpfullversion): New option.
27621 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27622 * doc/invoke.texi: Document -dumpfullversion.
27623 * doc/install.texi: Document --with-gcc-major-version-only.
27624 * configure: Regenerated.
27626 2017-01-17 Richard Biener <rguenther@suse.de>
27628 PR tree-optimization/71433
27629 * tree-vrp.c (register_new_assert_for): Merge same asserts
27630 on all incoming edges.
27631 (process_assert_insertions_for): Handle insertions at the
27634 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27636 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27637 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27639 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27642 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27645 2017-01-17 Alan Modra <amodra@gmail.com>
27648 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27649 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27650 symbolic stack limit when pic.
27652 2017-01-16 Martin Sebor <msebor@redhat.com>
27654 PR tree-optimization/78608
27655 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27657 2017-01-16 Jeff Law <law@redhat.com>
27660 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27661 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27662 for several include directories that may be relative to sysroot.
27663 * config/i386/x-mingw32 (gplus_includedir): Define.
27664 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27665 (native_system_includedir): Likewise.
27666 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27667 override if TARGET_SYSTEM_ROOT is defined.
27668 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27670 PR tree-optimization/79090
27671 PR tree-optimization/33562
27672 PR tree-optimization/61912
27673 PR tree-optimization/77485
27674 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27675 and computed trims into the dump file.
27677 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27679 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27681 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27684 * gimplify.c (gimplify_init_constructor): If want_value and
27685 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27689 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27690 sequence. Formatting fixes.
27691 (doloop_optimize): Formatting fixes.
27694 * gcc.c (debug_level_greater_than_spec_func): New function.
27695 (static_spec_functions): Add debug-level-gt spec function.
27696 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27698 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27699 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27700 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27701 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27702 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27703 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27705 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27707 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27708 QImode fixups to general and mask registers only.
27710 2017-01-16 Carl Love <cel@us.ibm.com>
27712 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27713 for built-in functions
27714 vector signed char vec_nabs (vector signed char)
27715 vector signed short vec_nabs (vector signed short)
27716 vector signed int vec_nabs (vector signed int)
27717 vector signed long long vec_nabs (vector signed long long)
27718 vector float vec_nabs (vector float)
27719 vector double vec_nabs (vector double)
27720 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27722 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27723 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27724 * doc/extend.texi: Update the documentation file for the new built-in
27727 2017-01-16 Martin Sebor <msebor@redhat.com>
27729 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27732 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27734 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27735 UNSPEC_VSX__XXSPLTD to require special splat handling.
27737 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27740 * system.h: Poison strndup.
27742 2017-01-16 Alan Modra <amodra@gmail.com>
27745 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27748 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27750 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27752 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27754 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27755 call recog here. Assert that INSN_CODE (insn) is non-negative.
27757 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27760 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27762 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27763 in the currently scheduled RTL fragment.
27765 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27767 PR rtl-optimization/78751
27768 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27771 2017-01-14 Jeff Law <law@redhat.com>
27773 PR tree-optimization/79090
27774 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27775 variable length stores.
27776 (compute_trims): Delete dead assignment to *trim_tail.
27777 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27780 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27782 PR rtl-optimization/78626
27783 PR rtl-optimization/78727
27784 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27785 of a block, and split such blocks after everything else is finished.
27787 2017-01-14 Alan Modra <amodra@gmail.com>
27790 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27791 target legitimate_combined_insn.
27792 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27793 (rs6000_legitimate_combined_insn): New function.
27794 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27796 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27797 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27798 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27800 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27802 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27804 2017-01-13 Jeff Law <law@redhat.com>
27806 PR tree-optimization/33562
27807 PR tree-optimization/61912
27808 PR tree-optimization/77485
27809 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27811 (delete_dead_assignment): Likewise.
27812 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27813 statement to delete_dead_call and delete_dead_assignment.
27815 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27818 * substring-locations.c (format_warning_va): Strengthen case 1 so
27819 that both endpoints of the substring must be within the format
27820 range for just the substring to be printed.
27822 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27824 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27825 * config/i386/i386.c (ix86_target_string): Add missing options
27826 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27827 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27828 flags_other and ix86_target_other to flags2_other. Display unknown
27830 (ix86_valid_target_attribute_inner_p): Add missing options and
27831 reorder options by implied ISAs, as in ix86_target_string.
27833 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27835 * hash-table.h (hash_table::too_empty_p): New function.
27836 (hash_table::expand): Use it.
27837 (hash_table::traverse): Likewise.
27838 (hash_table::empty_slot): Use sizeof (value_type) instead of
27839 sizeof (PTR) to convert bytes to elements. Shrink the table
27840 if the current size is excessive for the current number of
27843 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
27845 * ira-costs.c (record_reg_classes): Break from the inner loop
27846 early once alt_fail is known to be true. Update outer loop
27847 handling accordingly.
27849 2017-01-13 Jeff Law <law@redhat.com>
27851 * tree-ssa-dse.c (decrement_count): New function.
27852 (increment_start_addr, maybe_trim_memstar_call): Likewise.
27853 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
27854 when we know the partially dead statement is a mem* function.
27856 PR tree-optimization/61912
27857 PR tree-optimization/77485
27858 * tree-ssa-dse.c: Include expr.h.
27859 (maybe_trim_constructor_store): New function.
27860 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
27862 PR tree-optimization/33562
27863 PR tree-optimization/61912
27864 PR tree-optimization/77485
27865 * doc/invoke.texi: Document new dse-max-object-size param.
27866 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
27867 * tree-ssa-dse.c: Include params.h.
27868 (dse_store_status): New enum.
27869 (initialize_ao_ref_for_dse): New, partially extracted from
27871 (valid_ao_ref_for_dse, normalize_ref): New.
27872 (setup_live_bytes_from_ref, compute_trims): Likewise.
27873 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
27874 (maybe_trim_partially_dead_store): Likewise.
27875 (maybe_trim_complex_store): Likewise.
27876 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
27877 Track what bytes live from the original store. Return tri-state
27878 for dead, partially dead or live.
27879 (dse_dom_walker): Add constructor, destructor and new private members.
27880 (delete_dead_call, delete_dead_assignment): New extracted from
27882 (dse_optimize_stmt): Make a member of dse_dom_walker.
27883 Use initialize_ao_ref_for_dse.
27885 PR tree-optimization/33562
27886 PR tree-optimization/61912
27887 PR tree-optimization/77485
27888 * sbitmap.h (bitmap_count_bits): Prototype.
27889 (bitmap_clear_range, bitmap_set_range): Likewise.
27890 * sbitmap.c (bitmap_clear_range): New function.
27891 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
27893 2017-01-13 Martin Liska <mliska@suse.cz>
27896 * function.c (set_cfun): Add new argument force.
27897 * function.h (set_cfun): Likewise.
27898 * ipa-inline-transform.c (inline_call): Use the function when
27899 strict alising from is dropped for function we inline to.
27901 2017-01-13 Richard Biener <rguenther@suse.de>
27903 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
27904 for dumping GIMPLE INTEGER_CSTs.
27906 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27908 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
27909 to 201112L since C++17.
27911 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
27914 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
27915 if -fsanitize=kernel-address is present.
27917 2017-01-13 Richard Biener <rguenther@suse.de>
27919 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
27920 as _Literal ( type ) number in case usual suffixes do not
27921 preserve all information.
27923 2017-01-13 Richard Biener <rguenther@suse.de>
27925 PR tree-optimization/77283
27926 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
27927 and ssa-iterators.h.
27928 (is_feasible_trace): Implement a cost model based on joiner
27931 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
27934 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
27935 char or short to __float128/_Float128 directly.
27937 2017-01-12 Martin Sebor <msebor@redhat.com>
27939 to -Wformat-overflow.
27940 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
27941 (min_bytes_remaining): Same.
27942 (get_string_length): Same.
27943 (format_string): Same.
27944 (format_directive): Same.
27946 (pass_sprintf_length::handle_gimple_call): Same.
27948 2017-01-12 Jakub Jelinek <jakub@redhat.com>
27950 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
27951 info.nowrite calls with no lhs that can't throw. Return bool
27952 whether gsi_remove has been called or not.
27953 (pass_sprintf_length::handle_gimple_call): Return bool whether
27954 try_substitute_return_value called gsi_remove. Formatting fix.
27955 (pass_sprintf_length::execute): Don't use gsi_remove if
27956 handle_gimple_call returned true.
27959 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
27960 be removed due to side-effects, don't remove following barrier nor
27961 turn the successor edge into fallthru edge.
27963 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27966 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
27967 element-reversing loads and stores as not swappable.
27969 2017-01-12 Nathan Sidwell <nathan@acm.org>
27970 Nicolai Stange <nicstange@gmail.com>
27972 * combine.c (try_combine): Don't ignore result of overlap checking
27973 loop. Combine overlap & asm check into single loop.
27975 2017-01-12 Richard Biener <rguenther@suse.de>
27977 * tree-pretty-print.c (dump_generic_node): Provide -gimple
27978 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
27980 2017-01-12 Richard Biener <rguenther@suse.de>
27982 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
27983 and TS_TARGET_OPTION directly derive from TS_BASE.
27984 * tree-core.h (tree_optimization_option): Derive from tree_base.
27985 (tree_target_option): Likewise.
27987 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
27989 * config/i386/i386.c (memory_address_length): Increase len
27990 only when rip_relative_addr_p returns false.
27992 2017-01-11 Julia Koval <julia.koval@intel.com>
27994 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
27995 (OPTION_MASK_ISA_SGX_SET): New.
27996 (ix86_handle_option): Handle OPT_msgx.
27997 * config.gcc: Added sgxintrin.h.
27998 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
27999 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28000 * config/i386/i386.c (ix86_target_string): Add -msgx.
28002 (ix86_option_override_internal): Handle new options.
28003 (ix86_valid_target_attribute_inner_p): Add sgx.
28004 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28005 * config/i386/i386.opt: Add msgx.
28006 * config/i386/sgxintrin.h: New file.
28007 * config/i386/x86intrin.h: Add sgxintrin.h.
28009 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28012 * fold-const.c (maybe_nonzero_address): Return 1 for function
28014 (tree_single_nonzero_warnv_p): Don't handle function local objects
28018 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28021 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28024 * opts.c: Include "spellcheck.h"
28025 (struct string_fragment): New struct.
28026 (struct edit_distance_traits<const string_fragment &>): New
28028 (get_closest_sanitizer_option): New function.
28029 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28031 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28033 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28035 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28036 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28037 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28038 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28039 for initial die_offset if dwarf_split_debug_info.
28040 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28041 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28043 (output_skeleton_debug_sections): Formatting fix. Use
28044 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28045 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28047 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28049 * config/arm/cortex-a53.md: Add bypasses for
28050 cortex_a53_r2f_cvt.
28051 (cortex_a53_r2f): Only use for transfers.
28052 (cortex_a53_f2r): Likewise.
28053 (cortex_a53_r2f_cvt): Add reservation for conversions.
28054 (cortex_a53_f2r_cvt): Likewise.
28056 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28058 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28059 to all inlined functions, change static to extern.
28061 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28064 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28066 (arm_assemble_integer): Likewise.
28068 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28070 * config.gcc: Use new awk script to check CPU, FPU and architecture
28071 parameters for --with-... options.
28072 * config/arm/parsecpu.awk: New file
28073 * config/arm/arm-cpus.in: New file.
28074 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28076 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28078 * config/arm/t-arm: Update dependency rules.
28079 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28080 of processing .def files.
28081 * config/arm/genopt.sh: Deleted.
28082 * config/arm/gentune.sh: Deleted.
28083 * config/arm/arm-cores.def: Deleted.
28084 * config/arm/arm-arches.def: Deleted.
28085 * config/arm/arm-fpus.def: Deleted.
28086 * config/arm/arm-tune.md: Regenerated.
28087 * config/arm/arm-tables.opt: Regenerated.
28088 * config/arm/arm-cpu.h: New generated file.
28089 * config/arm/arm-cpu-data.h: New generated file.
28090 * config/arm/arm-cpu-cdata.h: New generated file.
28092 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28095 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28097 (input_varpool_node): Unpack dynamically_initialized bit.
28099 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28101 PR rtl-optimization/79032
28102 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28103 the alignment of the adjusted memory reference against that of MODE,
28104 instead of the alignment of the original memory reference.
28106 2017-01-11 Martin Jambor <mjambor@suse.cz>
28108 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28110 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28111 decorated functions.
28113 2017-01-11 Richard Biener <rguenther@suse.de>
28115 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28116 set range/nonnull info for PHI results. Do not set it on
28117 stmts marked for removal.
28119 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28121 * expr.c (store_field): In the bitfield case, fetch the return value
28122 from the registers before applying a single big-endian adjustment.
28123 Always do a final load for a BLKmode value not larger than a word.
28125 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28128 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28129 that we correctly handle column numbers greater than
28130 LINE_MAP_MAX_COLUMN_NUMBER.
28132 2017-01-10 Martin Sebor <msebor@redhat.com>
28134 PR middle-end/78245
28135 * gimple-ssa-sprintf.c (get_destination_size): Call
28136 {init,fini}object_sizes.
28137 * tree-object-size.c (addr_object_size): Adjust.
28138 (pass_through_call): Adjust.
28139 (pass_object_sizes::execute): Adjust.
28140 * tree-object-size.h (fini_object_sizes): Declare.
28142 2017-01-10 Martin Sebor <msebor@redhat.com>
28144 PR tree-optimization/78775
28145 * builtins.c (get_size_range): Move...
28146 * calls.c: ...to here.
28147 (alloc_max_size): Accept zero argument.
28148 (operand_signed_p): Remove.
28149 (maybe_warn_alloc_args_overflow): Call get_size_range.
28150 * calls.h (get_size_range): Declare.
28152 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28154 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28155 from TI's devices.csv file as of September 2016.
28156 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28158 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28160 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28161 * doc/invoke.texi: Likewise.
28162 * doc/md.texi: Likewise.
28163 * doc/objc.texi: Likewise.
28165 2017-01-10 Joshua Conner <joshconner@google.com>
28167 * config/arm/fuchsia-elf.h: New file.
28168 * config/fuchsia.h: New file.
28169 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28170 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28172 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28174 2016-01-10 Richard Biener <rguenther@suse.de>
28176 PR tree-optimization/79034
28177 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28178 Propagate out degenerate PHIs in the joiner.
28180 2017-01-10 Martin Liska <mliska@suse.cz>
28182 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28183 (sort_congruence_classes_by_decl_uid): Likewise.
28184 (sort_congruence_class_groups_by_decl_uid): Likewise.
28185 (sem_item_optimizer::merge_classes): Sort class, groups in these
28186 classes and members in the groups by DECL_UID of declarations.
28187 This would make merge operations stable.
28189 2017-01-10 Martin Liska <mliska@suse.cz>
28191 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28192 usage of m_classes_vec.
28193 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28194 (sem_item_optimizer::get_group_by_hash): Likewise.
28195 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28196 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28197 (sem_item_optimizer::verify_classes): Likewise.
28198 (sem_item_optimizer::process_cong_reduction): Likewise.
28199 (sem_item_optimizer::dump_cong_classes): Likewise.
28200 (sem_item_optimizer::merge_classes): Likewise.
28201 * ipa-icf.h (congruence_class_hash): Rename from
28202 congruence_class_group_hash. Remove declaration of m_classes_vec.
28204 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28206 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28207 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28208 * config.gcc: Add avx512vpopcntdqintrin.h.
28209 * config/i386/avx512vpopcntdqintrin.h: New.
28210 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28211 * config/i386/i386-builtin-types.def: Add new types.
28212 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28213 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28214 __builtin_ia32_vpopcountq_v8di_mask): New.
28215 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28216 __AVX512VPOPCNTDQ__.
28217 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28218 (PTA_AVX512VPOPCNTDQ): Define.
28219 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28220 TARGET_AVX512VPOPCNTDQ_P): Define.
28221 * config/i386/i386.opt: Add mavx512vpopcntdq.
28222 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28223 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28225 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28227 PR middle-end/77484
28228 * predict.def (PRED_CALL): Set to 67.
28230 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28232 * expr.c (store_field): In the bitfield case, if the value comes from
28233 a function call and is of an aggregate type returned in registers, do
28234 not modify the field mode; extract the value in all cases if the mode
28235 is BLKmode and the size is not larger than a word.
28237 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28240 * config/i386/cpuid.h: Fix undefined behavior.
28242 2017-01-04 Jeff Law <law@redhat.com>
28244 PR tree-optimization/79007
28245 PR tree-optimization/67955
28246 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28247 conservative for pt.null when flag_non_call_exceptions is on.
28249 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28251 PR translation/79019
28252 PR translation/79020
28253 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28254 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28255 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28257 * config/avr/avr.opt (maccumulate-args): Likewise.
28258 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28259 * common.opt (freport-bug): Likewise.
28260 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28261 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28262 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28263 translatable string.
28264 * config/i386/i386.c (function_value_32): Likewise.
28265 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28266 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28268 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28269 * common/config/msp430/msp430-common.c (msp430_handle_option):
28271 * symtab.c (symtab_node::verify_base): Likewise.
28272 * opts.c (set_debug_level): Likewise.
28273 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28274 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28275 missing whitespace to translatable strings.
28276 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28277 * config/sh/superh.h: Likewise.
28278 * config/i386/xopintrin.h: Likewise.
28279 * config/i386/znver1.md: Likewise.
28280 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28281 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28282 * double-int.h (struct double_int): Likewise.
28283 * double-int.c (div_and_round_double): Likewise.
28284 * wide-int.cc: Likewise.
28285 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28286 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28287 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28288 (crossjumps_occurred): ... this.
28289 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28292 PR tree-optimization/78899
28293 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28294 returning bool return struct loop *, NULL for failure and the new
28296 (versionable_outer_loop_p): Don't version outer loop if it has
28297 dont_vectorized bit set.
28298 (tree_if_conversion): When versioning outer loop, ensure
28299 tree_if_conversion is performed also on the inner loop of the
28300 non-vectorizable outer loop copy.
28301 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28302 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28303 prevent vectorization of it.
28304 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28305 the outer loop vectorization of the non-scalar version is attempted
28306 before vectorization of the inner loop in scalar version. If
28307 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28308 vectorization of its inner loop.
28309 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28310 has 2 inner loops, rename also on edges from bb whose single pred
28311 is outer_loop->header. Fix typo in function comment.
28313 2017-01-09 Martin Sebor <msebor@redhat.com>
28316 * asan.c (asan_emit_stack_protection): Increase local buffer size
28317 to avoid snprintf truncation warning.
28319 2017-01-09 Andrew Pinski <apinski@cavium.com>
28321 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28322 to reference thunderx2t99 for the tuning structure
28323 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28325 (thunderx2t99_extra_costs): This.
28326 * config/aarch64/aarch64-tune.md: Regenerate.
28327 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28328 (vulcan_addrcost_table): This.
28329 (vulcan_regmove_cost): Rename to ...
28330 (thunderx2t99_regmove_cost): This.
28331 (vulcan_vector_cost): Rename to ...
28332 (thunderx2t99_vector_cost): this.
28333 (vulcan_branch_cost): Rename to ...
28334 (thunderx2t99_branch_cost): This.
28335 (vulcan_tunings): Rename to ...
28336 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28337 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28339 2017-01-09 Martin Jambor <mjambor@suse.cz>
28343 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28344 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28345 (propagate_vr_accross_jump_function): Use the above function for all
28346 value range computations for pass-through jump functions and type
28347 converasion from explicit value range values.
28348 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28349 parameters from TYPE_ARG_TYPES.
28350 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28351 (ipa_write_node_info): Stream type of the actual argument.
28352 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28354 2017-01-09 Martin Liska <mliska@suse.cz>
28357 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28358 (lookup_compiler): Do not show error message with have_E.
28360 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28362 PR tree-optimization/78938
28363 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28364 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28365 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28366 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28369 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28371 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28374 2017-01-09 Richard Biener <rguenther@suse.de>
28376 PR tree-optimization/78997
28377 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28378 name condition properly.
28380 2017-01-09 Richard Biener <rguenther@suse.de>
28383 * dwarf2out.c (is_cxx): New overload with context.
28384 (is_naming_typedef_decl): Use it.
28386 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28388 * invoke.texi (Option Summary): Correct spacing in option lists
28389 and add line breaks to fix over-long lines.
28391 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28393 PR middle-end/17660
28395 * extend.texi (Common Variable Attributes): Add xref to GCC
28396 Internals manual to explain mode attribute keywords.
28398 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28401 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28402 and Preprocessor Options.
28403 (Options for Linking): Document -pthread here....
28404 (RS/6000 and PowerPC Options): ...not here.
28405 (Solaris 2 Options): ...or here.
28406 * doc/cppopts.texi: Document -pthread.
28408 2017-01-08 Martin Sebor <msebor@redhat.com>
28410 PR middle-end/77708
28411 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28412 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28413 New member functions.
28414 (format_directive): Used them.
28416 (pass_sprintf_length::handle_gimple_call): Same.
28417 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28418 to avoid truncation for any argument.
28419 (extract_affine_mul): Same.
28420 * tree.c (get_file_function_name): Same.
28422 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28424 PR middle-end/77484
28425 * predict.def (PRED_INDIR_CALL): Set to 86.
28427 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28429 PR preprocessor/54124
28430 * doc/cppopts.texi: Reformat -d subtable to list the full name
28431 of the options. Add cross-reference to the docs for the general
28432 compiler -d options.
28433 * doc/invoke.texi (Developer Options): Add cross-reference to the
28434 preprocessor-specific -d option documentation.
28436 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28438 PR preprocessor/13498
28439 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28440 redudant material, and reflect new command-line options.
28441 (System Headers): Likewise.
28443 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28445 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28446 -isystem, and -idirafter. Copy-edit.
28447 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28448 default for -ftrack-macro-expansion. Delete obsolete and
28449 badly-formatted implementation details about -fdebug-cpp output.
28450 * doc/cppwarnopts.texi: Copy-edit.
28452 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28455 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28456 that the transition from a max line width >= 1<<10 to narrower
28457 lines works correctly.
28459 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28461 * doc/options.texi (PerFunction): New.
28462 * opt-functions.awk (switch_flags): Map both Optimization and
28463 PerFunction to CL_OPTIMIZATION.
28464 * opth-gen.awk: Test for PerFunction flag along with
28466 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28467 it only when the latter is present. Skip those that don't in
28468 the hash function generator.
28469 * common.opt (fvar-tracking): Mark as PerFunction instead of
28471 (fvar-tracking-assignments): Likewise.
28472 (fvar-tracking-assignments-toggle): Likewise.
28473 (fvar-tracking-uninit): Likewise.
28475 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28477 PR translation/79018
28478 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28481 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28484 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28485 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28486 TARGET_LONG_JUMP_TABLE_OFFSETS.
28487 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28488 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28489 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28490 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28491 * config/m68k/m68k.md (tablejump expander): Likewise.
28492 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28493 TARGET_LONG_JUMP_TABLE_OFFSETS.
28494 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28495 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28497 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28498 David Holsgrove <david.holsgrove@xilinx.com>
28500 * common/config/microblaze/microblaze-common.c
28501 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28502 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28504 * config/microblaze/microblaze.c (microblaze_must_save_register)
28505 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28507 (microblaze_eh_return): New function.
28508 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28509 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28510 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28511 * config/microblaze/microblaze.md (eh_return): New pattern.
28513 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28515 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28516 GCC_DIAGNOSTIC_STRINGIFY): Define.
28518 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28520 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28522 * config/arm/arm.md (<mcrr>): New.
28524 * config/arm/arm.c (arm_arch5te): New.
28525 (arm_option_override): Set arm_arch5te.
28526 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28528 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28529 (arm_mcrr_qualifiers): ... this. New.
28530 (MRRC_QUALIFIERS): Define to...
28531 (arm_mrrc_qualifiers): ... this. New.
28532 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28534 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28535 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28536 (MRRCI, mrrc, MRRC): New.
28537 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28538 VUNSPEC_MRRC2): New.
28540 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28542 * config/arm/arm.md (<mcr>): New.
28544 * config/arm/arm.c (arm_coproc_builtin_available): Add
28545 support for mcr, mrc, mcr2 and mrc2.
28546 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28547 (arm_mcr_qualifiers): ... this. New.
28548 (MRC_QUALIFIERS): Define to ...
28549 (arm_mrc_qualifiers): ... this. New.
28550 (MCR_QUALIFIERS): Define to ...
28551 (arm_mcr_qualifiers): ... this. New.
28552 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28554 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28555 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28556 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28557 VUNSPEC_MRC2): New.
28559 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28561 * config/arm/arm.md (*ldc): New.
28565 * config/arm/arm.c (arm_coproc_builtin_available): Add
28566 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28567 (arm_coproc_ldc_stc_legitimate_address): New.
28568 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28569 'qualifier_const_pointer'.
28570 (LDC_QUALIFIERS): Define to...
28571 (arm_ldc_qualifiers): ... this. New.
28572 (STC_QUALIFIERS): Define to...
28573 (arm_stc_qualifiers): ... this. New.
28574 * config/arm/arm-protos.h
28575 (arm_coproc_ldc_stc_legitimate_address): New.
28576 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28577 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28578 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28579 stc2, stcl, stc2l): New.
28580 * config/arm/constraints.md (Uz): New.
28581 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28582 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28583 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28584 VUNSPEC_STC2L): New.
28586 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28588 * config/arm/arm.md (<cdp>): New.
28589 * config/arm/arm.c (neon_const_bounds): Rename this ...
28590 (arm_const_bounds): ... this.
28591 (arm_coproc_builtin_available): New.
28592 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28593 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28594 (CDP_QUALIFIERS): Define to...
28595 (arm_cdp_qualifiers): ... this. New.
28597 (arm_expand_builtin_args): Add case for 6 arguments.
28598 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28599 (arm_const_bounds): ... this.
28600 (arm_coproc_builtin_available): New.
28601 * config/arm/arm_acle.h (__arm_cdp): New.
28603 * config/arm/arm_acle_builtins.def (cdp): New.
28605 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28606 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28607 'arm_const_bounds'.
28608 * config/arm/types.md (coproc): New.
28609 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28610 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28611 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28612 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28614 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28616 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28617 (UBINOP_QUALIFIERS): New.
28619 (acle_builtin_data): New. Change comment.
28620 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28621 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28622 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28623 arm_acle_builtins.def.
28624 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28625 (arm_init_acle_builtins): New.
28626 (CRC32_BUILTIN): Remove.
28627 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28628 crc32cb, crc32ch and crc32cw.
28629 (arm_init_crc32_builtins): Remove.
28630 (arm_init_builtins): Use arm_init_acle_builtins rather
28631 than arm_init_crc32_builtins.
28632 (arm_expand_acle_builtin): New.
28633 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28634 * config/arm/arm_acle_builtins.def: New.
28636 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28638 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28639 (arm_builtin_datum): ... this.
28640 (arm_init_neon_builtin): Rename to ...
28641 (arm_init_builtin): ... this. Add a new parameters PREFIX
28642 and USE_SIG_IN_NAME.
28643 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28644 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28645 'arm_builtin_datum'.
28646 (arm_init_vfp_builtins): Likewise.
28647 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28648 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28649 (arm_expand_neon_args): Rename to ...
28650 (arm_expand_builtin_args): ... this. Rename builtin_arg
28651 enum values and differentiate between ARG_BUILTIN_MEMORY
28652 and ARG_BUILTIN_NEON_MEMORY.
28653 (arm_expand_neon_builtin_1): Rename to ...
28654 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28655 values, arm_expand_builtin_args and add bool parameter NEON.
28656 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28657 (arm_expand_vfp_builtin): Likewise.
28658 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28660 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28662 PR middle-end/77484
28663 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28664 * predict.c (tree_estimate_probability_bb): Reverse direction of
28665 polymorphic call predictor.
28667 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28669 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28670 (determine_pass_name_match): ...this new function and...
28671 (should_skip_pass_p): ...this new function.
28673 2017-01-06 Nathan Sidwell <nathan@acm.org>
28675 * ipa-visibility.c (function_and_variable_visibility): Reformat
28676 comments and long lines. Remove extrneous if.
28677 * symtab.c (symtab_node::make_decl_local): Fix code format.
28678 (symtab_node::set_section_for_node): Fix comment typo.
28680 2017-01-06 Martin Liska <mliska@suse.cz>
28683 * lra-constraints.c: Rename invariant to lra_invariant.
28684 * predict.c (set_even_probabilities): Initialize e to NULL.
28686 2017-01-05 Martin Sebor <msebor@redhat.com>
28688 PR tree-optimization/78910
28689 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28690 (format_integer): Correct off-by-one error in the handling
28691 of precision with negative numbers in signed conversions..
28693 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28695 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28697 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28699 PR tree-optimization/71016
28700 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28701 factor_out_conditional_conversion. Formatting fix.
28702 (factor_out_conditional_conversion): Add cond_stmt argument.
28703 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28704 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28707 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28709 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28710 read-rtl-function.o, and selftest-rtl.o.
28711 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28712 (selftest::aarch64_test_loading_full_dump): New function.
28713 (selftest::aarch64_run_selftests): New function.
28714 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28715 selftest::aarch64_run_selftests.
28716 * config/i386/i386.c
28717 (selftest::ix86_test_loading_dump_fragment_1): New function.
28718 (selftest::ix86_test_loading_call_insn): New function.
28719 (selftest::ix86_test_loading_full_dump): New function.
28720 (selftest::ix86_test_loading_unspec): New function.
28721 (selftest::ix86_run_selftests): Call the new functions.
28722 * emit-rtl.c (maybe_set_max_label_num): New function.
28723 * emit-rtl.h (maybe_set_max_label_num): New decl.
28724 * function.c (instantiate_decls): Guard call to
28725 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28726 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28728 * gensupport.c (gen_reader::gen_reader): Pass "false"
28729 for new "compact" param of rtx_reader.
28730 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28731 rather than an empty string for NULL strings.
28732 * read-md.c: Potentially include config.h rather than bconfig.h.
28733 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28734 (have_error): New global, copied from errors.c.
28735 (md_reader::read_name): Rename to...
28736 (md_reader::read_name_1): ...this, adding "out_loc" param,
28737 and converting "missing name or number" to returning false, rather
28739 (md_reader::read_name): Reimplement in terms of read_name_1.
28740 (md_reader::read_name_or_nil): New function.
28741 (md_reader::read_string): Handle "(nil)" by returning NULL.
28742 (md_reader::md_reader): Add new param "compact".
28743 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28744 (md_reader::read_file): New method.
28745 * read-md.h (md_reader::md_reader): Add new param "compact".
28746 (md_reader::read_file): New method.
28747 (md_reader::is_compact): New accessor.
28748 (md_reader::read_name): Convert return type from void to file_location.
28749 (md_reader::read_name_or_nil): New decl.
28750 (md_reader::read_name_1): New decl.
28751 (md_reader::m_compact): New field.
28752 (noop_reader::noop_reader): Pass "false" for new "compact" param
28754 (rtx_reader::rtx_reader): Add new "compact" param.
28755 (rtx_reader::read_rtx_operand): Make virtual and convert return
28756 type from void to rtx.
28757 (rtx_reader::read_until): New decl.
28758 (rtx_reader::handle_any_trailing_information): New virtual function.
28759 (rtx_reader::postprocess): New virtual function.
28760 (rtx_reader::finalize_string): New virtual function.
28761 (rtx_reader::m_in_call_function_usage): New field.
28762 (rtx_reader::m_reuse_rtx_by_id): New field.
28763 * read-rtl-function.c: New file.
28764 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28765 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28766 (selftest::verify_three_block_rtl_cfg): New decl.
28767 * read-rtl-function.h: New file.
28768 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28769 For host, include function.h, memmodel.h, and emit-rtl.h.
28770 (one_time_initialization): New function.
28771 (struct compact_insn_name): New struct.
28772 (compact_insn_names): New array.
28773 (find_code): Handle insn codes in compact dumps.
28774 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28775 (bind_subst_iter_and_attr): Likewise.
28776 (add_condition_to_string): Likewise.
28777 (add_condition_to_rtx): Likewise.
28778 (apply_attribute_uses): Likewise.
28779 (add_current_iterators): Likewise.
28780 (apply_iterators): Likewise.
28781 (initialize_iterators): Guard usage of apply_subst_iterator with
28782 #ifdef GENERATOR_FILE.
28783 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28784 (md_reader::read_mapping): Likewise.
28785 (add_define_attr_for_define_subst): Likewise.
28786 (add_define_subst_attr): Likewise.
28787 (read_subst_mapping): Likewise.
28788 (check_code_iterator): Likewise.
28789 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28791 (one_time_initialization): New function.
28792 (rtx_reader::read_until): New method.
28793 (read_flags): New function.
28794 (parse_reg_note_name): New function.
28795 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28796 Handle reuse_rtx ids.
28797 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28798 Add parsing support for RTL dumps, mirroring the special-cases in
28799 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28800 values, and calling handle_any_trailing_information.
28801 (rtx_reader::read_rtx_operand): Convert return type from void
28802 to rtx, returning return_rtx. Handle case 'e'. Call
28803 finalize_string on XSTR and XTMPL fields.
28804 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28805 "(nil)" values were omitted. Call the postprocess vfunc on the
28807 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28808 class ctor. Initialize m_in_call_function_usage. Call
28809 one_time_initialization.
28810 * rtl-tests.c (selftest::test_uncond_jump): Call
28811 set_new_first_and_last_insn.
28812 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28813 * selftest-rtl.c: New file.
28814 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28815 (selftest::get_insn_by_uid): New decl.
28816 * selftest-run-tests.c (selftest::run_tests): Call
28817 read_rtl_function_c_tests.
28818 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28819 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28822 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28824 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28825 operands in a special way. Assert that pos+len <= mode precision.
28827 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28829 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28830 3 argument Alias with unlimited for the negative form.
28831 (fno-vect-cost-model): Removed.
28833 2017-01-05 Martin Liska <mliska@suse.cz>
28835 * hsa-gen.c (gen_hsa_divmod): New function.
28836 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
28838 2017-01-05 Martin Liska <mliska@suse.cz>
28841 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
28844 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28846 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
28847 small constant length operands.
28849 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28851 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
28852 between loop iterations.
28854 2017-01-05 Martin Liska <mliska@suse.cz>
28857 * gimplify.c (gimplify_decl_expr): Compare to
28858 asan_poisoned_variables instread of checking flags.
28859 (gimplify_target_expr): Likewise.
28860 (gimplify_expr): Likewise.
28861 (gimplify_function_tree): Conditionally initialize
28862 asan_poisoned_variables.
28864 2017-01-04 Jeff Law <law@redhat.com>
28866 PR tree-optimizatin/78812
28867 * rtl.h (contains_mem_rtx_p): Prototype.
28868 * ifcvt.c (containts_mem_rtx_p): Move from here to...
28869 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
28870 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
28871 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
28872 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
28874 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28876 * input.c (assert_char_at_range): Default-initialize actual_range.
28878 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28880 * df-scan.c (df_ref_create_structure): Make regno unsigned,
28881 to match the caller.
28883 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28885 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
28886 insns after final jump in test to emit dummy move.
28888 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28890 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
28891 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
28893 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
28895 * multiple_target.c (create_dispatcher_calls): Init e_next.
28896 * tree-ssa-loop-split.c (split_loop): Init border.
28897 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
28900 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
28905 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
28906 (altivec_register_operand): Do not return true if the operand
28907 contains a SUBREG mixing SImode and SFmode.
28908 (vsx_register_operand): Likewise.
28909 (vsx_reg_sfsubreg_ok): New predicate.
28910 (vfloat_operand): Do not return true if the operand contains a
28911 SUBREG mixing SImode and SFmode.
28912 (vint_operand): Likewise.
28913 (vlogical_operand): Likewise.
28914 (gpc_reg_operand): Likewise.
28915 (int_reg_operand): Likewise.
28916 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
28917 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
28918 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
28920 (rs6000_emit_move_si_sf_subreg): New helper function.
28921 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
28922 fixup SUBREGs involving SImode and SFmode.
28923 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
28924 numbers for the new peephole2 optimization.
28925 (peephole2 for SFmode unions): New peephole2 to optimize cases in
28926 the GLIBC math library that do AND/IOR/XOR operations on single
28927 precision floating point.
28928 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
28929 target macros to say whether we need to avoid SUBREGs mixing
28931 (TARGET_ALLOW_SF_SUBREG): Likewise.
28932 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
28933 (UNSPEC_SI_FROM_SF): Likewise.
28934 (iorxor): Change spacing.
28935 (and_ior_xor): New iterator for AND, IOR, and XOR.
28936 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
28937 (movdi_from_sf_zero_ext): Likewise.
28938 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
28939 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
28940 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
28941 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
28942 (fms<mode>4): Likewise.
28943 (fnma<mode>4): Likewise.
28944 (fnms<mode>4): Likewise.
28945 (nfma<mode>4): Likewise.
28946 (nfms<mode>4): Likewise.
28948 2017-01-04 Marek Polacek <polacek@redhat.com>
28951 * doc/invoke.texi: Document -Wpointer-compare.
28953 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28955 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
28958 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
28959 descriptions for -gdwarf-5 and emit them as uleb128 instead of
28962 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
28965 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
28966 documentation of the powerpc_popcntb_ok attribute.
28967 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28968 code to issue warning messages if a requested CPU configuration is
28969 not supported by the binary (assembler and loader) toolchain.
28970 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
28971 made to define a built-in function that has been disabled.
28972 (paired_init_builtins): Add assertion to prevent ICE if attempt is
28973 made to define a built-in function that has been disabled.
28974 (altivec_init_builtins): Add comment explaining why definition
28975 of the DST built-in functions is not preceded by an assertion
28976 check. Add assertions to prevent ICE if attempts are made to
28977 define an altivec predicate or an abs* built-in function that has
28979 (htm_init_builtins): Add comment explaining why definition of the
28980 htm built-in functions is not preceded by an assertion check.
28982 2017-01-04 Jeff Law <law@redhat.com>
28984 PR tree-optimizatin/67955
28985 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
28986 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
28987 the points-to solution does not include pt_null. Use DECL_PT_UID
28990 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
28992 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
28993 Use gen_int_mode instead of gen_lopwart for const_int operands.
28995 2017-01-04 Jakub Jelinek <jakub@redhat.com>
28997 PR tree-optimization/71563
28998 * match.pd: Simplify X << Y into X if Y is known to be 0 or
28999 out of range value - has low bits known to be zero.
29001 2017-01-04 Alan Modra <amodra@gmail.com>
29003 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29004 * configure: Regenerate.
29005 * config.in: Regenerate.
29007 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29010 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29011 a substring of the message, but strcmp with the whole message. Ifdef
29012 ENABLE_NLS, translate the message first using dgettext.
29014 2017-01-03 Jeff Law <law@redhat.com>
29016 PR tree-optimizatin/78856
29017 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29018 (mark_threaded_blocks): Remove code to truncate thread paths that
29019 cross multiple loop headers. Instead invalidate the cached loop
29020 iteration information and handle case of a thread path walking
29021 into an irreducible region.
29023 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29026 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29027 assertions. Add support for doing the signbit if the IEEE 128-bit
29028 floating point value is in a GPR.
29029 * config/rs6000/rs6000.md (Fsignbit): Delete.
29030 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29031 Update the length attribute if the value is in a GPR.
29032 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29033 the sign or zero extension instruction, since the value is always 0/1.
29034 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29037 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29038 extracting SImode to a GPR register so that we can generate a
29039 store, limit the vector to be in a traditional Altivec register
29040 for the vextuwrx instruction.
29042 2017-01-03 Ian Lance Taylor <iant@google.com>
29044 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29046 2017-01-03 Martin Sebor <msebor@redhat.com>
29048 PR tree-optimization/78696
29049 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29050 precision. Use MPFR for %f for greater fidelity. Correct handling
29052 (pass_sprintf_length::compute_format_length): Set width and precision
29053 specified by asrerisk to void_node for vararg functions.
29054 (try_substitute_return_value): Adjust dump output.
29056 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29058 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29060 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29062 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29063 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29064 -mlra/-mno-lra was passed to the compiler.
29066 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29068 PR rtl-optimization/65618
29069 * emit-rtl.c (try_split): Move initialization of "before" and
29070 "after" to just before the call to emit_insn_after_setloc.
29072 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29074 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29076 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29078 * dwarf2out.c (gen_enumeration_type_die): When
29079 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29081 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29083 PR tree-optimization/78965
29084 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29085 Change first argument from const call_info & to call_info &. For %n
29086 set info.nowrite to false.
29088 PR middle-end/78901
29089 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29090 possibly throwing calls.
29092 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29093 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29094 and fns handling, rather than in a separate case SSA_NAME.
29096 2017-01-02 Jeff Law <law@redhat.com>
29098 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29099 fixes for first_period and second_period variables.
29101 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29104 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29105 (*insvqi_1): New insn pattern.
29106 (*insvqi_1_mem_rex64): Ditto.
29107 (*insvqi_2): Ditto.
29108 (*insvqi_3): Rename from *insvqi.
29110 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29112 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29114 * doc/cfg.texi (Edges): Remove reference to Java.
29115 (Maintaining the CFG): Ditto.
29117 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29119 PR middle-end/77674
29120 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29121 transparent aliases.
29123 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29125 PR middle-end/77484
29126 * predict.def (PRED_CALL): Update hitrate.
29127 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29128 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29129 into direct/indirect/polymorphic variants.
29131 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29133 Update copyright years.
29135 * gcc.c (process_command): Update copyright notice dates.
29136 * gcov-dump.c (print_version): Ditto.
29137 * gcov.c (print_version): Ditto.
29138 * gcov-tool.c (print_version): Ditto.
29139 * gengtype.c (create_file): Ditto.
29140 * doc/cpp.texi: Bump @copying's copyright year.
29141 * doc/cppinternals.texi: Ditto.
29142 * doc/gcc.texi: Ditto.
29143 * doc/gccint.texi: Ditto.
29144 * doc/gcov.texi: Ditto.
29145 * doc/install.texi: Ditto.
29146 * doc/invoke.texi: Ditto.
29148 Copyright (C) 2017 Free Software Foundation, Inc.
29150 Copying and distribution of this file, with or without modification,
29151 are permitted in any medium without royalty provided the copyright
29152 notice and this notice are preserved.