1 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4 when copying loop nest with only one inner loop.
6 2017-10-10 Richard Biener <rguenther@suse.de>
8 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
9 blocks if SCEV is active.
10 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13 (scev_initialize): Assert we are not yet initialized.
15 2017-10-10 Bin Cheng <bin.cheng@arm.com>
17 * tree-loop-distribution.c (generate_loops_for_partition): Remove
18 inner loop's exit stmt by making it always exit the loop, otherwise
19 we would generate an infinite empty loop.
21 2017-10-10 Bin Cheng <bin.cheng@arm.com>
23 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
24 renaming variables in new preheader if it's deleted.
26 2017-10-10 Bin Cheng <bin.cheng@arm.com>
28 * tree-loop-distribution.c (struct partition): Remove unused field
29 loops of the structure.
30 (partition_alloc, partition_free): Ditto.
31 (build_rdg_partition_for_vertex): Ditto.
33 2017-10-09 Jeff Law <law@redhat.com>
35 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
36 return type to match prototype and documentation.
38 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
40 * config/rs6000/rs6000.c (processor_costs): Move to ...
41 * config/rs6000/rs6000.h: ... here.
42 (rs6000_cost): Declare.
44 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
46 * except.c (setjmp_fn): New global variable.
47 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
48 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
49 if DONT_USE_BUILTIN_SETJMP is defined.
51 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
53 * target.def (insn_cost): New hook.
54 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
55 * doc/tm.texi: Regenerate.
56 * rtlanal.c (insn_cost): Use the new hook.
58 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
60 * combine.c (combine_validate_cost): Compute the new insn_cost,
61 not just pattern_cost.
62 (try_combine): Adjust comment.
64 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
66 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
68 * combine.c (uid_insn_cost): Adjust comment.
69 (combine_validate_cost): Adjust comment. Use pattern_cost instead
71 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
72 * dse.c (find_shift_sequence): Ditto.
73 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
74 (bb_valid_for_noce_process_p): Use pattern_cost.
75 * rtl.h (insn_rtx_cost): Delete.
76 (pattern_cost): New prototype.
77 (insn_cost): New prototype.
78 * rtlanal.c (insn_rtx_cost): Rename to...
79 (pattern_cost): ... this.
82 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
84 * config/i386/i386.md (*jcc_2): Remove insn pattern.
85 (*jcc<mode>_0_r_i387): Ditto.
86 (*jccxf_r_i387): Ditto.
87 (*jcc<mode>_r_i387): Ditto.
88 (*jccu<mode>_r_i387): Ditto.
89 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
90 (*jcc): Rename from *jcc_1.
92 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
94 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
95 deferred rescans after the lvx/stvx recombination pre-pass.
97 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
99 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
100 memory operation instruction support.
101 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
102 (rs6000-ibm-aix[789]*): Likewise.
103 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
104 Document new functions.
106 2017-10-09 Richard Biener <rguenther@suse.de>
108 PR tree-optimization/82397
109 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
110 equality only for semantically equal trees.
112 2017-10-09 Richard Biener <rguenther@suse.de>
114 PR tree-optimization/82449
115 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
116 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
117 allow constant addresses.
118 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
121 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
123 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
126 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
129 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
132 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
135 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
137 2017-10-09 Jakub Jelinek <jakub@redhat.com>
140 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
141 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
143 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
145 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
146 (WI_BINARY_PREDICATE_RESULT): Likewise.
147 (wi::binary_traits::operator_result): New type.
148 (wi::binary_traits::predicate_result): Likewise.
149 (generic_wide_int::operator~, unary generic_wide_int::operator-)
150 (generic_wide_int::operator==, generic_wide_int::operator!=)
151 (generic_wide_int::operator&, generic_wide_int::and_not)
152 (generic_wide_int::operator|, generic_wide_int::or_not)
153 (generic_wide_int::operator^, generic_wide_int::operator+
154 (binary generic_wide_int::operator-, generic_wide_int::operator*):
156 (operator~, unary operator-, operator==, operator!=, operator&)
157 (operator|, operator^, operator+, binary operator-, operator*): New
159 * expr.c (get_inner_reference): Use wi::bit_and_not.
160 * fold-const.c (fold_binary_loc): Likewise.
161 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
162 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
163 (bit_value_binop): Likewise.
164 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
165 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
166 (extract_range_from_binary_expr_1): Likewise.
167 (masked_increment): Likewise.
168 (simplify_bit_ops_using_ranges): Likewise.
170 2017-10-09 Martin Jambor <mjambor@suse.cz>
173 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
174 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
175 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
177 (hsa_fixup_mov_insn_type): New function.
178 (hsa_op_with_type::get_in_type): Use it.
179 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
180 immediates in an assert.
181 (hsa_op_with_type::extend_int_to_32bit): New method.
182 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
183 types. Convert to dest type if necessary.
184 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
185 (reg_for_gimple_ssa): Pass false as min32int to
186 hsa_type_for_scalar_tree_type.
187 (gen_hsa_addr): Fixup type when creating addresable temporary.
188 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
189 (gen_hsa_unary_operation): Extend operands and convert to dest type if
190 necessary. Call hsa_fixup_mov_insn_type.
191 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
192 extend operands and convert to dest type if necessary.
193 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
194 to dest type if necessary.
195 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
196 if conversion nt necessary and size matches.
197 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
198 to dest type if necessary.
199 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
200 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
202 (gen_hsa_clrsb): Likewise.
203 (gen_hsa_ffs): Likewise.
204 (gen_hsa_divmod): Extend operands and convert to dest type if
206 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
208 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
210 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
211 Remove empty default arguments. Use a brace block as output
213 (conditional return): Ditto.
215 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
216 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
218 (group_ending_nop): Ditto.
220 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
221 (splitters for those): Ditto.
223 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
225 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
226 a conditional jump (and the compare for it) so that pc_rtx is the
228 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
229 for the deleted and renamed ctr<mode>_internal[234] patterns.
230 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
231 Delete second conditional return pattern.
232 (ctr<mode>_internal2): Delete this second bdnz pattern.
233 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
234 (ctr<mode>_internal4): Delete this second bdz pattern.
236 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
238 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
239 (always_initialized_rtx_for_ssa_name_p): New predicate.
240 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
241 (finish_out_of_ssa): Free new field of SA.
242 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
243 * tree-ssa-coalesce.c: Include tree-ssa.h.
244 (get_parm_default_def_partitions): Remove extern keyword.
245 (get_undefined_value_partitions): New function.
246 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
247 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
249 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
251 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
253 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
255 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
257 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
260 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
262 * i386.c (has_dispatch): Disable for Ryzen.
264 2017-10-08 Olivier Hainque <hainque@adacore.com>
266 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
267 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
269 (thumb_set_return_address): Likewise.
271 2017-10-08 Olivier Hainque <hainque@adacore.com>
273 * common/config/arm/arm-common.c (arm_except_unwind_info):
274 Handle DWARF2_UNWIND_INFO.
276 2017-10-07 Michael Collison <michael.collison@arm.com>
278 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
281 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
283 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
284 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
285 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
286 defined, force the creation of a new block for a dispatch label.
288 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
290 * invoke.texi (Wsuggest-attribute=cold): Document.
291 * common.opt (Wsuggest-attribute=cold): New
292 * ipa-pure-const.c (warn_function_cold): New function.
293 * predict.c (compute_function_frequency): Use it.
294 * predict.h (warn_function_cold): Declare.
296 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
298 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
301 2017-10-06 Martin Liska <mliska@suse.cz>
303 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
304 keyword for member functions.
305 (struct sanopt_tree_couple): New struct.
306 (struct sanopt_tree_couple_hash): New function.
307 (struct sanopt_ctx): Add new hash_map.
308 (has_dominating_ubsan_ptr_check): New function.
309 (record_ubsan_ptr_check_stmt): Likewise.
310 (maybe_optimize_ubsan_ptr_ifn): Likewise.
311 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
312 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
314 2017-10-06 Sudakshina Das <sudi.das@arm.com>
317 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
318 aarch64_simd_valid_immediate on CONST_VECTORs.
319 (aarch64_reg_or_bic_imm): Likewise.
321 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
323 PR rtl-optimization/82396
324 * haifa-sched.c (ready_sort_real): Disable qsort checking.
326 2017-10-06 Sebastian Pop <sebpop@gmail.com>
328 * graphite-dependences.c (scop_get_reads): Move code to...
329 (scop_get_must_writes): Move code to...
330 (scop_get_may_writes): Move code to...
331 (scop_get_reads_and_writes): ... here.
332 (scop_get_dependences): Call scop_get_reads_and_writes.
334 2017-10-06 Jakub Jelinek <jakub@redhat.com>
336 PR tree-optimization/82434
337 * fold-const.h (can_native_encode_type_p,
338 can_native_encode_string_p): Remove.
339 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
340 don't encode anything, just return what would be otherwise returned.
341 (native_encode_fixed, native_encode_complex, native_encode_vector):
343 (native_encode_string): Likewise. Inline by hand
344 can_native_encode_string_p.
345 (can_native_encode_type_p): Remove.
346 (can_native_encode_string_p): Remove.
347 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
348 STRING_CSTs using can_native_encode_string_p, test all
349 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
350 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
351 argument from native_encode_expr.
352 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
353 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
356 2017-10-06 Richard Biener <rguenther@suse.de>
358 PR tree-optimization/82397
359 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
360 operand_equal_p but rely on data_ref_compare_tree for detecting
362 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
363 to match up with dr_group_sort_cmp.
365 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
368 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
370 * config/s390/s390-builtin-types.def: Regenerate.
372 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
375 * config/s390/s390-builtin-types.def: Regenerate.
376 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
377 Change flag from B_VXE to B_VX.
378 (s390_vec_min_dbl): Remove B_VXE flag.
380 2017-10-06 Richard Biener <rguenther@suse.de>
382 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
383 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
384 translate_isl_ast_to_gimple::is_valid_rename,
385 translate_isl_ast_to_gimple::get_rename,
386 translate_isl_ast_to_gimple::get_def_bb_for_const,
387 translate_isl_ast_to_gimple::get_new_name,
388 translate_isl_ast_to_gimple::collect_all_ssa_names,
389 translate_isl_ast_to_gimple::copy_loop_phi_args,
390 translate_isl_ast_to_gimple::collect_all_ssa_names,
391 translate_isl_ast_to_gimple::copy_loop_phi_args,
392 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
393 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
394 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
395 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
396 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
397 translate_isl_ast_to_gimple::copy_cond_phi_args,
398 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
399 translate_isl_ast_to_gimple::edge_for_new_close_phis,
400 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
401 translate_isl_ast_to_gimple::rename_uses,
402 translate_isl_ast_to_gimple::rename_all_uses): Remove.
403 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
404 (set_rename_for_each_def): Likewise.
405 (graphite_copy_stmts_from_block): Handle debug stmt resetting
406 here. Handle rewriting SCEV analyzable uses here.
407 (copy_bb_and_scalar_dependences): Generate code for PHI
409 (graphite_regenerate_ast_isl): Adjust.
410 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
411 (add_write, add_read): New functions.
412 (build_cross_bb_scalars_def): Use it and simplify.
413 (build_cross_bb_scalars_use): Likewise.
414 (graphite_find_cross_bb_scalar_vars): Inline into...
415 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
416 simulating out-of-SSA. Compute liveout and add dependencies.
417 (build_scops): Force an empty entry block.
418 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
420 (sese_build_liveouts): Declare.
421 (sese_trivially_empty_bb_p): Likewise.
422 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
423 compute liveout and debug_liveout.
424 (sese_bad_liveouts_use): Remove.
425 (sese_reset_debug_liveouts_bb): Likewise.
426 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
427 (sese_build_liveouts): Build liveout and debug_liveout and store
429 (new_sese_info): Adjust.
430 (free_sese_info): Likewise.
431 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
432 do not build liveout here.
433 (move_sese_in_condition): Adjust region entry.
434 (scev_analyzable_p): Match up with chrec_apply requirements.
435 (sese_trivially_empty_bb_p): New.
436 * tree-into-ssa.c (get_reaching_def): Properly support generating
437 default-defs for incremental rewrite of anonymous names.
439 2017-10-06 Richard Biener <rguenther@suse.de>
441 * graphite-sese-to-poly.c (extract_affine): For casts increasing
442 precision do not perform modulo reduction.
444 2017-10-06 Richard Biener <rguenther@suse.de>
446 PR tree-optimization/82436
447 * tree-vect-slp.c (vect_supported_load_permutation_p): More
448 conservatively choose the vectorization factor when checking
449 whether we can perform the required load permutation.
450 (vect_transform_slp_perm_load): Assert when we may not fail.
452 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
454 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
455 message for incompatible -msdata=* and -mcall-* options.
457 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
459 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
460 for post-reload scheduling.
462 2017-10-05 Tamar Christina <tamar.christina@arm.com>
464 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
466 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
468 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
469 to improve monte carlo in scimark.
471 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
473 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
474 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
475 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
476 pentium4_cost, nocona_cost): Set reassociation width to 1.
477 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
478 width to 2 for fp operations and 1 otherwise.
479 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
481 (atom_cost): Set reassociation width to 2.
482 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
483 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
484 (core_cost): Set fp reassociation width to 4 and vector to 2.
485 (ix86_reassociation_width): Rewrite using cost table; special case
486 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
487 and TARGET_AVX128_OPTIMAL.
488 * i386.h (processor_costs): Add
489 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
490 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
491 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
492 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
493 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
494 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
496 2017-10-05 Nathan Sidwell <nathan@acm.org>
498 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
500 2017-10-05 Tamar Christina <tamar.christina@arm.com>
502 * config/arm/arm.c (arm_test_fpu_data): New.
503 (arm_run_selftests): Call arm_test_fpu_data.
505 2017-10-04 Nathan Sidwell <nathan@acm.org>
507 * toplev.c (toplev::main): Remove excess parens on pretty_printer
509 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
511 2017-10-04 Sudakshina Das <sudi.das@arm.com>
513 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
514 check type for aarch64_simd_valid_immediate.
515 (aarch64_output_simd_mov_immediate): Update prototype.
516 (aarch64_simd_valid_immediate): Update prototype.
517 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
518 support for ORR-immediate.
519 (and<mode>3): modified pattern to add support for BIC-immediate.
520 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
521 now checks for valid immediate for BIC and ORR based on new enum
523 (aarch64_output_simd_mov_immediate): Function now used to output
524 BIC/ORR imm as well based on new enum argument.
525 * config/aarch64/constraints.md (Do): New vector immediate constraint.
527 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
528 (aarch64_reg_or_bic_imm): Likewise.
530 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
532 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
533 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
535 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
539 PR rtl-optimization/82396
540 * haifa-sched.c (autopref_multipass_init): Simplify
542 (autopref_rank_data): Simplify sort order.
543 * sched-int.h (autopref_multipass_data_): Remove
544 multi_mem_insn_p, min_offset and max_offset.
546 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
548 * doc/sourcebuild.texi: Document vect_peeling_profitable.
550 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
552 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
555 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
557 * doc/sourcebuild.texi: Document vect_long_mult.
559 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
561 PR tree-optimization/82413
562 * fold-const.c (build_range_check): Use widest_int when comparing
563 the maximum ETYPE value with HIGH.
565 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
567 PR rtl-optimization/82396
568 * haifa-sched.c (autopref_multipass_init): Simplify
570 (autopref_rank_data): Simplify sort order.
571 * sched-int.h (autopref_multipass_data_): Remove
572 multi_mem_insn_p, min_offset and max_offset.
574 2017-10-04 Jakub Jelinek <jakub@redhat.com>
576 PR tree-optimization/82381
577 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
578 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
581 PR tree-optimization/82374
582 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
583 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
584 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
585 current_function_decl to the new decl.
587 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
589 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
590 helper macro for IEEE float128 hardware built-in functions.
591 (SQRTF128_ODD): Add built-in functions with the round-to-odd
593 (TRUNCF128_ODD): Likewise.
594 (ADDF128_ODD): Likewise.
595 (SUBF128_ODD): Likewise.
596 (MULF128_ODD): Likewise.
597 (DIVF128_ODD): Likewise.
598 (FMAF128_ODD): Likewise.
599 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
600 UNSPEC_TRUNC_ROUND_TO_ODD.
601 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
602 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
603 floating point round to odd instructions.
604 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
605 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
606 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
607 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
608 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
609 (trunc<mode>sf2_hw): Change the truncate with round to odd
610 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
611 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
612 to odd hardware instructions.
613 (sub<mode>3_odd): Likewise.
614 (mul<mode>3_odd): Likewise.
615 (div<mode>3_odd): Likewise.
616 (sqrt<mode>2_odd): Likewise.
617 (fma<mode>4_odd): Likewise.
618 (fms<mode>4_odd): Likewise.
619 (nfma<mode>4_odd): Likewise.
620 (nfms<mode>4_odd): Likewise.
621 (trunc<mode>df2_odd): Change the truncate with round to odd
622 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
624 * doc/extend.texi (PowerPC built-in functions): Update documentation
625 for existing IEEE float128-bit built-in functions. Add built-in
626 functions that generate the IEEE 128-bit floating point round to
629 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
631 PR rtl-optimization/77729
632 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
633 to (X&(C1&~C2))|C2 transformations.
635 2017-10-03 Martin Jambor <mjambor@suse.cz>
637 PR tree-optimization/82363
638 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
639 mismatch, mark lacc written regardless of racc.
641 2017-10-03 Jakub Jelinek <jakub@redhat.com>
643 PR tree-optimization/82381
644 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
645 stmt_to_insert nor wheather SSA_NAMEs are default defs.
646 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
647 fallthrough into reassoc_stmt_dominates_stmt_p.
650 * combine.c (combine_instructions): Don't combine in unreachable
653 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
656 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
657 function to not use the have_cpu variable. Do not set cpu_index,
658 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
660 (rs6000_valid_attribute_p): Remove duplicate initializations of
661 old_optimize and func_optimize.
662 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
663 (rs6000_activate_target_options): Make global.
664 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
667 2017-10-02 Jakub Jelinek <jakub@redhat.com>
669 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
670 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
671 Set *poffset to 0 and *psize and *pmax_size to -1 if
672 *poffset + *psize overflows in HOST_WIDE_INT.
674 PR tree-optimization/82387
675 PR tree-optimization/82388
676 PR tree-optimization/82389
677 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
678 instead of live_bytes non-NULL.
680 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
683 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
686 2017-10-02 Richard Biener <rguenther@suse.de>
688 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
689 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
690 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
692 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
694 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
695 requested precision matches the type's.
696 * calls.c (alloc_max_size): Calculate the new candidate size as
697 a widest_int and use wi::to_widest when comparing it with the
698 current candidate size.
699 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
700 zero rather than integer_zero_node.
701 * match.pd: Check for a no-op conversion before using wi::add
702 rather than after. Use tree_to_uhwi when summing small shift
703 counts into an unsigned int.
705 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
706 Alan Hayward <alan.hayward@arm.com>
707 David Sherwood <david.sherwood@arm.com>
710 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
711 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
712 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
715 2017-10-02 Richard Biener <rguenther@suse.de>
717 PR tree-optimization/82355
718 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
719 a mapping for the enclosing loop but avoid generating one for
721 (copy_bb_and_scalar_dependences): Remove premature codegen
722 error on PHIs in blocks duplicated into multiple places.
723 * graphite-scop-detection.c
724 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
725 in the region use it as loop and nest to analyze the DR in.
726 (try_generate_gimple_bb): Likewise.
727 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
728 (add_loop_constraints): For blocks in a loop not in the region
729 create a dimension with a single iteration.
730 * sese.h (gbb_loop_at_index): Remove assert.
732 2017-10-01 Kevin Buettner <kevinb@redhat.com>
734 * omp-expand.c (adjust_context_scope): New function.
735 (expand_parallel_call): Call adjust_context_scope.
737 2017-10-01 Jeff Law <law@redhat.com>
739 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
740 dom_opt_dom_walker class with direct access to private members.
741 Add comments. Call test_for_singularity.
742 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
743 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
744 m_dummy_cond anymore.
745 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
747 (pass_dominator:execute): Build the dummy_cond here and pass it
748 to the dom_opt_dom_walker ctor.
749 (test_for_singularity): New function.
751 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
752 Maya Rashish <coypu@sdf.org>
754 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
755 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
756 (alpha*-*-netbsd*) Use nbsd_tm_file.
757 (arm*-*-netbsdelf*) Likewise.
758 (i[34567]86-*-netbsdelf*) Likewise.
759 (x86_64-*-netbsd*) Likewise.
760 (mips*-*-netbsd*) Likewise.
761 (powerpc-*-netbsd*) Likewise.
762 (sh*-*-netbsd*) Likewise.
763 (sparc-*-netbsdelf*) Likewise.
764 (sparc64-*-netbsd*) Likewise.
765 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
767 (vax-*-netbsdelf*) Likewise.
768 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
769 (UINT_FAST8_TYPE) Likewise.
770 (INT_FAST16_TYPE) Check CHAR_FAST16.
771 (UINT_FAST16_TYPE) Likewise.
773 2017-09-30 Jakub Jelinek <jakub@redhat.com>
776 * config/i386/i386.md
777 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
778 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
779 *divmodsi4_zext_2): New define_insn_and_split.
780 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
781 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
782 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
783 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
784 New define_insn_and_split.
785 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
786 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
787 operands[1] having DImode when mode is SImode.
789 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
790 always SImode for DIV and MOD in REG_EQUAL notes.
792 2017-09-29 Yury Gribov <tetra2005@gmail.com>
795 * match.pd: Fix handling of NaNs in pattern.
797 2017-09-29 Jeff Law <law@redhat.com>
799 * sbitmap.c (bitmap_bit_in_range_p): New function.
800 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
801 * tree-ssa-dse.c (live_bytes_read): New function.
802 (dse_classify_store): Ignore reads of dead bytes.
804 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
805 typos and whitespace errors.
806 * config/i386/predicates.md (address_no_seg_operand): Likewise.
807 * config/s390/s390.c (s390_emit_prologue): Likewise.
809 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
812 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
813 with a symbol for LRA.
815 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
817 PR rtl-optimization/82338
818 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
820 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
822 * genmodes.c (calc_wider_mode): Suppress qsort macro.
823 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
824 (qsort_chk): Declare.
825 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
826 (qsort_chk): New function.
828 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
830 PR tree-optimization/82337
831 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
832 phi definition if the PHI result appears in an abnormal PHI.
833 (find_basis_for_base_expr): Don't record a basis if the LHS of the
834 basis appears in an abnormal PHI.
836 2017-09-29 Richard Biener <rguenther@suse.de>
838 * graphite-isl-ast-to-gimple.c
839 (translate_isl_ast_to_gimple::set_codegen_error): New function.
840 (binary_op_to_tree): Use it.
841 (get_rename_from_scev): Likewise.
842 (copy_loop_phi_nodes): Likewise.
843 (copy_bb_and_scalar_dependences): Likewise.
844 (translate_pending_phi_nodes): Likewise.
846 2017-09-29 Jakub Jelinek <jakub@redhat.com>
849 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
850 for movabsq $(i32 << shift), r64.
852 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
854 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
855 index when encoding %esp as %rsp to avoid 0x67 prefix.
857 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
859 * config/i386/i386.md (*movsf_internal, *movdf_internal):
860 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
862 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
864 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
865 Extensions with more than 16 double VFP registers.
866 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
867 to_clear_mask and all code related to it. Replace the remaining
868 entry by a sbitmap and adapt code accordingly.
870 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
872 * brig-builtins.def: Change pure attributes to const.
874 2017-09-28 Joseph Myers <joseph@codesourcery.com>
876 * config.gcc (default_gnu_indirect_function): Default to yes for
877 sparc*-*-linux* with glibc.
879 2017-09-28 Joseph Myers <joseph@codesourcery.com>
881 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
882 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
883 when creating .init_array and .fini_array sections with priority
886 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
889 * config/aarch64/aarch64.c
890 (aarch64_builtin_support_vector_misalignment): Always return false
891 when misalignment is unknown.
893 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
895 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
896 this function to return false if the definition used by the swap
897 instruction is artificial, or if the memory address from which the
898 constant value is loaded is not represented by a base address held
899 in a register or if the base address register is a frame or stack
900 pointer. Additionally, return false if the base address of the
901 loaded constant is a SYMBOL_REF but is not considered to be a
903 (replace_swapped_load_constant): New function.
904 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
905 loaded constant vector with a load of a swapped constant vector.
907 2017-09-27 Carl Love <cel@us.ibm.com>
909 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
910 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
911 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
914 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
916 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
917 first, always call autopref_rank_data otherwise.
919 2017-09-27 Richard Biener <rguenther@suse.de>
921 * graphite-scop-detection.c (find_scop_parameters): Move
922 loop bound handling ...
923 (gather_bbs::before_dom_children): ... here, avoiding the need
924 to build scop_info->loop_nest.
925 (record_loop_in_sese): Remove.
926 * sese.h (sese_info_t::loop_nest): Remove.
927 * sese.c (new_sese_info): Do not allocate loop_nest.
928 (free_sese_info): Do not free loop_nest.
930 2017-09-27 Jakub Jelinek <jakub@redhat.com>
933 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
934 lhs from calls if the lhs has addressable type.
936 2017-09-27 Richard Biener <rguenther@suse.de>
938 * graphite.h (scop::max_alias_set): New member.
939 * graphite-scop-detection.c: Remove references to non-existing
941 (build_alias_sets): Record the maximum alias set used for drs.
942 (build_scops): Support zero as unlimited for
943 --param graphite-max-arrays-per-scop.
944 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
946 (build_poly_sr_1): ... here. Compute alias set based on the
947 maximum alias set used for drs rather than
948 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
950 2017-09-27 Richard Biener <rguenther@suse.de>
952 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
953 --param loop-block-tile-size=0 to disable tiling.
955 2017-09-27 Richard Biener <rguenther@suse.de>
957 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
958 (graphite-max-nb-scop-params): Document special value zero.
959 * domwalk.h (dom_walker::STOP): New symbolical constant.
960 (dom_walker::dom_walker): Add optional parameter for bb to
962 (dom_walker::~dom_walker): Declare.
963 (dom_walker::before_dom_children): Document STOP return value.
964 (dom_walker::m_user_bb_to_rpo): New member.
965 (dom_walker::m_bb_to_rpo): Likewise.
966 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
967 mapping here if not provided by the user.
968 (dom_walker::~dom_walker): Free bb to RPO mapping if not
969 provided by the user.
970 (dom_walker::STOP): Define.
971 (dom_walker::walk): Do not compute bb to RPO mapping here.
972 Support STOP return value from before_dom_children to stop
974 * graphite-optimize-isl.c (optimize_isl): If the schedule
975 is the same still generate code if -fgraphite-identity
976 or -floop-parallelize-all are given.
977 * graphite-scop-detection.c: Include cfganal.h.
978 (gather_bbs::gather_bbs): Get and pass through bb to RPO
980 (gather_bbs::before_dom_children): Return STOP for BBs
982 (build_scops): Compute bb to RPO mapping and pass it to
983 the domwalk. Treat --param graphite-max-nb-scop-params=0
984 as not limiting the number of params.
985 * graphite.c (graphite_initialize): Remove limit on the
986 number of basic-blocks in a function.
987 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
988 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
991 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
993 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
994 Adjust code to eliminate needing to do the shift right 32-bits
995 operation after XSCVDPSPN.
997 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
999 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1000 ((X / Y) != 0 -> X >= Y): Likewise.
1002 2017-09-26 Carl Love <cel@us.ibm.com>
1004 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1005 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1006 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1007 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1008 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1009 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1010 definitions and overloading.
1011 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1012 statement for P9V_BUILTIN_XST_LEN_R.
1013 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1014 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1015 define_expand and define_insn for the instructions and builtins.
1016 * doc/extend.texi: Update the built-in documentation file for the new
1018 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1019 define_insn for the instructions
1021 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1024 * gcc/config/netbsd-protos.h: New file.
1025 * gcc/config/netbsd.c: New file.
1026 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1027 * gcc/config/t-netbsd: New file.
1028 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1029 (tmake_file) Add t-netbsd.
1030 (extra_objs) Add netbsd.o.
1032 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1036 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1038 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1040 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1041 sign extension from a vector register to a GPR by doing a 32-bit
1042 direct move and then an EXTSW.
1043 (extendsi<mode>2 splitter): Likewise.
1044 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1045 right or vector extract after doing XSCVDPSPN. Use
1046 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1048 (movdi_from_sf_zero_ext): Likewise.
1049 (reload_gpr_from_vsxsf): Likewise.
1050 (p8_mfvsrd_4_disf): Delete, no longer used.
1051 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1052 then needing to move the SFmode to a GPR to use the XSCVDPSP
1053 instruction instead of FRSP and XSCVDPSPN.
1054 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1055 it is adjacent to the other XSCVSPDP insns.
1056 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1057 SFmode to be in traditional Altivec registers.
1058 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1059 (vsx_xscvspdpn): Likewise.
1060 (vsx_xscvspdpn_scalar): Likewise.
1062 2017-09-26 Martin Jambor <mjambor@suse.cz>
1064 * tree-sra.c (compare_access_positions): Put integral types first,
1065 stabilize sorting of integral types, remove conditions putting
1066 non-full-precision integers last.
1067 (sort_and_splice_var_accesses): Disable scalarization if a
1068 non-integert would be represented by a non-full-precision integer.
1070 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1072 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1073 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1074 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1075 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1076 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1077 conditionals inside the function instead of around it. Call
1078 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1079 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1081 2017-09-26 Richard Biener <rguenther@suse.de>
1083 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1085 (scop_detection::build_scop_breadth): ... this. Removed.
1086 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1087 (scop_detection::harmful_stmt_in_bb): Likewise.
1088 (scop_detection::graphite_can_represent_stmt): Likewise.
1089 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1090 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1091 (scop_detection::can_represent_loop_1): ... this. Removed.
1092 (scop_detection::harmful_loop_in_region): Simplify after inlining
1093 the above and remove more quadraticness.
1094 (build_scops): Adjust.
1095 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1098 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1101 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1102 REGNO (base) == SP_REG if base is a REG.
1105 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1106 if it is different SSA_NAME.
1107 (optimize_range_tests_cmp_bitwise): New function.
1108 (optimize_range_tests): Call it.
1110 2017-09-26 Richard Biener <rguenther@suse.de>
1112 PR tree-optimization/82321
1113 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1114 for the def being inside the loop.
1116 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1118 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1120 * config/s390/s390-builtins.def: Fix constraint on op4.
1122 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1124 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1125 independent expanders.
1126 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1127 ("vec_ordered", "vec_unordered"): New expanders.
1129 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1131 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1134 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1136 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1137 vec_unpacks_lo_v16qi.
1138 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1140 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1142 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1143 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1144 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1146 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1148 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1150 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1152 ("*vec_slb<mode>"): New insn pattern.
1153 ("vec_shr_<mode>"): New expander.
1154 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1155 and force the shift count operand to V16QImode.
1156 ("vec_srb<mode>"): Set shift count mode to V16QI.
1158 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1160 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1161 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1162 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1164 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1167 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1169 2017-09-26 Richard Biener <rguenther@suse.de>
1171 PR tree-optimization/82320
1172 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1175 2017-09-25 Jeff Law <law@redhat.com>
1177 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1178 prototype for new argument.
1179 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1180 mostly extracted from rs6000_emit_allocate_stack.
1181 (rs6000_emit_probe_stack_range_stack_clash): New function.
1182 (rs6000_emit_allocate_stack): Call
1183 rs6000_emit_probe_stack_range_stack_clash as needed.
1184 (rs6000_emit_probe_stack_range): Add additional argument
1185 to call to gen_probe_stack_range{si,di}.
1186 (output_probe_stack_range): New.
1187 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1188 (output_probe_stack_range_stack_clash): New.
1189 (rs6000_emit_prologue): Emit notes into dump file as requested.
1190 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1191 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1192 Add additional operand and pass it to output_probe_stack_range.
1194 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1196 PR tree-optimization/82163
1197 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1198 (checking_verify_loop_closed_ssa): New parameter.
1199 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1200 (check_loop_closed_ssa_stmt): Delete.
1201 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1202 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1203 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1206 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1208 * brig-builtins.def: Treat HSAIL barrier builtins as
1209 setjmp/longjump style functions.
1211 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1213 * target.def (constant_alignment): New hook.
1214 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1215 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1216 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1217 * doc/tm.texi: Regenerate.
1218 * targhooks.h (default_constant_alignment): Declare.
1219 (constant_alignment_word_strings): Likewise.
1220 * targhooks.c (default_constant_alignment): New function.
1221 (constant_alignment_word_strings): Likewise.
1222 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1223 instead of CONSTANT_ALIGNMENT.
1224 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1225 (force_const_mem): Likewise.
1226 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1227 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1228 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1229 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1230 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1232 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1233 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1234 constant_alignment_word_strings.
1235 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1236 (CONSTANT_ALIGNMENT): Likewise.
1237 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1238 (arm_constant_alignment): New function.
1239 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1240 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1241 constant_alignment_word_strings.
1242 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1243 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1244 constant_alignment_word_strings.
1245 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1246 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1247 (cris_constant_alignment): New function.
1248 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1249 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1250 (epiphany_constant_alignment): New function.
1251 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1252 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1253 constant_alignment_word_strings.
1254 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1255 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1256 constant_alignment_word_strings.
1257 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1258 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1259 constant_alignment_word_strings.
1260 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1261 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1262 * config/i386/i386.c (ix86_constant_alignment): Make static.
1263 Use the same interface as the target hook.
1264 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1265 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1266 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1267 constant_alignment_word_strings.
1268 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1269 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1270 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1271 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1272 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1273 constant_alignment_word_strings.
1274 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1275 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1276 constant_alignment_word_strings.
1277 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1278 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1279 constant_alignment_word_strings.
1280 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1281 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1283 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1284 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1285 * config/mips/mips.c (mips_constant_alignment): New function.
1286 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1287 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1288 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1289 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1290 (mmix_constant_alignment): Make static. Use the same interface
1292 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1293 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1294 constant_alignment_word_strings.
1295 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1296 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1297 constant_alignment_word_strings.
1298 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1299 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1300 constant_alignment_word_strings.
1301 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1302 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1303 (rs6000_constant_alignment): New function.
1304 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1305 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1306 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1307 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1308 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1309 (rs6000_constant_alignment): New function.
1310 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1311 * config/s390/s390.c (s390_constant_alignment): New function.
1312 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1313 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1314 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1315 constant_alignment_word_strings.
1316 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1317 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1318 (sparc_constant_alignment): New function.
1319 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1320 * config/spu/spu.c (spu_constant_alignment): New function.
1321 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1322 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1323 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1324 constant_alignment_word_strings.
1325 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1326 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1327 constant_alignment_word_strings.
1328 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1329 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1330 constant_alignment_word_strings.
1331 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1332 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1333 (visium_constant_alignment): New function.
1334 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1335 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1336 (xtensa_constant_alignment): New function.
1337 * system.h (CONSTANT_ALIGNMENT): Poison.
1339 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1341 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1342 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1343 (rs6000_builtin_valid_without_lhs): New helper function.
1344 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1345 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1347 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1349 * target.h (vec_perm_indices): Use unsigned short rather than
1351 (auto_vec_perm_indices): Likewise.
1352 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1353 Use unsigned int rather than unsigned char.
1354 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1356 2017-09-25 Richard Biener <rguenther@suse.de>
1358 * cfgloop.h (sort_sibling_loops): Declare.
1359 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1360 (sort_sibling_loops): New function sorting the sibling loop list
1362 * graphite.c (graphite_transform_loops): Sort sibling loops.
1364 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1366 * target.def (vec_perm_const_ok): Change sel parameter to
1368 * optabs-query.c (can_vec_perm_p): Update accordingly.
1369 * doc/tm.texi: Regenerate.
1370 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1371 auto_vec_perm_indices and remove separate nelt field.
1372 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1373 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1374 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1375 (aarch64_expand_vec_perm_const): Update accordingly.
1376 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1377 to vec_perm_indices.
1378 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1379 auto_vec_perm_indices and remove separate nelt field.
1380 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1381 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1382 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1384 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1385 to vec_perm_indices.
1386 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1387 sel to vec_perm_indices.
1388 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1389 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1390 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1392 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1395 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1398 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1399 on the FUNCTION_DECL function context if it has a DIE that is a
1402 2017-09-25 Richard Biener <rguenther@suse.de>
1404 PR tree-optimization/82285
1405 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1408 2017-09-25 Tom de Vries <tom@codesourcery.com>
1412 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1415 2017-09-25 Richard Biener <rguenther@suse.de>
1417 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1418 ISL errors other than isl_error_quota happen. Dump if the
1419 schedule is the same.
1420 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1421 errors instead of aborting inside ISL.
1423 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1426 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1427 of libgcc_eh for m64.
1428 * config/i386/darwin64.h: Likewise.
1430 2017-09-25 Richard Biener <rguenther@suse.de>
1433 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1434 attribute for incomplete types nor twice for complete ones.
1436 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1439 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1440 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1443 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1446 * config/i386/i386.opt (mprefer-avx256): Use
1447 ix86_target_flags variable.
1448 * config/i386/i386.c (ix86_target_string): Move
1449 -mprefer-avx256 to flag2_opts.
1451 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1454 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1455 and x != -1 | y != -1 into (x & y) != -1.
1457 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1459 * config.gcc: Add new case statement to set
1460 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1461 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1462 s390x-*-linux* case statements. Added aarch64 to the list of
1463 supported architectures.
1465 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1467 PR tree-optimization/82289
1468 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1469 STMT_VINFO_RELEVANT_P.
1471 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1472 Alan Hayward <alan.hayward@arm.com>
1473 David Sherwood <david.sherwood@arm.com>
1475 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1476 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1477 (extract_range_from_binary_expr_1): Don't call
1478 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1480 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * target.def (preferred_vector_alignment): New hook.
1485 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1487 * doc/tm.texi: Regenerate.
1488 * targhooks.h (default_preferred_vector_alignment): Declare.
1489 * targhooks.c (default_preferred_vector_alignment): New function.
1490 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1492 (DR_TARGET_ALIGNMENT): New macro.
1493 (aligned_access_p): Update commentary.
1494 (vect_known_alignment_in_bytes): New function.
1495 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1497 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1498 Calculate the misalignment based on the target alignment rather than
1500 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1501 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1502 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1503 the target alignment, rather than masking the element misalignment
1504 with the number of elements in a vector. Also use the target
1505 alignment when calculating the maximum number of peels.
1506 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1507 instead of TYPE_ALIGN_UNIT.
1508 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1509 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1510 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1511 (vect_create_data_ref_ptr): Likewise.
1512 (vect_setup_realignment): Realign by ANDing with
1513 -DR_TARGET_MISALIGNMENT.
1514 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1515 the number of peels based on DR_TARGET_ALIGNMENT.
1516 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1517 with the guaranteed alignment boundary when deciding whether
1519 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1520 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1521 (ensure_base_align): Remove stmt_info parameter. Get the
1522 target base alignment from DR_TARGET_ALIGNMENT.
1523 (vectorizable_store): Update call accordingly. Interpret
1524 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1526 (vectorizable_load): Likewise.
1528 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1529 Alan Hayward <alan.hayward@arm.com>
1530 David Sherwood <david.sherwood@arm.com>
1532 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1533 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1534 (vect_enhance_data_refs_alignment): Likewise.
1536 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1538 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1539 error. Only quit immediately if parsing is complete.
1540 (BEGIN): Initialize fatal_err and parse_done.
1541 (begin fpu, end fpu): Check number of arguments.
1542 (begin arch, end arch): Likewise.
1543 (begin cpu, end cpu): Likewise.
1544 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1545 (optalias): Likewise.
1547 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1549 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1550 * config/arm/arm-isa.h: Delete. Move definitions to ...
1551 * arm-cpus.in: ... here. Use new feature and fgroup values.
1552 * config/arm/arm.c (arm_option_override): Use lower case for feature
1554 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1555 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1556 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1558 (print_isa_bits_for): New function.
1559 (gen_isa): New function.
1560 (gen_comm_data): Use print_isa_bits_for.
1561 (define feature): New keyword.
1562 (define fgroup): New keyword.
1563 * config/arm/t-arm (TM_H): Remove.
1564 (GTM_H): Add arm-isa.h.
1565 (arm-isa.h): Add rule to generate file.
1566 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1567 case for feature bit names.
1569 2017-09-22 Richard Biener <rguenther@suse.de>
1571 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1573 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1574 print of no dependency loops ...
1575 * graphite.c (graphite_transform_loops): ... here.
1576 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1578 (same_close_phi_node, remove_duplicate_close_phi,
1579 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1580 (canonicalize_loop_closed_ssa): ... here and simplify.
1581 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1582 (optimize_isl): Use dump_printf_loc to tell when we stopped
1583 optimizing because of an ISL timeout.
1585 2017-09-22 Richard Biener <rguenther@suse.de>
1587 PR tree-optimization/82291
1588 * tree-if-conv.c (predicate_mem_writes): Make sure to
1589 remove writes in blocks predicated with false.
1591 2017-09-22 Richard Biener <rguenther@suse.de>
1593 * sese.c: Include cfganal.h.
1594 (if_region_set_false_region): Remove.
1595 (create_if_region_on_edge): Likewise.
1596 (move_sese_in_condition): Re-implement without destroying
1599 2017-09-22 Richard Biener <rguenther@suse.de>
1601 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1602 Verify both BBs contain loop PHI nodes before dispatching to
1604 (graphite_regenerate_ast_isl): Do not recompute dominators,
1605 do not verify three times. Restructure for clarity.
1606 * graphite-scop-detection.c (same_close_phi_node,
1607 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1608 defined_in_loop_p, canonicalize_loop_closed_ssa,
1609 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1610 checking and SSA rewrite, move to ...
1611 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1612 (graphite_initialize): Do not pass in ctx, do not reset the
1613 SCEV cache, compute only dominators.
1614 (graphite_transform_loops): Allocate ISL ctx after
1615 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1616 Maintain post-dominators only around build_scops.
1617 * sese.c (if_region_set_false_region): Make static. Free
1618 and recompute dominators.
1619 (move_sese_in_condition): Assert we don't get called with
1620 post-dominators computed.
1621 * sese.h (if_region_set_false_region): Remove.
1623 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1625 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1626 mode attribute for TARGET_AVX512VL.
1628 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1630 * config/i386/i386.opt (mprefer-avx256): New option.
1631 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1633 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1634 for TARGET_PREFER_AVX256.
1635 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1637 2017-09-21 Jeff Law <law@redhat.com>
1639 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1640 Fix dump output if the only stack space is for pushed registers.
1642 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1644 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1647 2017-09-21 Martin Sebor <msebor@redhat.com>
1650 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1651 code (in C++) or code that triggers warnings.
1653 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1655 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1657 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1659 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1660 * sched-rgn.c: Ditto.
1661 * sel-sched-ir.c: Ditto.
1663 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1665 * toplev.h (set_random_seed): Adjust return type.
1666 * toplev.c (init_local_tick): Move eager initialization of random_seed
1667 to get_random_seed. Adjust comment.
1668 (init_random_seed): Inline to get_random_seed, delete.
1669 (get_random_seed): Initialize random_seed lazily.
1670 (set_random_seed): Do not return previous value.
1671 (print_switch_value): Do not call get_random_seed.
1673 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1675 * cgraph.c (delete_function_version): New, broken out from...
1676 (cgraph_node::delete_function_version): ...here. Rename to
1677 cgraph_node::delete_function_version_by_decl. Update all uses.
1678 (cgraph_node::remove): Call delete_function_version.
1680 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1683 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1684 VAR_DECLs to which addressable non-volatile parameters are mapped
1685 and for id->retvar after the return value assignment. Clear
1686 id->retval and id->retbnd after inlining.
1688 2017-09-21 Richard Biener <rguenther@suse.de>
1690 PR tree-optimization/82276
1691 PR tree-optimization/82244
1692 * tree-vrp.c (build_assert_expr_for): Set
1693 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1695 (remove_range_assertions): Revert earlier change.
1697 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1700 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1702 2017-09-21 Richard Biener <rguenther@suse.de>
1704 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1705 Restore valid IL after code generation errors.
1706 * graphite.c (graphite_transform_loops): Diagnose code
1707 generation issues as MSG_MISSED_OPTIMIZATION and continue
1708 with processing SCOPs.
1710 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1711 Alan Hayward <alan.hayward@arm.com>
1712 David Sherwood <david.sherwood@arm.com>
1714 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1715 rather than choosing between plus_constant and gen_rtx_<CODE>.
1716 * expr.c (emit_push_insn): Likewise.
1717 (expand_expr_real_2): Likewise.
1719 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1720 Alan Hayward <alan.hayward@arm.com>
1721 David Sherwood <david.sherwood@arm.com>
1723 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1725 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1726 Alan Hayward <alan.hayward@arm.com>
1727 David Sherwood <david.sherwood@arm.com>
1729 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1730 calling tree_to_uhwi.
1732 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1733 Alan Hayward <alan.hayward@arm.com>
1734 David Sherwood <david.sherwood@arm.com>
1736 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1737 INTEGER_CST rather than a negative test for ADDR_EXPR.
1739 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1744 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1746 2017-09-21 Richard Biener <rguenther@suse.de>
1748 PR tree-optimization/71351
1749 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1750 graphite_create_new_loop_guard): Remove, fold remaining parts
1752 (translate_isl_ast_node_for): ... here and simplify.
1754 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1757 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1758 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1759 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1760 alternative always use QI mode, for -Os imov (=R,R) alternative
1761 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1764 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1765 Jeff Law <law@redhat.com>
1767 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1768 (allocate_stack_space): New function, partially extracted from
1770 (s390_emit_prologue): Track offset to most recent stack probe.
1771 Code to allocate space moved into allocate_stack_space.
1772 Dump actions when no stack is allocated.
1773 (s390_prologue_plus_offset): New function.
1774 (s390_emit_stack_probe): Likewise.
1776 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1778 * common.opt (Wa, Wl, Wp, g, gz=): Add
1780 (gno-column-info): Remove.
1781 (gcolumn-info): Drop RejectNegative.
1783 (gno-record-gcc-switches): Remove.
1784 (grecord-gcc-switches): Drop RejectNegative.
1785 (gno-split-dwarf): Remove.
1786 (gsplit-dwarf): Drop RejectNegative.
1787 (gno-strict-dwarf): Remove.
1788 (gstrict-dwarf): Drop RejectNegative.
1789 * config/darwin.opt (gfull, gused): Add RejectNegative.
1790 * dwarf2out.c (gen_producer_string): Drop
1791 gno-record-gcc-switches handler.
1792 * optc-gen.awk: Add g to prefixes with negative forms.
1793 * opts-common.c (remapping_prefix_p): New.
1794 (find_opt): Check it.
1795 (generate_canonical_option): Test g prefix.
1796 (option_map): Add -gno- mapping.
1797 (add_misspelling_candidates): Check remapping_prefix_p.
1799 2017-09-20 Jeff Law <law@redhat.com>
1801 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1802 thinko in stack clash protection support.
1804 * explow.c (compute_stack_clash_protection_loop_data): Use
1805 CONST_INT_P instead of explicit test. Verify object is a
1806 CONST_INT_P before looking at INTVAL.
1807 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1808 instead of explicit test.
1810 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1813 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1814 address instead of to r1 and r11.
1816 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1818 * config.gcc: Support "knm".
1819 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1820 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1822 * config/i386/i386.c (m_KNM): Define.
1823 (processor_target_table): Add "knm".
1825 (ix86_option_override_internal): Add "knm".
1826 (ix86_issue_rate): Add PROCESSOR_KNM.
1827 (ix86_adjust_cost): Ditto.
1828 (ia32_multipass_dfa_lookahead): Ditto.
1829 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1830 (fold_builtin_cpu): Add M_INTEL_KNM.
1831 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1832 (processor_type): Add PROCESSOR_KNM.
1833 * config/i386/x86-tune.def: Add m_KNM.
1834 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1836 2017-09-20 Richard Biener <rguenther@suse.de>
1838 PR tree-optimization/80213
1839 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1840 are allowed in empty BBs as well.
1841 (canonicalize_loop_closed_ssa): Also look for other complex
1843 (scop_detection::get_sese): Include the loop-closed PHI block
1845 (scop_detection::merge_sese): Remove code adding extra blocks.
1846 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
1847 (build_scops): Assert the final returned scop is invalid.
1849 2017-09-20 Richard Biener <rguenther@suse.de>
1851 PR tree-optimization/82264
1852 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
1854 (vn_phi_lookup): Likewise.
1855 (vn_phi_insert): Likewise.
1857 2017-09-20 Jakub Jelinek <jakub@redhat.com>
1859 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
1860 that fits into uhwi or shwi, add DW_AT_const_value regardless
1861 of early_dwarf without going through RTL, using add_AT_unsigned
1864 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
1865 (DEBUG_LTO_ABBREV_SECTION): Likewise.
1866 (DEBUG_LTO_MACINFO_SECTION): Likewise.
1867 (DEBUG_MACRO_SECTION): Likewise.
1868 (DEBUG_LTO_MACRO_SECTION): Likewise.
1869 (DEBUG_STR_DWO_SECTION): Likewise.
1870 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
1871 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
1872 (DEBUG_LTO_DWO_LINE_SECTION): Define.
1873 (DEBUG_LTO_LINE_STR_SECTION): Define.
1874 (init_sections_and_labels): Initialize debug_line_str_section
1875 variable. Initialize debug_loc_section for -gdwarf-5 to
1876 DEBUG_LOCLISTS_SECTION. Formatting fixes.
1878 2017-09-20 Richard Biener <rguenther@suse.de>
1880 * graphite-sese-to-poly.c (extract_affine): Properly handle
1881 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
1883 2017-09-20 Richard Biener <rguenther@suse.de>
1885 PR tree-optimization/81373
1886 * graphite-scop-detection.c (build_cross_bb_scalars_def):
1887 Force SESE live-out defs to be handled even if they are
1890 2017-09-19 Jeff Law <law@redhat.com>
1892 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
1893 nothing for stack adjustments with REG_STACK_CHECK.
1894 * sched-deps.c (parse_add_or_inc): Reject insns with
1895 REG_STACK_CHECK from dependency breaking.
1896 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
1897 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
1898 * reg-notes.def (STACK_CHECK): New note.
1900 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
1901 (ix86_expand_prologue): Dump stack clash info as needed.
1902 Call ix86_adjust_stack_and_probe_stack_clash as needed.
1904 * function.c (dump_stack_clash_frame_info): New function.
1905 * function.h (dump_stack_clash_frame_info): Prototype.
1906 (enum stack_clash_probes): New enum.
1908 * config/alpha/alpha.c (alpha_expand_prologue): Also check
1909 flag_stack_clash_protection.
1910 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
1911 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
1912 (arm_frame_pointer_required): Likewise.
1913 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
1914 (ia64_expand_prologue): Likewise.
1915 * config/mips/mips.c (mips_expand_prologue): Likewise.
1916 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
1917 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1918 (sparc_flat_expand_prologue): Likewise.
1919 * config/spu/spu.c (spu_expand_prologue): Likewise.
1921 * explow.c: Include "params.h".
1922 (anti_adjust_stack_and_probe_stack_clash): New function.
1923 (get_stack_check_protect): Likewise.
1924 (compute_stack_clash_protection_loop_data): Likewise.
1925 (emit_stack_clash_protection_loop_start): Likewise.
1926 (emit_stack_clash_protection_loop_end): Likewise.
1927 (allocate_dynamic_stack_space): Use get_stack_check_protect.
1928 Use anti_adjust_stack_and_probe_stack_clash.
1929 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
1930 (emit_stack_clash_protection_loop_start): Likewise.
1931 (emit_stack_clash_protection_loop_end): Likewise.
1932 * rtl.h (get_stack_check_protect): Prototype.
1933 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
1934 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
1935 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
1937 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
1939 * doc/tm.texi: Rebuilt.
1940 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
1941 get_stack_check_protect.
1942 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
1943 * config/arm/arm.c (arm_expand_prologue): Likewise.
1944 (arm_frame_pointer_required): Likewise.
1945 * config/i386/i386.c (ix86_expand_prologue): Likewise.
1946 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
1947 * config/mips/mips.c (mips_expand_prologue): Likewise.
1948 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
1949 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
1950 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
1951 (sparc_flat_expand_prologue): Likewise.
1953 * common.opt (-fstack-clash-protection): New option.
1954 * flag-types.h (enum stack_check_type): Note difference between
1955 -fstack-check= and -fstack-clash-protection.
1956 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
1957 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
1958 * toplev.c (process_options): Issue warnings/errors for cases
1959 not handled with -fstack-clash-protection.
1960 * doc/invoke.texi (-fstack-clash-protection): Document new option.
1961 (-fstack-check): Note additional problem with -fstack-check=generic.
1962 Note that -fstack-check is primarily for Ada and refer users
1963 to -fstack-clash-protection for stack-clash-protection.
1964 Document new params for stack clash protection.
1966 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
1968 * config/i386/i386.c (ix86_split_long_move): Do not handle
1969 address used for LEA in a special way.
1971 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
1973 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
1975 2017-09-19 Martin Sebor <msebor@redhat.com>
1978 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
1979 of incompatible types.
1981 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
1983 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1984 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
1985 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1986 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
1988 2017-09-19 Richard Biener <rguenther@suse.de>
1990 PR tree-optimization/82244
1991 * tree-vrp.c (remove_range_assertions): Do not propagate
1992 a constant to abnormals but replace the assert with a copy.
1994 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
1996 PR rtl-optimization/57878
1997 PR rtl-optimization/68988
1998 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
1999 avoidance test involving non_reload_pseudos. Move frequency test
2000 below the general fragmentation avoidance test.
2002 2017-09-19 Richard Biener <rguenther@suse.de>
2004 PR tree-optimization/69728
2005 * graphite-sese-to-poly.c (schedule_error): New global.
2006 (add_loop_schedule): Handle empty domain by failing the
2008 (build_original_schedule): Handle schedule_error.
2010 2017-09-19 Richard Biener <rguenther@suse.de>
2012 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2013 Do not iterate to sibling loops but only to siblings of inner
2016 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2019 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2020 that the registers are different.
2022 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2024 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2025 to processor_model and "amdfam17h" to arch_names_table.
2026 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2028 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2031 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2033 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2034 Alan Hayward <alan.hayward@arm.com>
2035 David Sherwood <david.sherwood@arm.com>
2037 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2039 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2041 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2042 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2043 vect_schedule_slp_instance.
2044 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2045 Update call to vect_slp_analyze_node_operations. Simplify return
2047 (vect_slp_analyze_bb_1): Update call accordingly.
2048 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2049 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2050 (vect_schedule_slp): Update call accordingly.
2052 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2053 Alan Hayward <alan.hayward@arm.com>
2054 David Sherwood <david.sherwood@arm.com>
2056 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2057 with types that aren't in fact scalar.
2059 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2061 * tree-vect-slp.c (vect_record_max_nunits): New function,
2063 (vect_build_slp_tree_1): ...here.
2064 (vect_build_slp_tree_2): Call it for phis too.
2066 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2068 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2069 to vect_get_vec_def_for_operand when getting the mask operand.
2071 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2072 Alan Hayward <alan.hayward@arm.com>
2073 David Sherwood <david.sherwood@arm.com>
2075 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2078 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2079 Alan Hayward <alan.hayward@arm.com>
2080 David Sherwood <david.sherwood@arm.com>
2082 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2083 calculation for vector booleans.
2085 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2086 Alan Hayward <alan.hayward@arm.com>
2087 David Sherwood <david.sherwood@arm.com>
2089 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2091 (vect_transform_stmt): ...here.
2092 (vect_analyze_stmt): Use it instead of calling
2093 vectorizable_live_operation directly.
2095 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2097 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2098 non-SIMT targets in acc vector loops.
2100 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2102 * configure.ac: Add arc and check if assembler supports gdwarf2.
2103 * configure: Regenerate.
2105 2017-09-18 Richard Biener <rguenther@suse.de>
2107 PR tree-optimization/82220
2108 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2109 epilogue niters from the min_profitable_iters compute.
2111 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2114 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2115 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2116 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2118 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2121 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2122 switching to a new text section.
2124 2017-09-18 Richard Biener <rguenther@suse.de>
2126 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2128 (build_alias_set): Reject aliases with no access function.
2130 2017-09-18 Richard Biener <rguenther@suse.de>
2132 PR tree-optimization/79622
2133 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2135 (build_cross_bb_scalars_use): Likewise.
2137 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2139 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2141 2017-09-18 Alan Modra <amodra@gmail.com>
2144 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2145 stack_pointer_rtx for count 0. Update comments. Break up
2146 large rtl expression.
2148 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2150 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2151 Increase to 20 bytes.
2152 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2153 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2154 or avx version of the stub.
2156 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2159 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2160 compute the minimum stack alignment. Also update preferred stack
2161 boundary for leaf functions.
2163 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2165 PR tree-optimization/82228
2166 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2169 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2171 * common/config/nds32/nds32-common.c
2172 (nds32_option_optimization_table): Refine formatting.
2173 (nds32_option_optimization_table): Use -fsched-pressure and
2174 -fomit-frame-pointer for specific optimization level.
2176 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2178 * config/nds32/nds32.c: Refine formatting and comments.
2179 * config/nds32/nds32.h: Likewise.
2180 * config/nds32/nds32.md: Likewise.
2181 * config/nds32/nds32-cost.c: Likewise.
2182 * config/nds32/nds32-isr.c: Likewise.
2183 * config/nds32/nds32-md-auxiliary.c: Likewise.
2184 * config/nds32/nds32-multiple.md: Likewise.
2185 * config/nds32/nds32-predicates.c: Likewise.
2187 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2188 Jakub Jelinek <jakub@redhat.com>
2190 Add support for -std=c++2a.
2191 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2193 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2195 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2198 * doc/extend.texi (Common Function Attributes): Add
2199 references to ARM, AArch64, and S/390 specific attributes.
2200 (Function Specific Option Pragmas): Add AArch64 and S/390
2201 to list of back ends that support the target pragma.
2203 2017-09-15 Nathan Sidwell <nathan@acm.org>
2205 * doc/standards.texi: Fix C++17 description. Update URLs for
2208 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2210 * common.opt (Wcast-align=strict): New warning option.
2211 * doc/invoke.texi: Document -Wcast-align=strict.
2213 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2215 * cgraph.h (cgraph_thunk_info): Add comments.
2216 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2217 assert for VIRTUAL_* arguments stricter.
2219 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2221 PR tree-optimization/71026
2222 * match.pd: Move RDIV patterns from fold-const.c
2223 * fold-const.c (distribute_real_division): Removed.
2224 (fold_binary_loc): Remove calls to distribute_real_divison.
2226 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2228 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2229 c++1z and gnu++1z as deprecated. Change other references to
2230 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2231 Change -Wc++1z-compat to -Wc++17-compat.
2232 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2233 * dwarf2out.c (highest_c_language): Handle C++17.
2234 (gen_compile_unit_die): Likewise.
2236 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2238 PR rtl-optimization/82192
2239 * combine.c (make_extraction): Don't look through non-paradoxical
2240 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2243 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2244 Alan Hayward <alan.hayward@arm.com>
2245 David Sherwood <david.sherwood@arm.com>
2247 * target.def (function_arg_offset): New hook.
2248 * targhooks.h (default_function_arg_offset): Declare.
2249 * targhooks.c (default_function_arg_offset): New function.
2250 * function.c (locate_and_pad_parm): Use
2251 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2252 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2253 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2254 * doc/tm.texi: Regenerate.
2255 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2256 * config/spu/spu.c (spu_function_arg_offset): New function.
2257 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2258 * system.h (FUNCTION_ARG_OFFSET): Poison.
2260 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayard <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * target.def (truly_noop_truncation): New hook.
2265 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2266 than TRULY_NOOP_TRUNCATION.
2267 * hooks.h (hook_bool_uint_uint_true): Declare.
2268 * hooks.c (hook_bool_uint_uint_true): New function.
2269 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2270 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2271 * doc/tm.texi: Regenerate.
2272 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2273 rather than TRULY_NOOP_TRUNCATION in comments.
2274 (simplify_comparison): Likewise.
2275 (record_truncated_value): Likewise.
2276 * expmed.c (extract_bit_field_1): Likewise.
2277 (extract_split_bit_field): Likewise.
2278 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2279 instead of TRULY_NOOP_TRUNCATION.
2280 * function.c (assign_parm_setup_block): Likewise.
2281 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2282 * rtlhooks.c: Include target.h.
2283 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2284 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2285 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2286 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2287 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2288 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2289 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2290 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2291 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2292 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2293 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2294 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2295 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2296 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2297 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2298 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2299 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2300 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2301 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2302 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2303 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2304 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2305 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2306 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2307 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2308 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2309 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2310 rather than TRULY_NOOP_TRUNCATION in comments.
2311 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2312 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2313 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2314 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2315 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2316 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2317 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2318 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2319 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2320 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2321 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2322 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2323 rather than TRULY_NOOP_TRUNCATION in comments.
2324 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2325 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2326 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2327 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2328 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2329 TRULY_NOOP_TRUNCATION condition.
2330 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2331 (TRULY_NOOP_TRUNCATION): Delete.
2332 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2333 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2334 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2335 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2336 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2337 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2338 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2339 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2340 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2341 rather than TRULY_NOOP_TRUNCATION in comments.
2342 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2343 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2344 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2345 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2346 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2347 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2349 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2352 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2353 (*cmp_ior): Likewise.
2354 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2355 (*ior_scc_scc_cmp): Likewise.
2356 (*and_scc_scc): Likewise.
2357 (*and_scc_scc_cmp): Likewise.
2359 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2360 Alan Hayard <alan.hayward@arm.com>
2361 David Sherwood <david.sherwood@arm.com>
2363 * target.def (can_change_mode_class): New hook.
2364 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2365 (hard_regno_nregs): Likewise.
2366 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2367 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2368 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2369 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2370 (LOAD_EXTEND_OP): Update accordingly.
2371 * doc/tm.texi: Regenerate.
2372 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2373 CANNOT_CHANGE_MODE_CLASS.
2374 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2375 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2376 * combine.c (simplify_set): Update accordingly.
2377 * emit-rtl.c (validate_subreg): Likewise.
2378 * recog.c (general_operand): Likewise.
2379 * regcprop.c (mode_change_ok): Likewise.
2380 * reload1.c (choose_reload_regs): Likewise.
2381 (inherit_piecemeal_p): Likewise.
2382 * rtlanal.c (simplify_subreg_regno): Likewise.
2383 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2384 instead of CANNOT_CHANGE_MODE_CLASS.
2385 (reload_cse_simplify_operands): Likewise.
2386 * reload.c (push_reload): Use targetm.can_change_mode_class
2387 instead of CANNOT_CHANGE_MODE_CLASS.
2388 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2389 REG_CANNOT_CHANGE_MODE_P.
2390 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2391 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2392 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2393 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2394 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2395 (arm_can_change_mode_class): New function.
2396 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2397 than CANNOT_CHANGE_MODE_CLASS in comments.
2398 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2399 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2400 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2401 (ix86_can_change_mode_class): ...this new function, inverting the
2402 sense of the return value.
2403 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2404 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2405 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2406 (ia64_can_change_mode_class): New function.
2407 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2408 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2409 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2410 (m32c_can_change_mode_class): ...this new function, inverting the
2411 sense of the return value.
2412 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2413 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2414 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2415 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2416 (mips_can_change_mode_class): ...this new function, inverting the
2417 sense of the return value.
2418 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2419 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2420 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2421 (msp430_can_change_mode_class): New function.
2422 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2423 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2424 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2425 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2426 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2427 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2428 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2429 (pa_cannot_change_mode_class): Replace with...
2430 (pa_can_change_mode_class): ...this new function, inverting the
2431 sense of the return value.
2432 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2433 than CANNOT_CHANGE_MODE_CLASS in comments.
2434 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2435 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2436 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2437 (pdp11_cannot_change_mode_class): Replace with...
2438 (pdp11_can_change_mode_class): ...this new function, inverting the
2439 sense of the return value.
2440 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2441 * config/powerpcspe/powerpcspe-protos.h
2442 (rs6000_cannot_change_mode_class_ptr): Delete.
2443 * config/powerpcspe/powerpcspe.c
2444 (rs6000_cannot_change_mode_class_ptr): Delete.
2445 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2446 (rs6000_option_override_internal): Assign to
2447 targetm.can_change_mode_class instead of
2448 rs6000_cannot_change_mode_class_ptr.
2449 (rs6000_cannot_change_mode_class): Replace with...
2450 (rs6000_can_change_mode_class): ...this new function, inverting the
2451 sense of the return value.
2452 (rs6000_debug_cannot_change_mode_class): Replace with...
2453 (rs6000_debug_can_change_mode_class): ...this new function.
2454 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2455 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2456 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2457 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2458 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2460 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2461 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2462 (rs6000_option_override_internal): Assign to
2463 targetm.can_change_mode_class instead of
2464 rs6000_cannot_change_mode_class_ptr.
2465 (rs6000_cannot_change_mode_class): Replace with...
2466 (rs6000_can_change_mode_class): ...this new function, inverting the
2467 sense of the return value.
2468 (rs6000_debug_cannot_change_mode_class): Replace with...
2469 (rs6000_debug_can_change_mode_class): ...this new function.
2470 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2471 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2472 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2473 (s390_can_change_mode_class): ...this new function, inverting the
2474 sense of the return value.
2475 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2476 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2477 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2478 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2479 (sh_cannot_change_mode_class): Replace with...
2480 (sh_can_change_mode_class): ...this new function, inverting the
2481 sense of the return value.
2482 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2483 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2484 (sparc_can_change_mode_class): New function.
2485 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2486 * config/spu/spu.c (spu_can_change_mode_class): New function.
2487 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2488 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2489 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2490 (visium_can_change_mode_class): New function.
2491 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2493 2017-09-15 Richard Biener <rguenther@suse.de>
2495 PR tree-optimization/82217
2496 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2497 but not undefined case.
2499 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2502 * postreload.c (reload_cse_simplify_operands): Skip
2503 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2505 2017-09-15 Richard Biener <rguenther@suse.de>
2507 PR tree-optimization/68823
2508 * graphite-scop-detection.c (build_alias_set): If we have a
2509 possible dependence check whether we can handle them by just
2510 looking at the DRs DR_ACCESS_FNs.
2511 (build_scops): If build_alias_set fails, fail the SCOP.
2513 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2515 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2516 to support float128 built-in functions that require the ISA 3.0
2518 (BU_FLOAT128_3_HW): Likewise.
2519 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2521 (FMAF128): Likewise.
2523 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2524 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2525 floating point instructions.
2526 (rs6000_invalid_builtin): Likewise.
2527 (rs6000_builtin_mask_names): Likewise.
2528 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2529 (RS6000_BTM_FLOAT128_HW): Likewise.
2530 (RS6000_BTM_COMMON): Likewise.
2531 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2533 * doc/extend.texi (RS/6000 built-in functions): Document the
2534 IEEE 128-bit floating point square root and fused multiply-add
2537 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2539 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2540 reg (r2) isn't in the set of registers defined in the prologue.
2542 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2543 Alan Hayward <alan.hayward@arm.com>
2544 David Sherwood <david.sherwood@arm.com>
2546 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2547 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2548 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2549 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2550 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2552 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2553 max_vectorization_factor.
2554 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2556 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2557 Alan Hayward <alan.hayward@arm.com>
2558 David Sherwood <david.sherwood@arm.com>
2560 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2561 (vect_worthwhile_without_simd_p): Declare.
2562 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2563 (vectorizable_reduction): Use it.
2564 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2565 (vectorizable_operation): Likewise.
2567 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2568 Alan Hayward <alan.hayward@arm.com>
2569 David Sherwood <david.sherwood@arm.com>
2571 * tree-vectorizer.h (vect_get_num_copies): New function.
2572 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2573 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2574 (vectorizable_induction): Likewise.
2575 (vectorizable_live_operation): Likewise.
2576 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2577 (vectorizable_bswap): Likewise.
2578 (vectorizable_call): Likewise.
2579 (vectorizable_conversion): Likewise.
2580 (vectorizable_assignment): Likewise.
2581 (vectorizable_shift): Likewise.
2582 (vectorizable_operation): Likewise.
2583 (vectorizable_store): Likewise.
2584 (vectorizable_load): Likewise.
2585 (vectorizable_condition): Likewise.
2586 (vectorizable_comparison): Likewise.
2587 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2589 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2590 Alan Hayward <alan.hayward@arm.com>
2591 David Sherwood <david.sherwood@arm.com>
2593 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2594 of vect_init_vector.
2596 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2597 Alan Hayward <alan.hayward@arm.com>
2598 David Sherwood <david.sherwood@arm.com>
2600 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2601 an inline wrapper that provides a location.
2602 (gimple_build_vector): Likewise.
2603 * gimple-fold.c (gimple_build_vector_from_val): New function.
2604 (gimple_build_vector): Likewise.
2605 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2606 functions to build the initial value. Always return a gimple value.
2607 (get_initial_defs_for_reduction): Likewise. Only compute
2609 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2610 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2611 (vectorizable_induction): Use gimple_build_vector rather than
2614 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2615 Alan Hayward <alan.hayward@arm.com>
2616 David Sherwood <david.sherwood@arm.com>
2618 * target.h (vec_perm_indices): New typedef.
2619 (auto_vec_perm_indices): Likewise.
2620 * optabs-query.h: Include target.h
2621 (can_vec_perm_p): Take a vec_perm_indices *.
2622 * optabs-query.c (can_vec_perm_p): Likewise.
2623 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2624 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2625 * tree-vect-generic.c (lower_vec_perm): Likewise.
2626 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2627 (vect_grouped_load_supported): Likewise.
2628 (vect_shift_permute_load_chain): Likewise.
2629 (vect_permute_store_chain): Use auto_vec_perm_indices.
2630 (vect_permute_load_chain): Likewise.
2631 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2632 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2633 Update uses of can_vec_perm_p.
2634 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2635 mode with a number of elements. Take a vec_perm_indices *.
2636 (vect_create_epilog_for_reduction): Update accordingly.
2637 Use auto_vec_perm_indices.
2638 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2639 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2640 (vect_transform_slp_perm_load): Likewise.
2641 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2642 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2643 (vect_gen_perm_mask_checked): Likewise.
2644 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2645 (vect_gen_perm_mask_checked): Likewise.
2646 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2647 (vectorizable_store): Likewise.
2648 (vectorizable_load): Likewise.
2649 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2650 (vectorizable_bswap): Likewise.
2652 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2653 Alan Hayward <alan.hayward@arm.com>
2654 David Sherwood <david.sherwood@arm.com>
2656 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2657 * tree.c (build_vector): Likewise.
2658 (build_vector_from_ctor): Update accordingly.
2659 (build_vector_from_val): Likewise.
2660 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2661 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2662 * tree-vect-generic.c (add_rshift): Likewise.
2663 (expand_vector_divmod): Likewise.
2664 (optimize_vector_constructor): Likewise.
2665 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2666 (vect_transform_slp_perm_load): Likewise.
2667 (vect_schedule_slp_instance): Likewise.
2668 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2669 (vectorizable_call): Likewise.
2670 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2671 * expmed.c (make_tree): Likewise.
2672 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2673 a vector passed to build_vector.
2674 (fold_convert_const): Likewise.
2675 (exact_inverse): Likewise.
2676 (fold_ternary_loc): Likewise.
2677 (fold_relational_const): Likewise.
2678 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2679 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2680 (const_unop): Likewise. Store the reduction accumulator in a
2681 variable rather than an array.
2682 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2683 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2684 the new vector, rather than constructing it after the input arrays.
2685 (native_interpret_vector): Use auto_vec<tree> when building
2686 a vector passed to build_vector. Add elements in order.
2687 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2688 auto_vec<tree> when building a vector passed to build_vector.
2689 (vect_create_epilog_for_reduction): Likewise.
2690 (vectorizable_induction): Likewise.
2691 (get_initial_def_for_reduction): Likewise. Fix indentation of
2693 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2695 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2696 passed to build_vector.
2698 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2699 Alan Hayward <alan.hayward@arm.com>
2700 David Sherwood <david.sherwood@arm.com>
2702 * tree-core.h (tree_base::u): Add an "nelts" field.
2703 (tree_vector): Use VECTOR_CST_NELTS as the length.
2704 * tree.c (tree_size): Likewise.
2705 (make_vector): Initialize VECTOR_CST_NELTS.
2706 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2707 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2708 TYPE_VECTOR_SUBPARTS.
2709 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2710 as the number of units, setting it from VECTOR_CST_NELTS.
2711 (const_vector_from_tree): Likewise.
2712 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2713 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2714 (fold_negate_expr_1): Likewise.
2715 (fold_convert_const): Likewise.
2716 (const_binop): Likewise. Differentiate the number of output and
2718 (const_unop): Likewise.
2719 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2720 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2721 in cases that did the opposite.
2723 2017-09-14 Richard Biener <rguenther@suse.de>
2725 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2728 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2730 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2732 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2735 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2736 if and where to split a bb, except for splitting before debug insn
2737 sequences followed by non-label real insn. Delete debug insns
2738 in between basic blocks.
2740 * combine.c (make_compound_operation_int): Formatting fixes.
2742 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2743 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2744 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2745 * config/sol2.h (LINK_EH_SPEC): Likewise.
2746 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2747 * config/s390/linux.h (LINK_SPEC): Likewise.
2748 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2749 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2750 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2751 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2752 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2753 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2754 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2755 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2757 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2759 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2761 (ENDFILE_LINUX_SPEC): Likewise.
2762 (LINK_EH_SPEC): Likewise.
2763 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2764 (LINK_OS_LINUX_SPEC32): Likewise.
2765 (LINK_OS_LINUX_SPEC64): Likewise.
2766 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2767 (LINK_OS_LINUX_SPEC): Likewise.
2769 2017-09-13 Martin Liska <mliska@suse.cz>
2772 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2773 CASE_HIGH is NULL_TREE.
2775 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2776 Alan Hayward <alan.hayward@arm.com>
2777 David Sherwood <david.sherwood@arm.com>
2779 * target.def (secondary_memory_needed): New hook.
2780 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2781 instead of SECONDARY_MEMORY_NEEDED.
2782 (secondary_memory_needed_mode): Likewise.
2783 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2784 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2785 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2786 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2787 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2788 * doc/tm.texi: Regenerate.
2789 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2790 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2791 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2792 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2793 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2794 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2795 mode argument first and change the reg_class arguments to reg_class_t.
2796 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2797 Make static. Update the call to inline_secondary_memory_needed.
2798 (ix86_register_move_cost): Update the call to
2799 inline_secondary_memory_needed.
2800 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2801 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2803 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2804 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2806 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2807 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2808 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2809 and match hook interface. Add comment from mips.h.
2810 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2811 * config/mmix/mmix.md (truncdfsf2): Refer to
2812 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2814 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2815 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2816 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2817 (pa_secondary_memory_needed): New function.
2818 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2819 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2820 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2821 (pdp11_secondary_memory_needed): Make static and match hook interface.
2822 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2823 * config/powerpcspe/powerpcspe-protos.h
2824 (rs6000_secondary_memory_needed_ptr): Delete.
2825 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2827 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2828 (rs6000_option_override_internal): Assign to
2829 targetm.secondary_memory_needed rather than
2830 rs6000_secondary_memory_needed_ptr.
2831 (rs6000_secondary_memory_needed): Match hook interface.
2832 (rs6000_debug_secondary_memory_needed): Likewise.
2833 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2834 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2835 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2836 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2837 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2838 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2840 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
2841 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2842 (rs6000_option_override_internal): Assign to
2843 targetm.secondary_memory_needed rather than
2844 rs6000_secondary_memory_needed_ptr.
2845 (rs6000_secondary_memory_needed): Match hook interface.
2846 (rs6000_debug_secondary_memory_needed): Likewise.
2847 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
2848 * config/s390/s390.c (s390_secondary_memory_needed): New function.
2849 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2850 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
2851 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2852 (sparc_secondary_memory_needed): New function.
2853 * lra-constraints.c (check_and_process_move): Refer to
2854 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2856 (curr_insn_transform): Likewise.
2857 (process_alt_operands): Use targetm.secondary_memory_needed
2858 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2859 (check_secondary_memory_needed_p): Likewise.
2860 (choose_split_class): Likewise.
2861 * reload.c: Unconditionally include code that was previously
2862 conditional on SECONDARY_MEMORY_NEEDED.
2863 (push_secondary_reload): Use targetm.secondary_memory_needed
2864 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2865 (push_reload): Likewise.
2866 * reload1.c: Unconditionally include code that was previously
2867 conditional on SECONDARY_MEMORY_NEEDED.
2868 (choose_reload_regs): Use targetm.secondary_memory_needed
2869 instead of TARGET_SECONDARY_MEMORY_NEEDED.
2870 (gen_reload): Likewise.
2871 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
2873 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2874 Alan Hayward <alan.hayward@arm.com>
2875 David Sherwood <david.sherwood@arm.com>
2877 * target.def (secondary_memory_needed_mode): New hook:
2878 * targhooks.c (default_secondary_memory_needed_mode): Declare.
2879 * targhooks.h (default_secondary_memory_needed_mode): New function.
2880 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
2881 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
2882 * doc/tm.texi: Regenerate.
2883 * lra-constraints.c (check_and_process_move): Use
2884 targetm.secondary_memory_needed_mode instead of
2885 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
2886 (curr_insn_transform): Likewise.
2887 * reload.c (get_secondary_mem): Likewise.
2888 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2889 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
2891 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2892 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2893 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
2894 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2895 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
2897 * config/powerpcspe/powerpcspe-protos.h
2898 (rs6000_secondary_memory_needed_mode): Delete.
2899 * config/powerpcspe/powerpcspe.c
2900 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2901 (rs6000_secondary_memory_needed_mode): Make static.
2902 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2903 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
2905 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2907 (rs6000_secondary_memory_needed_mode): Make static.
2908 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2909 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
2910 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
2911 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
2912 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
2914 (sparc_secondary_memory_needed_mode): New function.
2915 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
2917 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
2919 * config/aarch64/constraints.md (Umq): New constraint.
2920 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
2922 (mov<mode>): Update condition.
2924 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2926 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
2927 when bitposition is the same.
2929 2017-09-13 Richard Biener <rguenther@suse.de>
2931 * dwarf2out.c (output_die_symbol): Remove.
2932 (output_die): Do not output a DIEs symbol.
2934 2017-09-13 Richard Biener <rguenther@suse.de>
2937 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
2938 default-def to avoid breaking iterator update with the weird
2939 interaction with cgraph_update_edges_for_call_stmt_node.
2941 2017-09-13 Richard Biener <rguenther@suse.de>
2943 * tree-cfg.c (verify_gimple_assign_binary): Add verification
2944 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
2945 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
2946 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
2948 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2950 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
2951 Disable pc relative literal load irrespective of
2952 TARGET_FIX_ERR_A53_84341 for default.
2954 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
2956 * config/sparc/sparc.c (output_return): Output the source location of
2957 the insn in the delay slot, if any.
2958 (output_sibcall): Likewise.
2960 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
2963 * common.opt (-static-pie): New alias.
2964 (shared): Negate static-pie.
2965 (-no-pie): Update help text.
2967 (static-pie): New option.
2968 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
2969 -static-pie support.
2970 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2971 (LINK_EH_SPEC): Likewise.
2972 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2973 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2974 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
2975 * gcc.c (LINK_COMMAND_SPEC): Likewise.
2976 (init_gcc_specs): Likewise.
2977 (init_spec): Likewise.
2978 (display_help): Update help message for -pie.
2979 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
2982 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
2984 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
2985 (movdi_aarch64): Likewise.
2986 (movti_aarch64): Likewise.
2988 2017-09-12 Simon Wright <simon@pushface.org>
2991 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
2992 calculation of the minor version, always output as 0.
2994 2017-09-12 Jakub Jelinek <jakub@redhat.com>
2997 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
2998 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
2999 on it early, rather than manual conversion late. For
3000 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3001 instead of performing manual conversion.
3003 2017-09-12 Carl Love <cel@us.ibm.com>
3005 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3006 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3008 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3009 VMULOSW): Add definitions.
3010 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3011 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3012 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3013 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3014 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3016 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3018 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3020 (movti_aarch64): Likewise.
3021 (movdf_aarch64): Likewise.
3022 (movtf_aarch64): Likewise.
3023 (load_pairdi): Likewise.
3024 (store_pairdi): Likewise.
3025 (load_pairdf): Likewise.
3026 (store_pairdf): Likewise.
3027 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3028 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3029 (ldr_got_small_<mode>): Likewise.
3030 (ldr_got_small_28k_<mode>): Likewise.
3031 (ldr_got_tiny): Likewise.
3032 * config/aarch64/iterators.md (ldst_sz): New.
3033 (ldpstp_sz): Likewise.
3034 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3036 (thunderx_load): Split load_8 to load_16.
3037 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3039 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3040 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3041 (xgene1_store_pair): Split store_8 to store_16.
3042 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3043 (falkor_st_0_st_sd): Split store_8 to store_16.
3045 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3047 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3048 and store1/2/3/4 to store_4/8/12/16.
3049 * config/aarch64/aarch64.md: Update for rename.
3050 * config/arm/arm.md: Likewise.: Likewise.
3051 * config/arm/arm.c: Likewise.
3052 * config/arm/thumb1.md: Likewise.
3053 * config/arm/thumb2.md: Likewise.
3054 * config/arm/vfp.md: Likewise.
3055 * config/arm/arm-generic.md: Likewise.
3056 * config/arm/arm1020e.md: Likewise.
3057 * config/arm/arm1026ejs.md: Likewise.
3058 * config/arm/arm1136jfs.md: Likewise.
3059 * config/arm/arm926ejs.md: Likewise.
3060 * config/arm/cortex-a15.md: Likewise.
3061 * config/arm/cortex-a17.md: Likewise.
3062 * config/arm/cortex-a5.md: Likewise.
3063 * config/arm/cortex-a53.md: Likewise.
3064 * config/arm/cortex-a57.md: Likewise.
3065 * config/arm/cortex-a7.md: Likewise.
3066 * config/arm/cortex-a8.md: Likewise.
3067 * config/arm/cortex-a9.md: Likewise.
3068 * config/arm/cortex-m4.md: Likewise.
3069 * config/arm/cortex-m7.md: Likewise.
3070 * config/arm/cortex-r4.md: Likewise.
3071 * config/arm/exynos-m1.md: Likewise.
3072 * config/arm/fa526.md: Likewise.
3073 * config/arm/fa606te.md: Likewise.
3074 * config/arm/fa626te.md: Likewise.
3075 * config/arm/fa726te.md: Likewise.
3076 * config/arm/fmp626.md: Likewise.
3077 * config/arm/iwmmxt.md: Likewise.
3078 * config/arm/ldmstm.md: Likewise.
3079 * config/arm/marvell-pj4.md: Likewise.
3080 * config/arm/xgene1.md: Likewise.
3081 * config/aarch64/thunderx.md: Likewise.
3082 * config/aarch64/thunderx2t99.md: Likewise.
3083 * config/aarch64/falkor.md: Likewise.
3085 2017-09-12 Martin Liska <mliska@suse.cz>
3087 * attribs.c (private_lookup_attribute): New function.
3088 * attribs.h (private_lookup_attribute): Declared here.
3089 (lookup_attribute): Called from this place.
3091 2017-09-12 Richard Biener <rguenther@suse.de>
3093 PR tree-optimization/82157
3094 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3095 stmts with side-effects.
3097 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3098 Alan Hayward <alan.hayward@arm.com>
3099 David Sherwood <david.sherwood@arm.com>
3101 * target.def (hard_regno_nregs): New hook.
3102 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3103 * targhooks.h (default_hard_regno_nregs): Declare.
3104 * targhooks.c (default_hard_regno_nregs): New function.
3105 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3106 (TARGET_HARD_REGNO_NREGS): ...this hook.
3107 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3108 (CLASS_MAX_NREGS): Likewise.
3109 * doc/tm.texi: Regenerate.
3110 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3111 instead of HARD_REGNO_NREGS.
3112 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3113 HARD_REGNO_NREGS in the comment.
3114 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3115 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3116 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3117 Return an unsigned int.
3118 (TARGET_HARD_REGNO_NREGS): Redefine.
3119 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3120 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3121 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3122 (arc_hard_regno_nregs): New function.
3123 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3124 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3125 (arm_hard_regno_nregs): New function.
3126 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3127 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3128 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3129 (TARGET_HARD_REGNO_NREGS): Redefine.
3130 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3131 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3132 (HARD_REGNO_NREGS): Delete.
3133 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3134 (cr16_hard_regno_nregs): New function.
3135 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3136 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3137 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3138 (cris_hard_regno_nregs): New function.
3139 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3140 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3141 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3142 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3143 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3144 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3145 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3146 (frv_hard_regno_nregs): Make static. Take and return an
3148 (frv_class_max_nregs): Remove outdated copy of documentation.
3149 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3150 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3151 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3152 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3153 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3154 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3155 (TARGET_HARD_REGNO_NREGS): Redefine.
3156 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3157 (CLASS_MAX_NREGS): Update comment.
3158 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3159 (ia64_hard_regno_nregs): New function.
3160 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3161 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3162 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3163 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3164 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3166 (m32c_hard_regno_nregs): Likewise. Make static.
3167 (TARGET_HARD_REGNO_NREGS): Redefine.
3168 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3169 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3170 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3171 (m68k_hard_regno_nregs): New function.
3172 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3173 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3174 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3175 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3176 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3177 Take and return an unsigned int.
3178 (TARGET_HARD_REGNO_NREGS): Redefine.
3179 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3180 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3181 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3182 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3183 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3184 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3185 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3186 (msp430_hard_regno_nregs): Make static. Take and return an
3188 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3189 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3190 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3191 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3192 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3193 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3194 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3195 (TARGET_HARD_REGNO_NREGS): Redefine.
3196 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3197 (PA_HARD_REGNO_NREGS): ...this.
3198 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3199 (PA_HARD_REGNO_NREGS): ...this.
3200 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3201 (pa_hard_regno_nregs): New function.
3202 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3203 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3204 (pdp11_hard_regno_nregs): New function.
3205 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3206 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3207 (rs6000_hard_regno_nregs_hook): New function.
3208 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3209 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3210 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3211 Take and return an unsigned int. Move earlier in file.
3212 (TARGET_HARD_REGNO_NREGS): Redefine.
3213 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3214 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3215 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3216 (rl78_hard_regno_nregs): Make static. Take and return an
3218 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3219 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3220 (rs6000_hard_regno_nregs_hook): New function.
3221 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3222 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3223 (TARGET_HARD_REGNO_NREGS): Redefine.
3224 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3225 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3226 instead of HARD_REGNO_NREGS.
3227 (s390_hard_regno_nregs): New function.
3228 (s390_hard_regno_mode_ok): Add comment from s390.h.
3229 (TARGET_HARD_REGNO_NREGS): Redefine.
3230 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3231 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3232 (sh_hard_regno_nregs): New function.
3233 (sh_pass_in_reg_p): Use it.
3234 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3235 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3236 (sparc_hard_regno_nregs): New function.
3237 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3238 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3239 (spu_function_arg_advance): Use it, supplying a valid register number.
3240 (TARGET_HARD_REGNO_NREGS): Redefine.
3241 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3242 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3243 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3244 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3245 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3246 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3247 (CLASS_MAX_NREGS): Remove copy of old documentation.
3248 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3249 (visium_hard_regno_nregs): New function.
3250 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3251 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3252 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3253 xtensa_hard_regno_nregs): New function.
3254 * system.h (HARD_REGNO_NREGS): Poison.
3256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3258 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3259 hard_regno_nregs instead of HARD_REGNO_NREGS.
3260 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3261 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3262 (c6x_expand_epilogue): Likewise.
3263 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3264 (frv_read_iacc_argument): Likewise.
3265 * config/sh/sh.c: Include regs.h.
3266 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3267 (regs_used): Likewise.
3268 (output_stack_adjust): Likewise.
3269 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3270 * expmed.c: Include regs.h.
3271 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3272 * ree.c: Include regs.h.
3273 (combine_reaching_defs): Use hard_regno_nregs instead of
3275 (add_removable_extension): Likewise.
3277 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3279 * regs.h (hard_regno_nregs): Turn into a function.
3280 (end_hard_regno): Update accordingly.
3281 * caller-save.c (setup_save_areas): Likewise.
3282 (save_call_clobbered_regs): Likewise.
3283 (replace_reg_with_saved_mem): Likewise.
3284 (insert_restore): Likewise.
3285 (insert_save): Likewise.
3286 * combine.c (can_change_dest_mode): Likewise.
3287 (move_deaths): Likewise.
3288 (distribute_notes): Likewise.
3289 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3290 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3291 (rs6000_split_multireg_move): Likewise.
3292 (rs6000_register_move_cost): Likewise.
3293 (rs6000_memory_move_cost): Likewise.
3294 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3295 (rs6000_split_multireg_move): Likewise.
3296 (rs6000_register_move_cost): Likewise.
3297 (rs6000_memory_move_cost): Likewise.
3298 * cselib.c (cselib_reset_table): Likewise.
3299 (cselib_lookup_1): Likewise.
3300 * emit-rtl.c (set_mode_and_regno): Likewise.
3301 * function.c (aggregate_value_p): Likewise.
3302 * ira-color.c (setup_profitable_hard_regs): Likewise.
3303 (check_hard_reg_p): Likewise.
3304 (calculate_saved_nregs): Likewise.
3305 (assign_hard_reg): Likewise.
3306 (improve_allocation): Likewise.
3307 (calculate_spill_cost): Likewise.
3308 * ira-emit.c (modify_move_list): Likewise.
3309 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3310 (ira_hard_reg_in_set_p): Likewise.
3311 * ira.c (setup_reg_mode_hard_regset): Likewise.
3312 (clarify_prohibited_class_mode_regs): Likewise.
3313 (check_allocation): Likewise.
3314 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3315 (lra_setup_reg_renumber): Likewise.
3316 (setup_try_hard_regno_pseudos): Likewise.
3317 (spill_for): Likewise.
3318 (assign_hard_regno): Likewise.
3319 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3320 * lra-constraints.c (in_class_p): Likewise.
3321 (lra_constraint_offset): Likewise.
3322 (simplify_operand_subreg): Likewise.
3323 (lra_constraints): Likewise.
3324 (split_reg): Likewise.
3325 (split_if_necessary): Likewise.
3326 (invariant_p): Likewise.
3327 (inherit_in_ebb): Likewise.
3328 * lra-lives.c (process_bb_lives): Likewise.
3329 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3330 (get_hard_regs): Likewise.
3331 (do_remat): Likewise.
3332 * lra-spills.c (assign_spill_hard_regs): Likewise.
3333 * mode-switching.c (create_pre_exit): Likewise.
3334 * postreload.c (reload_combine_recognize_pattern): Likewise.
3335 * recog.c (peep2_find_free_register): Likewise.
3336 * regcprop.c (kill_value_regno): Likewise.
3337 (set_value_regno): Likewise.
3338 (copy_value): Likewise.
3339 (maybe_mode_change): Likewise.
3340 (find_oldest_value_reg): Likewise.
3341 (copyprop_hardreg_forward_1): Likewise.
3342 * regrename.c (check_new_reg_p): Likewise.
3343 (regrename_do_replace): Likewise.
3344 * reload.c (push_reload): Likewise.
3345 (combine_reloads): Likewise.
3346 (find_dummy_reload): Likewise.
3347 (operands_match_p): Likewise.
3348 (find_reloads): Likewise.
3349 (find_equiv_reg): Likewise.
3350 (reload_adjust_reg_for_mode): Likewise.
3351 * reload1.c (count_pseudo): Likewise.
3352 (count_spilled_pseudo): Likewise.
3353 (find_reg): Likewise.
3354 (clear_reload_reg_in_use): Likewise.
3355 (free_for_value_p): Likewise.
3356 (allocate_reload_reg): Likewise.
3357 (choose_reload_regs): Likewise.
3358 (reload_adjust_reg_for_temp): Likewise.
3359 (emit_reload_insns): Likewise.
3360 (delete_output_reload): Likewise.
3361 * rtlanal.c (subreg_get_info): Likewise.
3362 * sched-deps.c (sched_analyze_reg): Likewise.
3363 * sel-sched.c (init_regs_for_mode): Likewise.
3364 (mark_unavailable_hard_regs): Likewise.
3365 (choose_best_reg_1): Likewise.
3366 (verify_target_availability): Likewise.
3367 * valtrack.c (dead_debug_insert_temp): Likewise.
3368 * var-tracking.c (track_loc_p): Likewise.
3369 (emit_note_insn_var_location): Likewise.
3370 * varasm.c (make_decl_rtl): Likewise.
3371 * reginfo.c (choose_hard_reg_mode): Likewise.
3372 (init_reg_modes_target): Refer directly to
3373 this_target_regs->x_hard_regno_nregs.
3375 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3377 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3378 instead of hard_regno_nregs.
3380 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3382 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3383 end_hard_regno instead of hard_regno_nregs.
3384 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3385 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3386 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3387 * ira-color.c (improve_allocation): Likewise.
3388 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3389 * lra-lives.c (mark_regno_live): Likewise.
3390 (mark_regno_dead): Likewise.
3391 * lra-remat.c (operand_to_remat): Likewise.
3392 * lra.c (collect_non_operand_hard_regs): Likewise.
3393 * postreload.c (reload_combine_note_store): Likewise.
3394 (move2add_valid_value_p): Likewise.
3395 * reload.c (regno_clobbered_p): Likewise.
3397 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3399 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3401 * config/v850/v850.c (v850_reorg): Likewise.
3402 * reload.c (refers_to_regno_for_reload_p): Likewise.
3403 (find_equiv_reg): Likewise.
3404 * reload1.c (reload_reg_reaches_end_p): Likewise.
3406 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3408 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3410 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3411 * config/arm/arm.c (output_move_neon): Likewise.
3412 (arm_attr_length_move_neon): Likewise.
3413 (neon_split_vcombine): Likewise.
3414 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3415 (c6x_mark_reg_written): Likewise.
3416 (c6x_dwarf_register_span): Likewise.
3417 * config/i386/i386.c (ix86_save_reg): Likewise.
3418 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3419 (rws_access_reg): Likewise.
3420 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3421 * mode-switching.c (create_pre_exit): Likewise.
3422 * ree.c (combine_reaching_defs): Likewise.
3423 (add_removable_extension): Likewise.
3424 * regcprop.c (find_oldest_value_reg): Likewise.
3425 (copyprop_hardreg_forward_1): Likewise.
3426 * reload.c (reload_inner_reg_of_subreg): Likewise.
3427 (push_reload): Likewise.
3428 (combine_reloads): Likewise.
3429 (find_dummy_reload): Likewise.
3430 (reload_adjust_reg_for_mode): Likewise.
3431 * reload1.c (find_reload_regs): Likewise.
3432 (forget_old_reloads_1): Likewise.
3433 (reload_reg_free_for_value_p): Likewise.
3434 (reload_adjust_reg_for_temp): Likewise.
3435 (emit_reload_insns): Likewise.
3436 (delete_output_reload): Likewise.
3437 * sel-sched.c (choose_best_reg_1): Likewise.
3438 (choose_best_pseudo_reg): Likewise.
3440 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3441 Alan Hayward <alan.hayward@arm.com>
3442 David Sherwood <david.sherwood@arm.com>
3444 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3445 * target.def (slow_unaligned_access): New hook.
3446 * targhooks.h (default_slow_unaligned_access): Declare.
3447 * targhooks.c (default_slow_unaligned_access): New function.
3448 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3449 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3450 * doc/tm.texi: Regenerate.
3451 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3452 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3453 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3455 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3456 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3458 (rs6000_slow_unaligned_access): New function.
3459 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3460 (expand_block_compare): Likewise.
3461 (expand_strn_compare): Likewise.
3462 (rs6000_rtx_costs): Likewise.
3463 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3464 (riscv_slow_unaligned_access): Likewise.
3465 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3466 (riscv_slow_unaligned_access_p): ...this and make static.
3467 (riscv_option_override): Update accordingly.
3468 (riscv_slow_unaligned_access): New function.
3469 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3470 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3471 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3472 (rs6000_slow_unaligned_access): New function.
3473 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3474 (rs6000_rtx_costs): Likewise.
3475 * config/rs6000/rs6000-string.c (expand_block_compare)
3476 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3477 of SLOW_UNALIGNED_ACCESS.
3478 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3479 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3480 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3481 of SLOW_UNALIGNED_ACCESS.
3482 * expmed.c (simple_mem_bitfield_p): Likewise.
3483 * expr.c (alignment_for_piecewise_move): Likewise.
3484 (emit_group_load_1): Likewise.
3485 (emit_group_store): Likewise.
3486 (copy_blkmode_from_reg): Likewise.
3487 (emit_push_insn): Likewise.
3488 (expand_assignment): Likewise.
3489 (store_field): Likewise.
3490 (expand_expr_real_1): Likewise.
3491 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3492 * lra-constraints.c (simplify_operand_subreg): Likewise.
3493 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3494 * gimple-ssa-store-merging.c: Likewise in block comment at start
3496 * tree-ssa-strlen.c: Include target.h.
3497 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3498 of SLOW_UNALIGNED_ACCESS.
3499 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3501 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3503 PR rtl-optimization/82185
3504 * expmed.c (emit_store_flag_int): Only test tem if it has been
3507 2017-09-12 Richard Biener <rguenther@suse.de>
3510 * match.pd ((FTYPE) N CMP CST): Fix typo.
3512 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3514 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3516 (mips_near_type_p): Add 'short_call' attribute as a synonym
3518 * doc/extend.texi (short_call): Document new function attribute.
3520 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3523 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3524 assertion check that in the condition.
3525 (get_atomic_generic_size): Likewise. Before testing if parameter
3526 has pointer type, if it has array type, call for C++
3527 default_conversion to perform array-to-pointer conversion.
3529 2017-09-12 Richard Biener <rguenther@suse.de>
3531 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3532 for operations we cannot scalarize.
3534 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3536 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3537 vectors heap vectors. Clean up comments.
3538 Make visited_bbs a reference.
3539 (profitable_jump_thread_path): Make GC
3540 vectors heap vectors. Clean up comments.
3542 (convert_and_register_jump_thread_path): Make GC vectors heap
3544 (check_subpath_and_update_thread_path): Same. Clean up comments.
3545 Make visited_bbs a reference.
3546 (handle_phi): Abstract common code to to
3547 register_jump_thread_path_if_profitable.
3548 Rename VAR_BB to DEF_BB.
3550 Make GC vectors heap vectors.
3551 Make visited_bbs a reference.
3552 (handle_assignment): Same.
3553 (register_jump_thread_path_if_profitable): New.
3554 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3556 Make GC vectors heap vectors. Clean up comments.
3557 Make visited_bbs a reference.
3558 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3559 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3562 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3565 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3566 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3568 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3570 Revert r251800 and r251799.
3572 2017-09-11 Martin Jambor <mjambor@suse.cz>
3575 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3576 arguments in advance.
3577 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3578 use it to find predecessor edges.
3579 (naive_outof_ssa): Collect vector of predecessors.
3581 2017-09-08 Jason Merrill <jason@redhat.com>
3583 PR c++/70029 - ICE with ref-qualifier and -flto
3584 * langhooks.h (struct lang_hooks_for_types): Add
3585 copy_lang_qualifiers.
3586 * attribs.c (build_type_attribute_qual_variant): Use it.
3587 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3589 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3590 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3592 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3595 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3596 (*mulsi3_sp64): New instruction.
3597 (mulsi3): New expander.
3599 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3601 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3603 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3605 * sancov.c: Include memmodel.h.
3607 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3610 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3613 2017-09-07 Carl Love <cel@us.ibm.com>
3615 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3616 the sldi instruction.
3618 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3620 * sancov.c: Include tm_p.h.
3622 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3625 * output.h (switch_to_other_text_partition): New declaration.
3626 * varasm.c (switch_to_other_text_partition): New function.
3627 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3628 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3629 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3630 to the other text partition before emitting LCL label and switch back
3631 after emitting the word after it.
3633 2017-09-07 Richard Biener <rguenther@suse.de>
3635 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3636 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3637 critical edges here, after loop init.
3638 (pass_data_pre): Remove PROP_no_crit_edges flags.
3639 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3640 for valueization of call args to avoid leaking VN_TOP.
3641 (visit_use): Assert we do not visit default defs.
3642 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3643 Use error_mark_node to more easily detect leaking VN_TOP.
3644 All default-defs are varying, not VN_TOP. Mark them visited.
3645 (run_scc_vn): Make code match comment.
3647 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3649 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3650 OPTION_MASK_FLOAT128_KEYWORD.
3651 (POWERPC_MASKS): Likewise.
3652 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3653 support for the -mfloat128-type option, and make -mfloat128
3654 default on PowerPC Linux systems. Define or undefine
3655 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3656 Define __float128 to be __ieee128 if IEEE 128-bit support is
3657 enabled, or undefine it.
3658 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3659 Delete defining __FLOAT128_TYPE__.
3660 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3661 -mfloat128-type option and make -mfloat128 default on PowerPC
3663 (TARGET_FLOAT128_TYPE): Likewise.
3664 (-mfloat128-type): Likewise.
3665 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3666 Delete the -mfloat128-type option and make -mfloat128 default on
3667 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3668 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3669 128-bit floating point is enabled. Change tests from using
3670 -mfloat128-type to -mfloat128.
3671 (rs6000_mangle_type): Use the correct mangling for the __float128
3672 type even if normal long double is restricted to 64-bits.
3673 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3675 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3676 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3677 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3679 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3680 documentation for -mfloat128.
3682 2017-09-06 Olivier Hainque <hainque@adacore.com>
3684 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3686 2017-09-06 Wish Wu <wishwu007@gmail.com>
3687 Jakub Jelinek <jakub@redhat.com>
3689 * asan.c (initialize_sanitizer_builtins): Add
3690 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3691 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3692 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3693 BT_FN_VOID_UINT64_PTR variables.
3694 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3695 (BT_FN_VOID_UINT16_UINT16): Likewise.
3696 (BT_FN_VOID_UINT32_UINT32): Likewise.
3697 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3698 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3699 (BT_FN_VOID_UINT64_PTR): Likewise.
3700 * common.opt (flag_sanitize_coverage): New variable.
3701 (fsanitize-coverage=trace-pc): Remove.
3702 (fsanitize-coverage=): Add.
3703 * flag-types.h (enum sanitize_coverage_code): New enum.
3704 * fold-const.c (fold_range_test): Disable non-short-circuit
3705 optimization if flag_sanitize_coverage.
3706 (fold_truth_andor): Likewise.
3707 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3708 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3709 (coverage_sanitizer_opts): New array.
3710 (get_closest_sanitizer_option): Add OPTS argument, handle also
3711 OPT_fsanitize_coverage_.
3712 (parse_sanitizer_options): Adjusted to also handle
3713 OPT_fsanitize_coverage_.
3714 (common_handle_option): Add OPT_fsanitize_coverage_.
3715 * sancov.c (instrument_comparison, instrument_switch): New function.
3716 (sancov_pass): Add trace-cmp support.
3717 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3718 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3719 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3720 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3721 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3722 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3723 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3724 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3725 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3726 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3728 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3730 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3731 error. Only quit immediately if parsing is complete.
3732 (BEGIN): Initialize fatal_err and parse_done.
3733 (begin fpu, end fpu): Check number of arguments.
3734 (begin arch, end arch): Likewise.
3735 (begin cpu, end cpu): Likewise.
3736 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3737 (optalias): Likewise.
3739 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3741 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3742 * config/arm/arm-isa.h: Delete. Move definitions to ...
3743 * arm-cpus.in: ... here. Use new feature and fgroup values.
3744 * config/arm/arm.c (arm_option_override): Use lower case for feature
3746 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3747 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3748 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3750 (print_isa_bits_for): New function.
3751 (gen_isa): New function.
3752 (gen_comm_data): Use print_isa_bits_for.
3753 (define feature): New keyword.
3754 (define fgroup): New keyword.
3755 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3756 (arm-isa.h): Add rule to generate file.
3757 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3758 case for feature bit names.
3760 2017-09-06 Richard Biener <rguenther@suse.de>
3762 * tree-ssa-pre.c (NECESSARY): Remove.
3763 (create_expression_by_pieces): Do not touch pass-local flags.
3764 (insert_into_preds_of_block): Likewise.
3765 (do_pre_regular_insertion): Likewise.
3766 (eliminate_insert): Likewise.
3767 (eliminate_dom_walker::before_dom_children): Likewise.
3768 (fini_eliminate): Do not look at inserted_exprs.
3769 (mark_operand_necessary): Remove.
3770 (remove_dead_inserted_code): Replace with simple work-list
3771 algorithm based on inserted_exprs and SSA uses.
3772 (pass_pre::execute): Re-order fini_eliminate and
3773 remove_dead_inserted_code.
3775 2017-09-06 Olivier Hainque <hainque@adacore.com>
3777 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3778 for VxWorks 7. Adjust surrounding comments.
3780 2017-09-06 Richard Biener <rguenther@suse.de>
3782 * gimple-ssa-strength-reduction.c
3783 (find_candidates_dom_walker::before_dom_children): Also allow
3786 2017-09-06 Richard Biener <rguenther@suse.de>
3788 PR tree-optimization/82108
3789 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3790 for gap in the non-permutation SLP case.
3792 2017-09-06 Martin Jambor <mjambor@suse.cz>
3794 PR tree-optimization/82078
3795 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3796 add_access_to_work_queue...
3797 (build_accesses_from_assign): ...here.
3798 (propagate_all_subaccesses): Make sure racc is the group
3799 representative, if there is one.
3801 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3804 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3807 2017-09-06 Richard Biener <rguenther@suse.de>
3809 * gimple-ssa-strength-reduction.c
3810 (find_candidates_dom_walker::before_doom_children): Use a
3811 type and not a mode check.
3813 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3816 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3818 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3820 2017-09-05 Jeff Law <law@redhat.com>
3822 PR tree-optimization/64910
3823 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3824 cases where we have 3 or more operands.
3826 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3829 * omp-low.c (lower_omp_for): Recompute tree invariant if
3830 gimple_omp_for_initial/final is ADDR_EXPR.
3833 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3834 into gimple val before gimplification fo the COND_EXPR.
3836 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3838 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3839 REGION_COPY argument.
3840 (thread_through_all_blocks): Remove unused argument to
3841 duplicate_thread_path.
3843 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3844 Alan Hayward <alan.hayward@arm.com>
3845 David Sherwood <david.sherwood@arm.com>
3847 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
3848 Take a scalar_mode rather than a machine_mode.
3849 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3850 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
3851 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3852 (aarch64_gen_adjusted_ldpstp): Likewise.
3853 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
3855 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3856 Alan Hayward <alan.hayward@arm.com>
3857 David Sherwood <david.sherwood@arm.com>
3859 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
3860 Take a scalar_int_mode instead of a machine_mode.
3861 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3862 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3863 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3864 (aarch64_simd_attr_length_rglist): Delete.
3865 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
3866 a scalar_int_mode instead of a machine_mode.
3867 (aarch64_add_offset): Likewise.
3868 (aarch64_internal_mov_immediate): Likewise
3869 (aarch64_add_constant_internal): Likewise.
3870 (aarch64_add_constant): Likewise.
3871 (aarch64_movw_imm): Likewise.
3872 (aarch64_rtx_arith_op_extract_p): Likewise.
3873 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
3874 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
3875 Remove assert that the mode isn't a vector.
3876 (aarch64_output_scalar_simd_mov_immediate): Likewise.
3877 (aarch64_expand_mov_immediate): Update calls after above changes.
3878 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
3879 (aarch64_and_bitmask_imm): Check for scalar integer modes.
3880 (aarch64_move_imm): Likewise.
3881 (aarch64_can_const_movi_rtx_p): Likewise.
3882 (aarch64_strip_extend): Likewise.
3883 (aarch64_extr_rtx_p): Likewise.
3884 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
3885 a CONST_INT when the mode parameter is VOIDmode.
3886 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
3888 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3890 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
3891 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3892 (bitwise_type_for_mode): Update accordingly.
3894 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3896 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
3897 * stor-layout.c (mode_for_size_tree): Likewise.
3898 (mode_for_array): Update accordingly.
3899 (layout_decl): Likewise.
3900 (compute_record_mode): Likewise. Only set the mode once.
3902 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3904 * target.def (get_mask_mode): Change return type to opt_mode.
3906 * doc/tm.texi: Regenerate.
3907 * targhooks.h (default_get_mask_mode): Return an opt_mode.
3908 * targhooks.c (default_get_mask_mode): Likewise.
3909 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3910 * optabs-query.c (can_vec_mask_load_store_p): Update use of
3911 targetm.get_mask_mode.
3912 * tree.c (build_truth_vector_type): Likewise.
3914 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3916 * machmode.h (mode_for_vector): Return an opt_mode.
3917 * stor-layout.c (mode_for_vector): Likewise.
3918 (mode_for_int_vector): Update accordingly.
3919 (layout_type): Likewise.
3920 * config/i386/i386.c (emit_memmov): Likewise.
3921 (ix86_expand_set_or_movmem): Likewise.
3922 (ix86_expand_vector_init): Likewise.
3923 (ix86_get_mask_mode): Likewise.
3924 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
3926 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
3927 * expmed.c (extract_bit_field_1): Likewise.
3928 * expr.c (expand_expr_real_2): Likewise.
3929 * optabs-query.c (can_vec_perm_p): Likewise.
3930 (can_vec_mask_load_store_p): Likewise.
3931 * optabs.c (expand_vec_perm): Likewise.
3932 * targhooks.c (default_get_mask_mode): Likewise.
3933 * tree-vect-stmts.c (vectorizable_store): Likewise.
3934 (vectorizable_load): Likewise.
3935 (get_vectype_for_scalar_type_and_size): Likewise.
3937 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3939 * machmode.h (mode_for_int_vector): New function.
3940 * stor-layout.c (mode_for_int_vector): Likewise.
3941 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
3942 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
3943 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3944 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3945 (s390_expand_vcond): Likewise.
3947 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3949 * machmode.h (opt_machine_mode): New type.
3950 (opt_mode<T>): Allow construction from anything that can be
3952 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
3953 (mode_for_size): Return an opt_machine_mode.
3954 * stor-layout.c (mode_for_size): Likewise.
3955 (mode_for_size_tree): Update call accordingly.
3956 (bitwise_mode_for_mode): Likewise.
3957 (make_fract_type): Likewise.
3958 (make_accum_type): Likewise.
3959 * caller-save.c (replace_reg_with_saved_mem): Update call
3961 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3962 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3963 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3964 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
3965 * expmed.c (extract_bit_field_1): Likewise.
3966 * reload.c (get_secondary_mem): Likewise.
3967 * varasm.c (assemble_integer): Likewise.
3968 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
3971 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3973 * machmode.h (decimal_float_mode_for_size): New function.
3974 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
3975 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
3976 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
3977 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
3978 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
3979 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
3981 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3983 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
3984 (get_builtin_sync_mode): Likewise.
3985 (expand_ifn_atomic_compare_exchange): Likewise.
3986 (expand_builtin_atomic_clear): Likewise.
3987 (expand_builtin_atomic_test_and_set): Likewise.
3988 (fold_builtin_atomic_always_lock_free): Likewise.
3989 * calls.c (compute_argument_addresses): Likewise.
3990 (emit_library_call_value_1): Likewise.
3991 (store_one_arg): Likewise.
3992 * combine.c (combine_instructions): Likewise.
3993 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
3994 * config/arm/arm.c (arm_function_value): Likewise.
3995 (aapcs_allocate_return_reg): Likewise.
3996 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
3997 * config/i386/i386.c (construct_container): Likewise.
3998 (ix86_gimplify_va_arg): Likewise.
3999 (ix86_expand_sse_cmp): Likewise.
4000 (emit_memmov): Likewise.
4001 (emit_memset): Likewise.
4002 (expand_small_movmem_or_setmem): Likewise.
4003 (ix86_expand_pextr): Likewise.
4004 (ix86_expand_pinsr): Likewise.
4005 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4006 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4008 * config/mips/mips.c (mips_function_value_1) Likewise.
4009 (mips_block_move_straight): Likewise.
4010 (mips_expand_ins_as_unaligned_store): Likewise.
4011 * config/powerpcspe/powerpcspe.c
4012 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4013 (rs6000_darwin64_record_arg_flush): Likewise.
4014 * config/rs6000/rs6000.c
4015 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4016 (rs6000_darwin64_record_arg_flush): Likewise.
4017 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4018 (sparc_function_value_1): Likewise.
4019 * config/spu/spu.c (adjust_operand): Likewise.
4020 (spu_emit_branch_or_set): Likewise.
4021 (arith_immediate_p): Likewise.
4022 * emit-rtl.c (gen_lowpart_common): Likewise.
4023 * expr.c (expand_expr_real_1): Likewise.
4024 * function.c (assign_parm_setup_block): Likewise.
4025 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4026 * reload1.c (alter_reg): Likewise.
4027 * stor-layout.c (mode_for_vector): Likewise.
4028 (layout_type): Likewise.
4030 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4032 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4033 (spu_convert_move): Likewise.
4034 * lower-subreg.c (resolve_simple_move): Likewise.
4036 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4039 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4040 define_insn to a define_expand.
4041 (altivec_vsum2sws_direct): New define_insn.
4042 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4044 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4046 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4049 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4054 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4055 realign stack if stack alignment needed is less than incoming
4058 2017-09-05 Marek Polacek <polacek@redhat.com>
4061 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4064 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4066 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4068 2017-09-05 Richard Biener <rguenther@suse.de>
4070 PR tree-optimization/82084
4071 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4073 2017-09-05 Richard Biener <rguenther@suse.de>
4075 PR tree-optimization/82102
4076 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4078 2017-09-05 Martin Liska <mliska@suse.cz>
4080 PR tree-optimization/82032
4081 * tree-cfg.c (generate_range_test): New function.
4082 * tree-cfg.h (generate_range_test): Declared here.
4083 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4084 (cleanup_control_expr_graph): Use it.
4085 * tree-switch-conversion.c (try_switch_expansion): Remove
4087 (emit_case_nodes): Use generate_range_test.
4089 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4092 * config/i386/i386.md (*<btsc><mode>_mask): Add
4093 TARGET_USE_BT to insn constraint.
4094 (*btr<mode>_mask): Ditto.
4096 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4098 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4099 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4101 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4104 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4105 TARGET_NEON and TARGET_IWMMXT.
4106 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4107 TARGET_NEON and TARGET_IWMMXT.
4108 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4110 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4112 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4113 (ix86_rewrite_tls_address): Ditto.
4114 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4115 (ix86_rewrite_tls_address_1): Ditto.
4116 (ix86_rewrite_tls_address): Ditto.
4117 * config/i386/predicates.md (tls_address_pattern): New predicate.
4118 * config/i386/i386.md (TLS address splitter): New splitter.
4120 2017-09-04 Richard Biener <rguenther@suse.de>
4122 PR tree-optimization/82084
4123 * fold-const.h (can_native_encode_string_p): Declare.
4124 * fold-const.c (can_native_encode_string_p): Factor out from ...
4125 (native_encode_string): ... here.
4126 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4127 vectorizing stores from constants we later cannot handle.
4129 2017-09-04 Marek Polacek <polacek@redhat.com>
4132 * doc/invoke.texi: Update -Wtautological-compare documentation.
4134 2017-09-04 Jeff Law <law@redhat.com>
4136 PR tree-optimization/64910
4137 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4138 swap the first and last operand if the last is a constant.
4140 2017-09-04 Marek Polacek <polacek@redhat.com>
4143 * convert.c (do_narrow): When sanitizing signed integer overflows,
4144 bail out for signed types.
4145 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4147 2017-09-04 Richard Biener <rguenther@suse.de>
4149 PR tree-optimization/82060
4150 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4151 Move devirtualization after stmt folding and before EH/AB/noreturn
4152 cleanup to get the stmt refs canonicalized. Use a bool instead
4153 of gimple_modified_p since that doesn't work for NOPs. Schedule
4154 NOPs generated by folding for removal.
4156 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4157 Alan Hayward <alan.hayward@arm.com>
4158 David Sherwood <david.sherwood@arm.com>
4160 * coretypes.h (pad_direction): New enum.
4161 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4162 (FUNCTION_ARG_PADDING): Likewise.
4163 * target.def (function_arg_padding): New hook.
4164 * targhooks.h (default_function_arg_padding): Declare.
4165 * targhooks.c (default_function_arg_padding): New function.
4166 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4167 (TARGET_FUNCTION_ARG_PADDING): ...this.
4168 * doc/tm.texi: Regenerate.
4169 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4170 instead of direction.
4171 (compute_argument_addresses): Likewise.
4172 (load_register_parameters): Likewise.
4173 (emit_library_call_value_1): Likewise.
4174 (store_one_arg): Use targetm.calls.function_arg_padding instead
4175 of FUNCTION_ARG_PADDING.
4176 (must_pass_in_stack_var_size_or_pad): Likewise.
4177 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4178 (emit_group_store): Likewise.
4179 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4180 instead of FUNCTION_ARG_PADDING.
4181 (emit_push_insn): Likewise, and propagate enum change throughout
4183 * function.h (direction): Delete.
4184 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4186 * function.c (assign_parm_find_stack_rtl): Likewise.
4187 (assign_parm_setup_block_p): Likewise.
4188 (assign_parm_setup_block): Likewise.
4189 (gimplify_parameters): Likewise.
4190 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4191 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4193 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4194 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4195 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4196 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4197 (aarch64_function_arg_padding): ...this new function.
4198 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4199 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4200 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4201 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4202 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4203 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4204 (arm_pad_arg_upward): Replace with...
4205 (arm_function_arg_padding): ...this new function.
4206 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4208 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4209 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4210 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4211 (ia64_hpux_function_arg_padding): Replace with...
4212 (ia64_function_arg_padding): ...this new function. Use pad_direction
4213 instead of direction. Check for TARGET_HPUX.
4214 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4215 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4216 (iq2000_function_arg_padding): New function.
4217 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4218 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4219 (mips_function_arg_padding): ...this new function.
4220 (mips_pad_reg_upward): Update accordingly.
4221 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4222 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4223 targetm.calls.function_arg_padding.
4224 (FUNCTION_ARG_PADDING): Delete.
4225 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4226 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4227 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4228 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4229 (nios2_block_reg_padding): Return pad_direction instead of direction.
4230 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4231 instead of direction.
4232 (nios2_function_arg_padding): Likewise. Make static.
4233 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4234 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4235 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4236 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4237 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4238 (pa_function_arg_padding): Make static. Return pad_direction instead
4240 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4241 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4242 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4243 instead of direction. Use targetm.calls.function_arg_padding.
4244 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4245 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4246 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4247 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4248 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4250 (function_arg_padding): Rename to...
4251 (rs6000_function_arg_padding): ...this. Make static. Return
4252 pad_direction instead of direction.
4253 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4254 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4255 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4256 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4257 instead of direction. Use targetm.calls.function_arg_padding.
4258 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4259 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4260 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4261 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4262 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4263 (function_arg_padding): Rename to...
4264 (rs6000_function_arg_padding): ...this. Make static. Return
4265 pad_direction instead of direction.
4266 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4267 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4268 * config/s390/s390.c (s390_function_arg_padding): New function.
4269 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4270 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4271 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4272 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4273 (function_arg_padding): Rename to...
4274 (sparc_function_arg_padding): ...this. Make static. Return
4275 pad_direction instead of direction.
4276 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4277 * config/spu/spu.c (spu_function_arg_padding): New function.
4278 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4279 * system.h (FUNCTION_ARG_PADDING): Poison.
4281 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4282 Alan Hayward <alan.hayward@arm.com>
4283 David Sherwood <david.sherwood@arm.com>
4285 * target.def (modes_tieable_p): New hook.
4286 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4287 (TARGET_MODES_TIEABLE_P): ...this.
4288 * doc/tm.texi.in: Regenerate.
4289 * hooks.h (hook_bool_mode_mode_true): Declare.
4290 * hooks.c (hook_bool_mode_mode_true): New function.
4291 * combine.c (subst): Use targetm.modes_tieable_p instead of
4293 * dse.c (find_shift_sequence): Likewise.
4294 * expmed.c (extract_low_bits): Likewise.
4295 * lower-subreg.c: Include target.h.
4296 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4298 * rtlanal.c (rtx_cost): Likewise.
4299 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4300 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4301 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4302 (TARGET_MODES_TIEABLE_P): Redefine.
4303 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4304 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4305 (TARGET_MODES_TIEABLE_P): Redefine.
4306 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4307 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4308 (arc_modes_tieable_p): New function.
4309 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4310 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4311 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4312 (arm_modes_tieable_p): Make static.
4313 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4314 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4315 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4316 (TARGET_MODES_TIEABLE_P): Redefine.
4317 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4318 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4319 (TARGET_MODES_TIEABLE_P): Redefine.
4320 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4321 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4322 (cr16_modes_tieable_p): New function.
4323 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4324 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4325 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4326 (TRULY_NOOP_TRUNCATION): Update comment.
4327 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4328 (TRULY_NOOP_TRUNCATION): Update comment.
4329 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4330 (frv_modes_tieable_p): New function.
4331 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4332 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4333 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4334 (TARGET_MODES_TIEABLE_P): Redefine.
4335 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4336 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4337 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4338 (TARGET_MODES_TIEABLE_P): Redefine.
4339 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4340 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4341 (ia64_modes_tieable_p): New function.
4342 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4343 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4344 (iq2000_modes_tieable_p): New function.
4345 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4346 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4347 (lm32_modes_tieable_p): New function.
4348 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4349 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4350 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4351 (TARGET_MODES_TIEABLE_P): Redefine.
4352 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4353 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4354 (m32r_modes_tieable_p): New function.
4355 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4356 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4357 (m68k_modes_tieable_p): New function.
4358 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4359 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4360 (mcore_modes_tieable_p): New function.
4361 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4362 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4364 (TARGET_MODES_TIEABLE_P): Redefine.
4365 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4366 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4367 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4368 (TARGET_MODES_TIEABLE_P): Redefine.
4369 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4370 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4371 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4372 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4373 (mn10300_modes_tieable_p): ...this and make static.
4374 (TARGET_MODES_TIEABLE_P): Redefine.
4375 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4376 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4377 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4378 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4379 (msp430_modes_tieable_p): Make static.
4380 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4381 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4382 (TARGET_MODES_TIEABLE_P): Redefine.
4383 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4384 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4385 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4386 (TARGET_MODES_TIEABLE_P): Redefine.
4387 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4388 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4389 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4390 (TARGET_MODES_TIEABLE_P): Redefine.
4391 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4392 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4393 (pdp11_modes_tieable_p): New function.
4394 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4395 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4396 (rs6000_modes_tieable_p): New function.
4397 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4398 * config/powerpcspe/powerpcspe.md: Update comment.
4399 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4400 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4401 (TARGET_MODES_TIEABLE_P): Redefine.
4402 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4403 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4404 (rl78_modes_tieable_p): New function.
4405 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4406 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4407 (rs6000_modes_tieable_p): New function.
4408 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4409 * config/rs6000/rs6000.md: Update comment.
4410 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4411 * config/rx/rx.c (rx_modes_tieable_p): New function.
4412 (TARGET_MODES_TIEABLE_P): Redefine.
4413 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4414 * config/s390/s390.c (s390_modes_tieable_p): New function.
4415 (TARGET_MODES_TIEABLE_P): Redefine.
4416 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4417 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4418 (sh_modes_tieable_p): New function.
4419 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4420 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4421 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4422 (sparc_modes_tieable_p): Make static.
4423 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4424 * config/spu/spu.c (spu_modes_tieable_p): New function.
4425 (TARGET_MODES_TIEABLE_P): Redefine.
4426 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4427 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4428 (TARGET_MODES_TIEABLE_P): Redefine.
4429 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4430 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4431 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4432 * config/v850/v850.c (v850_modes_tieable_p): New function.
4433 (TARGET_MODES_TIEABLE_P): Redefine.
4434 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4435 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4436 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4437 (visium_modes_tieable_p): New function.
4438 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4439 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4440 (xtensa_modes_tieable_p): New function.
4441 * system.h (MODES_TIEABLE_P): Poison.
4443 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4444 Alan Hayward <alan.hayward@arm.com>
4445 David Sherwood <david.sherwood@arm.com>
4447 * target.def (hard_regno_mode_ok): New hook.
4448 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4449 (TARGET_HARD_REGNO_MODE_OK): ...this.
4450 * doc/tm.texi.in: Regenerate.
4451 * hooks.h (hook_bool_uint_mode_true): Declare.
4452 * hooks.c (hook_bool_uint_mode_true): New function.
4453 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4455 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4456 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4457 instead of HARD_REGNO_MODE_OK.
4458 * caller-save.c: Include target.h.
4459 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4461 * combine.c (can_combine_p): Likewise.
4462 (combinable_i3pat): Likewise.
4463 (can_change_dest_mode): Likewise.
4464 * expr.c (init_expr_target): Likewise.
4465 (convert_move): Likewise.
4466 (convert_modes): Likewise.
4467 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4468 (setup_prohibited_mode_move_regs): Likewise.
4469 * ira.h (target_ira): Likewise.
4470 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4471 * lra-constraints.c (process_alt_operands): Likewise.
4472 (split_reg): Likewise.
4473 * recog.c (peep2_find_free_register): Likewise.
4474 * ree.c (combine_reaching_defs): Likewise.
4475 * regcprop.c (maybe_mode_change): Likewise.
4476 * reginfo.c (init_reg_sets_1): Likewise.
4477 (choose_hard_reg_mode): Likewise.
4478 (simplifiable_subregs): Likewise.
4479 * regrename.c (check_new_reg_p): Likewise.
4480 * reload.c (find_valid_class): Likewise.
4481 (find_valid_class_1): Likewise.
4482 (reload_inner_reg_of_subreg): Likewise.
4483 (push_reload): Likewise.
4484 (combine_reloads): Likewise.
4485 (find_dummy_reload): Likewise.
4486 (find_reloads): Likewise.
4487 * reload1.c (find_reg): Likewise.
4488 (set_reload_reg): Likewise.
4489 (allocate_reload_reg): Likewise.
4490 (choose_reload_regs): Likewise.
4491 (reload_adjust_reg_for_temp): Likewise.
4492 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4493 (simplify_subreg_regno): Likewise.
4494 * sel-sched.c (init_regs_for_mode): Likewise.
4495 * varasm.c (make_decl_rtl): Likewise.
4496 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4497 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4499 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4500 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4502 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4503 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4504 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4505 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4506 (arc_mode_class): Delete.
4507 (HARD_REGNO_MODE_OK): Delete.
4508 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4509 (arc_hard_regno_mode_ok): Rename old array to...
4510 (arc_hard_regno_mode_ok_modes): ...this.
4511 (arc_conditional_register_usage): Update accordingly.
4512 (arc_mode_class): Make static.
4513 (arc_hard_regno_mode_ok): New function.
4514 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4515 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4516 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4517 (arm_hard_regno_mode_ok): Make static.
4518 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4520 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4521 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4522 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4524 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4525 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4526 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4527 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4528 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4529 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4530 * config/bfin/predicates.md (valid_reg_operand): Use
4531 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4532 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4533 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4535 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4536 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4537 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4538 (cr16_hard_regno_mode_ok): Make static and return a bool.
4539 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4540 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4541 (cris_hard_regno_mode_ok): New function.
4542 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4543 (epiphany_mode_class): Delete.
4544 (HARD_REGNO_MODE_OK): Delete.
4545 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4546 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4547 (hard_regno_mode_ok): Rename to...
4548 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4549 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4550 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4552 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4553 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4554 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4555 (frv_hard_regno_mode_ok): Make static and return a bool.
4556 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4558 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4559 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4560 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4561 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4563 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4564 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4565 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4566 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4568 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4569 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4570 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4571 (ia64_hard_regno_mode_ok): New function.
4572 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4573 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4574 (iq2000_hard_regno_mode_ok): New function.
4575 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4576 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4577 (lm32_hard_regno_mode_ok): New function.
4578 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4579 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4580 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4581 instead of HARD_REGNO_MODE_OK.
4582 (m32c_hard_regno_ok): Rename to...
4583 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4584 (m32c_cannot_change_mode_class): Update accordingly.
4585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4586 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4587 (m32r_mode_class): Delete.
4588 (HARD_REGNO_MODE_OK): Delete.
4589 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4590 (m32r_hard_regno_mode_ok): Rename to...
4591 (m32r_hard_regno_modes): ...this.
4592 (m32r_mode_class): Make static.
4593 (m32r_hard_regno_mode_ok): New function.
4594 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4595 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4596 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4597 (m68k_hard_regno_mode_ok): Make static.
4598 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4599 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4600 (mcore_hard_regno_mode_ok): New function.
4601 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4602 (HARD_REGNO_MODE_OK): Delete.
4603 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4605 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4606 (microblaze_hard_regno_mode_ok): New function.
4607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4608 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4609 (mips_hard_regno_mode_ok): Delete.
4610 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4611 (mips_hard_regno_mode_ok_p): ...this and make static.
4612 (mips_hard_regno_mode_ok_p): Rename to...
4613 (mips_hard_regno_mode_ok_uncached): ...this.
4614 (mips_hard_regno_mode_ok): New function.
4615 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4616 of HARD_REGNO_MODE_OK.
4617 (mips_option_override): Update after above name changes.
4618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4619 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4620 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4621 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4622 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4624 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4625 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4626 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4627 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4628 (msp430_hard_regno_mode_ok): Make static and return a bool.
4629 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4630 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4631 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4633 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4634 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4635 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4636 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4637 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4638 (PA_HARD_REGNO_MODE_OK): ...this
4639 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4640 (PA_HARD_REGNO_MODE_OK): ...this.
4641 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4642 (pa_hard_regno_mode_ok): New function.
4643 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4644 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4645 (pdp11_hard_regno_mode_ok): New function.
4646 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4647 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4649 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4651 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4652 (rs6000_hard_regno_mode_ok): Rename to...
4653 (rs6000_hard_regno_mode_ok_uncached): ...this.
4654 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4655 (rs6000_hard_regno_mode_ok): New function.
4656 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4657 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4658 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4659 (riscv_hard_regno_mode_ok): ...this and make static.
4660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4661 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4662 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4663 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4664 (rl78_hard_regno_mode_ok): Make static and return bool.
4665 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4666 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4668 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4669 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4670 (rs6000_hard_regno_mode_ok): Rename to...
4671 (rs6000_hard_regno_mode_ok_uncached): ...this.
4672 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4673 (rs6000_hard_regno_mode_ok): New function.
4674 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4675 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4677 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4678 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4679 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4681 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4682 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4683 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4684 (sh_hard_regno_mode_ok): Make static.
4685 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4686 instead of HARD_REGNO_MODE_OK.
4687 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4688 (sparc_mode_class): Delete.
4689 (HARD_REGNO_MODE_OK): Delete.
4690 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4691 (hard_regno_mode_classes): Make static.
4692 (sparc_mode_class): Likewise.
4693 (sparc_hard_regno_mode_ok): New function.
4694 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4695 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4696 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4698 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4699 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4700 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4701 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4702 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4703 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4704 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4705 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4706 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4707 (visium_hard_regno_mode_ok): New function.
4708 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4709 instead of HARD_REGNO_MODE_OK.
4710 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4711 (HARD_REGNO_MODE_OK): Delete.
4712 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4713 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4714 (xtensa_option_override): Update accordingly.
4715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4716 (xtensa_hard_regno_mode_ok): New function.
4717 * system.h (HARD_REGNO_MODE_OK): Poison.
4719 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4720 Alan Hayward <alan.hayward@arm.com>
4721 David Sherwood <david.sherwood@arm.com>
4723 * target.def (hard_regno_call_part_clobbered): New hook.
4724 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4725 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4726 * doc/tm.texi: Regenerate.
4727 * hooks.h (hook_bool_uint_mode_false): Declare.
4728 * hooks.c (hook_bool_uint_mode_false): New function.
4729 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4730 * cselib.c (cselib_process_insn): Use
4731 targetm.hard_regno_call_part_clobbered instead of
4732 HARD_REGNO_CALL_PART_CLOBBERED.
4733 * ira-conflicts.c (ira_build_conflicts): Likewise.
4734 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4735 * lra-constraints.c (need_for_call_save_p): Likewise.
4736 * lra-lives.c: Include target.h.
4737 (check_pseudos_live_through_calls): Use
4738 targetm.hard_regno_call_part_clobbered instead of
4739 HARD_REGNO_CALL_PART_CLOBBERED.
4740 * regcprop.c: Include target.h.
4741 (copyprop_hardreg_forward_1): Use
4742 targetm.hard_regno_call_part_clobbered instead of
4743 HARD_REGNO_CALL_PART_CLOBBERED.
4744 * reginfo.c (choose_hard_reg_mode): Likewise.
4745 * regrename.c (check_new_reg_p): Likewise.
4746 * reload.c (find_equiv_reg): Likewise.
4747 * reload1.c (emit_reload_insns): Likewise.
4748 * sched-deps.c (deps_analyze_insn): Likewise.
4749 * sel-sched.c (init_regs_for_mode): Likewise.
4750 (mark_unavailable_hard_regs): Likewise.
4751 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4752 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4753 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4755 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4756 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4757 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4759 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4761 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4762 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4763 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4765 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4766 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4767 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4769 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4770 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4772 * config/powerpcspe/powerpcspe.c
4773 (rs6000_hard_regno_call_part_clobbered): New function.
4774 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4775 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4776 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4778 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4779 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4780 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4782 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4783 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4784 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4786 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4787 Alan Hayward <alan.hayward@arm.com>
4788 David Sherwood <david.sherwood@arm.com>
4790 * rtl.h (subreg_memory_offset): Declare.
4791 * emit-rtl.c (subreg_memory_offset): New function.
4792 * expmed.c (store_bit_field_1): Use it.
4793 * expr.c (undefined_operand_subword_p): Likewise.
4794 * simplify-rtx.c (simplify_subreg): Likewise.
4796 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4798 PR rtl-optimization/57448
4801 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4802 using atomic_load pattern.
4803 (expand_atomic_store): Likewise.
4805 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4808 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4809 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4810 handling. Use replace_call_with_value with NULL instead of
4811 gsi_replace, unlink_stmt_vdef and release_defs.
4813 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4816 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4818 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4821 * rtl.h (emit_library_call_value_1): Declare.
4822 (emit_library_call): Replace declaration with a series of overloads.
4823 Remove the parameter count argument.
4824 (emit_library_call_value): Likewise.
4825 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4826 with an "rtx_mode_t *".
4827 (emit_library_call_value): Delete.
4828 (emit_library_call): Likewise.
4829 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4830 (asan_emit_allocas_unpoison): Likewise.
4831 * builtins.c (expand_builtin_powi): Likewise.
4832 (expand_asan_emit_allocas_unpoison): Likewise.
4833 * cfgexpand.c (expand_main_function): Likewise.
4834 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4835 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4836 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4837 * config/arm/arm.c (arm_trampoline_init): Likewise.
4838 (arm_call_tls_get_addr): Likewise.
4839 (arm_expand_divmod_libfunc): Likewise.
4840 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
4841 (smulsi3_highpart): Likewise.
4842 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
4843 (c6x_expand_compare): Likewise.
4844 (c6x_expand_movmem): Likewise.
4845 * config/frv/frv.c (frv_trampoline_init): Likewise.
4846 * config/i386/i386.c (ix86_trampoline_init): Likewise.
4847 (ix86_expand_divmod_libfunc): Likewise.
4848 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
4849 (ia64_expand_compare): Likewise.
4850 (ia64_profile_hook): Likewise.
4851 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
4852 (nonlocal_goto): Likewise.
4853 (restore_stack_nonlocal): Likewise.
4854 * config/m32r/m32r.c (block_move_call): Likewise.
4855 (m32r_trampoline_init): Likewise.
4856 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
4857 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
4858 (m68k_call_m68k_read_tp): Likewise.
4859 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
4860 (microblaze_expand_divide): Likewise.
4861 * config/mips/mips.h (mips_args): Likewise.
4862 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
4863 (MIPS_ICACHE_SYNC): Likewise.
4864 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
4865 (nios2_trampoline_init): Likewise.
4866 * config/pa/pa.c (hppa_tls_call): Likewise.
4867 (pa_trampoline_init): Likewise.
4868 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
4869 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
4870 (expand_strn_compare): Likewise.
4871 (rs6000_generate_compare): Likewise.
4872 (rs6000_expand_float128_convert): Likewise.
4873 (output_profile_hook): Likewise.
4874 (rs6000_trampoline_init): Likewise.
4875 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
4876 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
4877 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
4878 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
4879 (rs6000_generate_compare): Likewise.
4880 (rs6000_expand_float128_convert): Likewise.
4881 (output_profile_hook): Likewise.
4882 (rs6000_trampoline_init): Likewise.
4883 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
4884 * config/sh/sh.c (sh_trampoline_init): Likewise.
4885 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
4886 (sparc_emit_float_lib_cmp): Likewise.
4887 (sparc32_initialize_trampoline): Likewise.
4888 (sparc64_initialize_trampoline): Likewise.
4889 (sparc_profile_hook): Likewise.
4890 * config/spu/spu.c (ea_load_store): Likewise.
4891 * config/spu/spu.md (floatunssidf2): Likewise.
4892 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
4893 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
4894 * config/visium/visium.c (expand_block_move_4): Likewise.
4895 (expand_block_move_2): Likewise.
4896 (expand_block_move_1): Likewise.
4897 (expand_block_set_4): Likewise.
4898 (expand_block_set_2): Likewise.
4899 (expand_block_set_1): Likewise.
4900 (visium_trampoline_init): Likewise.
4901 (visium_profile_hook): Likewise.
4902 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
4903 (xtensa_setup_frame_addresses): Likewise.
4904 (xtensa_trampoline_init): Likewise.
4905 * except.c (sjlj_emit_function_enter): Likewise.
4906 (sjlj_emit_function_exit): Likewise.
4907 * explow.c (allocate_dynamic_stack_space): Likewise.
4908 (probe_stack_range): Likewise.
4909 * expr.c (convert_mode_scalar): Likewise.
4910 * optabs.c (expand_binop): Likewise.
4911 (expand_twoval_binop_libfunc): Likewise.
4912 (expand_unop): Likewise.
4913 (prepare_cmp_insn): Likewise.
4914 (prepare_float_lib_cmp): Likewise.
4915 (expand_float): Likewise.
4916 (expand_fix): Likewise.
4917 (expand_fixed_convert): Likewise.
4918 (maybe_emit_sync_lock_test_and_set): Likewise.
4919 (expand_atomic_compare_and_swap): Likewise.
4920 (expand_mem_thread_fence): Likewise.
4921 (expand_atomic_fetch_op): Likewise.
4923 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
4925 * doc/generic.texi (OpenACC): Adjust URL.
4926 * doc/invoke.texi (C Dialect Options): Ditto.
4928 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
4930 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
4931 predicate for operand 1. Add (m,<S>) constraint.
4932 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
4933 Prevent memory operand 1 with register operand 2.
4935 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
4937 PR rtl-optimization/82024
4938 * combine.c (try_combine): If the combination result is a PARALLEL,
4939 and we only need to retain the SET in there that would be placed
4940 at I2, check that we can place that at I3 instead, before doing so.
4942 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4945 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
4947 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
4948 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
4951 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
4952 Jeff Law <law@redhat.com>
4954 * varasm.c (bss_initializer_p): Do not put constants into .bss
4955 (categorize_decl_for_section): Handle bss_initializer_p returning
4956 false when DECL_INITIAL is NULL.
4958 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4961 * config/s390/s390.c (s390_can_inline_p): New function.
4963 2017-09-01 Jeff Law <law@redhat.com>
4965 PR tree-optimization/82052
4966 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
4967 Always initialize the returned slot after a hash table miss
4968 when INSERT is true.
4970 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
4972 * config/s390/s390.md (mem_signal_fence): Remove.
4973 * doc/md.texi (mem_signal_fence): Remove.
4974 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
4976 * target-insns.def (mem_signal_fence): Remove.
4978 2017-09-01 Jakub Jelinek <jakub@redhat.com>
4981 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
4984 * asan.c (create_odr_indicator): Strip name encoding from assembler
4985 name before appending it after __odr_asan_.
4987 2017-09-01 Martin Liska <mliska@suse.cz>
4989 PR tree-optimization/82059
4990 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
4991 frequency only when an edge is redirected.
4993 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
4995 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
4996 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
4997 (arc_conditional_register_usage): Remove ARC600 lp_count
4999 (arc_file_start): Emit Tag_ARC_CPU_variation.
5000 (arc_can_use_doloop_p): New conditions to use ZOLs.
5001 (hwloop_fail): New function.
5002 (hwloop_optimize): Likewise.
5003 (hwloop_pattern_reg): Likewise.
5004 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5005 (arc_reorg_loops): New function, calls reorg_loops.
5006 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5007 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5009 (arc_loop_hazard): Remove function, functionality moved into
5011 (arc_hazard): Remove arc_loop_hazard call.
5012 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5013 into hwloop_optimize.
5014 (arc_label_align): Remove ZOL handling.
5015 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5016 * config/arc/arc.md (doloop_begin): Remove pattern.
5017 (doloop_begin_i): Likewise.
5018 (doloop_end_i): Likewise.
5019 (doloop_fallback): Likewise.
5020 (doloop_fallback_m): Likewise.
5021 (doloop_end): Reimplement expand.
5022 (arc_lp): New pattern for LP instruction.
5023 (loop_end): New pattern.
5024 (loop_fail): Likewise.
5025 (decrement_and_branch_until_zero): Likewise.
5026 * config/arc/arc.opt (mlpc-width): New option.
5027 * doc/invoke.texi (mlpc-width): Document option.
5029 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5031 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5032 (arc_ccfsm_advance): Fix checking for delay slots.
5033 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5035 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5037 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5039 (movhi_insn): Update store instruction constraint which are saving
5040 6-bit short immediates.
5041 (movsi_insn): Consider also short scaled load operations.
5042 (zero_extendhisi2_i): Use Usd constraint instead of T.
5043 (extendhisi2_i): Add q constraint.
5044 (arc_clzsi2): Add type and length attributes.
5045 (arc_ctzsi2): Likewise.
5046 * config/arc/constraints.md (Usc): Update constraint, the
5047 assembler can parse two relocations for a single instruction.
5049 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5051 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5052 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5054 2017-08-31 Olivier Hainque <hainque@adacore.com>
5056 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5057 match as powerpc-wrs-vxworks*.
5059 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5061 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5062 register constraint for by-element operand.
5063 (aarch64_mls_elt_merge<mode>): Likewise.
5065 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5067 * config/arc/arc.c (arc_can_follow_jump): Check for short
5070 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5072 * config.gcc: Use g.opt for arc.
5073 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5074 functionality moved to ...
5075 (legitimate_scaled_address_p): New function, ...here.
5076 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5077 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5078 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5080 (arc_override_options): Handle G option.
5081 (arc_output_pic_addr_const): Correct function definition.
5082 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5083 (arc_decl_anon_ns_mem_p): Delete.
5084 (arc_in_small_data_p): Overhaul this function to take into
5085 consideration the value given via G option.
5086 (arc_rewrite_small_data_1): Renamed and corrected old
5087 arc_rewrite_small_data function.
5088 (arc_rewrite_small_data): New function.
5089 (small_data_pattern): Don't use pic_offset_table_rtx.
5090 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5091 * config/arc/simdext.md (movmisalignv2hi): Use
5092 prepare_move_operands function.
5094 (movmisalign*): Likewise.
5095 * doc/invoke.texi (ARC options): Document -G option.
5097 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5099 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5101 * config/arc/arc.c (arc_print_operand): Output scalled address for
5102 sdata whenever is possible.
5103 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5104 load/stores are available.
5105 (compact_sda_memory_operand): Check for the alignment required by
5106 code density instructions.
5107 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5109 * config/arc/constraints.md (Usd): Update constraint.
5110 (Us0): New constraint.
5111 (Usc): Update constraint.
5113 2017-08-31 Richard Biener <rguenther@suse.de>
5116 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5119 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5121 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5122 Resize type_signature.
5124 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5125 Alan Hayward <alan.hayward@arm.com>
5126 David Sherwood <david.sherwood@arm.com>
5128 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5129 subregs whose inner modes can be stored in GPRs.
5130 (aarch64_classify_index): Likewise.
5132 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5133 Alan Hayward <alan.hayward@arm.com>
5134 David Sherwood <david.sherwood@arm.com>
5136 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5137 (V_INT_EQUIV): ...this.
5138 (v_cmp_result): Rename to...
5139 (v_int_equiv): ...this.
5140 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5141 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5142 (copysign<mode>3): Likewise.
5143 (aarch64_simd_bsl<mode>_internal): Likewise.
5144 (aarch64_simd_bsl<mode>): Likewise.
5145 (vec_cmp<mode><mode>): Likewise.
5146 (vcond<mode><mode>): Likewise.
5147 (vcond<v_cmp_mixed><mode>): Likewise.
5148 (vcondu<mode><v_cmp_mixed>): Likewise.
5149 (aarch64_cm<optab><mode>): Likewise.
5150 (aarch64_cmtst<mode>): Likewise.
5151 (aarch64_fac<optab><mode>): Likewise.
5152 (vec_perm_const<mode>): Likewise.
5153 (vcond_mask_<mode><v_cmp_result>): Rename to...
5154 (vcond_mask_<mode><v_int_equiv>): ...this.
5155 (vec_cmp<mode><v_cmp_result>): Rename to...
5156 (vec_cmp<mode><v_int_equiv>): ...this.
5158 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5159 Alan Hayward <alan.hayward@arm.com>
5160 David Sherwood <david.sherwood@arm.com>
5162 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5164 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5165 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5166 (UNSPEC_LD4_DREG): New unspecs.
5167 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5168 (aarch64_ld2<mode>_dreg_be): Replace with...
5169 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5171 (aarch64_ld3<mode>_dreg_le)
5172 (aarch64_ld3<mode>_dreg_be): Replace with...
5173 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5175 (aarch64_ld4<mode>_dreg_le)
5176 (aarch64_ld4<mode>_dreg_be): Replace with...
5177 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5180 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5182 PR tree-optimization/81987
5183 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5184 insert an initializer in a location not dominated by the stride
5187 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5189 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5190 on the entire header of the finally block in the fallthru case.
5192 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5194 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5196 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5198 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5199 rs6000_emit_move_from_cr and call renamed function.
5200 (rs6000_emit_prologue): Call renamed functions.
5201 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5202 movesi_from_cr, remove volatile CRs.
5204 2017-08-30 Jon Beniston <jon@beniston.com>
5205 Richard Biener <rguenther@suse.de>
5207 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5208 of VECTOR_MODE_P check.
5209 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5210 element vector types.
5212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5214 * df.h (df_read_modify_subreg_p): Remove in favor of...
5215 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5216 const_rtx instead of an rtx.
5217 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5218 * df-problems.c (df_word_lr_mark_ref): Likewise.
5219 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5220 (mark_pseudo_reg_dead): Likewise.
5221 (mark_ref_dead): Likewise.
5222 * reginfo.c (init_subregs_of_mode): Likewise.
5223 * sched-deps.c (sched_analyze_1): Likewise.
5224 * df-scan.c (df_def_record_1): Likewise.
5225 (df_uses_record): Likewise.
5226 (df_read_modify_subreg_p): Remove in favor of...
5227 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5228 const_rtx instead of an rtx.
5230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5231 Alan Hayward <alan.hayward@arm.com>
5232 David Sherwood <david.sherwood@arm.com>
5234 * rtl.h (partial_subreg_p): New function.
5235 * caller-save.c (save_call_clobbered_regs): Use it.
5236 * calls.c (expand_call): Likewise.
5237 * combine.c (combinable_i3pat): Likewise.
5238 (simplify_set): Likewise.
5239 (make_extraction): Likewise.
5240 (make_compound_operation_int): Likewise.
5241 (gen_lowpart_or_truncate): Likewise.
5242 (force_to_mode): Likewise.
5243 (make_field_assignment): Likewise.
5244 (reg_truncated_to_mode): Likewise.
5245 (record_truncated_value): Likewise.
5246 (move_deaths): Likewise.
5247 * cse.c (record_jump_cond): Likewise.
5248 (cse_insn): Likewise.
5249 * cselib.c (cselib_lookup_1): Likewise.
5250 * expmed.c (extract_bit_field_using_extv): Likewise.
5251 * function.c (assign_parm_setup_reg): Likewise.
5252 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5253 * ira-build.c (create_insn_allocnos): Likewise.
5254 * lra-coalesce.c (merge_pseudos): Likewise.
5255 * lra-constraints.c (match_reload): Likewise.
5256 (simplify_operand_subreg): Likewise.
5257 (curr_insn_transform): Likewise.
5258 * lra-lives.c (process_bb_lives): Likewise.
5259 * lra.c (new_insn_reg): Likewise.
5260 (lra_substitute_pseudo): Likewise.
5261 * regcprop.c (mode_change_ok): Likewise.
5262 (maybe_mode_change): Likewise.
5263 (copyprop_hardreg_forward_1): Likewise.
5264 * reload.c (push_reload): Likewise.
5265 (find_reloads): Likewise.
5266 (find_reloads_subreg_address): Likewise.
5267 * reload1.c (alter_reg): Likewise.
5268 (eliminate_regs_1): Likewise.
5269 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5271 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5273 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5274 back to if statements, including unpack.
5276 2017-08-30 Martin Liska <mliska@suse.cz>
5279 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5281 (func_checker::compare_asm_inputs_outputs): ... this function.
5282 (func_checker::compare_gimple_asm): Use the function to compare
5283 also ASM constrains.
5284 * ipa-icf-gimple.h: Rename the function.
5286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5287 Alan Hayward <alan.hayward@arm.com>
5288 David Sherwood <david.sherwood@arm.com>
5290 * coretypes.h (complex_mode): New type.
5291 * gdbhooks.py (build_pretty_printer): Handle it.
5292 * machmode.h (complex_mode): New class.
5293 (complex_mode::includes_p): New function.
5294 (is_complex_int_mode): Likewise.
5295 (is_complex_float_mode): Likewise.
5296 * genmodes.c (get_mode_class): Handle complex mode classes.
5297 * function.c (expand_function_end): Use is_complex_int_mode.
5299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5300 Alan Hayward <alan.hayward@arm.com>
5301 David Sherwood <david.sherwood@arm.com>
5303 * coretypes.h (scalar_mode_pod): New typedef.
5304 * gdbhooks.py (build_pretty_printer): Handle it.
5305 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5306 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5307 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5308 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5312 Alan Hayward <alan.hayward@arm.com>
5313 David Sherwood <david.sherwood@arm.com>
5315 * machmode.h (mode_for_vector): Take a scalar_mode instead
5317 * stor-layout.c (mode_for_vector): Likewise.
5318 * explow.c (promote_mode): Use as_a <scalar_mode>.
5319 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5322 Alan Hayward <alan.hayward@arm.com>
5323 David Sherwood <david.sherwood@arm.com>
5325 * target.def (preferred_simd_mode): Take a scalar_mode
5326 instead of a machine_mode.
5327 * targhooks.h (default_preferred_simd_mode): Likewise.
5328 * targhooks.c (default_preferred_simd_mode): Likewise.
5329 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5330 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5331 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5332 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5333 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5334 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5335 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5336 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5338 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5339 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5340 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5341 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5342 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5343 * doc/tm.texi: Regenerate.
5344 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5348 Alan Hayward <alan.hayward@arm.com>
5349 David Sherwood <david.sherwood@arm.com>
5351 * target.def (scalar_mode_supported_p): Take a scalar_mode
5352 instead of a machine_mode.
5353 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5354 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5355 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5356 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5357 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5358 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5359 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5360 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5361 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5362 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5363 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5364 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5365 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5366 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5368 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5369 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5370 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5371 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5372 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5374 * doc/tm.texi: Regenerate.
5376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5377 Alan Hayward <alan.hayward@arm.com>
5378 David Sherwood <david.sherwood@arm.com>
5380 * coretypes.h (opt_scalar_mode): New typedef.
5381 * gdbhooks.py (build_pretty_printers): Handle it.
5382 * machmode.h (mode_iterator::get_2xwider): Add overload for
5384 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5386 * expr.c (convert_mode_scalar): Likewise.
5387 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5388 * optabs.c (expand_float): Likewise.
5389 (expand_fix): Likewise.
5390 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5393 Alan Hayward <alan.hayward@arm.com>
5394 David Sherwood <david.sherwood@arm.com>
5396 * optabs.c (expand_float): Explicitly check for scalars before
5397 using a branching expansion.
5398 (expand_fix): Likewise.
5400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5401 Alan Hayward <alan.hayward@arm.com>
5402 David Sherwood <david.sherwood@arm.com>
5404 * expr.c (convert_mode): Split scalar handling out into...
5405 (convert_mode_scalar): ...this new function. Treat the modes
5408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5409 Alan Hayward <alan.hayward@arm.com>
5410 David Sherwood <david.sherwood@arm.com>
5412 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5414 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5417 Alan Hayward <alan.hayward@arm.com>
5418 David Sherwood <david.sherwood@arm.com>
5420 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5421 rather than a machine_mode.
5422 (fixed_from_string): Likewise.
5423 (fixed_convert): Likewise.
5424 (fixed_convert_from_int): Likewise.
5425 (fixed_convert_from_real): Likewise.
5426 (real_convert_from_fixed): Likewise.
5427 * fixed-value.c (fixed_from_double_int): Likewise.
5428 (fixed_from_string): Likewise.
5429 (fixed_convert): Likewise.
5430 (fixed_convert_from_int): Likewise.
5431 (fixed_convert_from_real): Likewise.
5432 (real_convert_from_fixed): Likewise.
5433 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5436 Alan Hayward <alan.hayward@arm.com>
5437 David Sherwood <david.sherwood@arm.com>
5439 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5440 of separate mode class checks. Do not allow vector modes here.
5441 (immed_wide_int_const): Use as_a <scalar_mode>.
5442 * explow.c (trunc_int_for_mode): Likewise.
5443 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5444 (wi::shwi): Likewise.
5445 (wi::min_value): Likewise.
5446 (wi::max_value): Likewise.
5447 * dwarf2out.c (loc_descriptor): Likewise.
5448 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5452 Alan Hayward <alan.hayward@arm.com>
5453 David Sherwood <david.sherwood@arm.com>
5455 * tree.h (SCALAR_TYPE_MODE): New macro.
5456 * expr.c (expand_expr_addr_expr_1): Use it.
5457 (expand_expr_real_2): Likewise.
5458 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5459 (fold_convert_const_fixed_from_int): Likewise.
5460 (fold_convert_const_fixed_from_real): Likewise.
5461 (native_encode_fixed): Likewise
5462 (native_encode_complex): Likewise
5463 (native_encode_vector): Likewise.
5464 (native_interpret_fixed): Likewise.
5465 (native_interpret_real): Likewise.
5466 (native_interpret_complex): Likewise.
5467 (native_interpret_vector): Likewise.
5468 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5469 (simd_clone_adjust_argument_types): Likewise.
5470 (simd_clone_init_simd_arrays): Likewise.
5471 (simd_clone_adjust): Likewise.
5472 * stor-layout.c (layout_type): Likewise.
5473 * tree.c (build_minus_one_cst): Likewise.
5474 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5475 * tree-inline.c (estimate_move_cost): Likewise.
5476 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5477 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5478 (vectorizable_reduction): Likewise.
5479 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5480 (vect_recog_mixed_size_cond_pattern): Likewise.
5481 (check_bool_pattern): Likewise.
5482 (adjust_bool_pattern): Likewise.
5483 (search_type_for_mask_1): Likewise.
5484 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5485 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5486 (vectorizable_load): Likewise.
5487 (vectorizable_store): Likewise.
5488 * ubsan.c (ubsan_encode_value): Likewise.
5489 * varasm.c (output_constant): Likewise.
5491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5492 Alan Hayward <alan.hayward@arm.com>
5493 David Sherwood <david.sherwood@arm.com>
5495 * coretypes.h (scalar_mode): New class.
5496 * machmode.h (scalar_mode): Likewise.
5497 (scalar_mode::includes_p): New function.
5498 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5499 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5500 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5501 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5502 * expmed.c (store_bit_field_1): Likewise.
5503 (extract_bit_field_1): Likewise.
5504 * expr.c (write_complex_part): Likewise.
5505 (read_complex_part): Likewise.
5506 (emit_move_complex_push): Likewise.
5507 (expand_expr_real_2): Likewise.
5508 * function.c (assign_parm_setup_reg): Likewise.
5509 (assign_parms_unsplit_complex): Likewise.
5510 * optabs.c (expand_binop): Likewise.
5511 * rtlanal.c (subreg_get_info): Likewise.
5512 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5513 * varasm.c (output_constant_pool_2): Likewise.
5515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5516 Alan Hayward <alan.hayward@arm.com>
5517 David Sherwood <david.sherwood@arm.com>
5519 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5521 (expmed_mult_highpart_optab): Likewise.
5522 (expmed_mult_highpart): Likewise.
5524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5525 Alan Hayward <alan.hayward@arm.com>
5526 David Sherwood <david.sherwood@arm.com>
5528 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5529 instead of a machine_mode.
5530 (builtin_memset_read_str): Likewise.
5531 * builtins.c (c_readstr): Likewise.
5532 (builtin_memcpy_read_str): Likewise.
5533 (builtin_strncpy_read_str): Likewise.
5534 (builtin_memset_read_str): Likewise.
5535 (builtin_memset_gen_str): Likewise.
5536 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5537 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5538 instead of a machine_mode.
5539 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5541 (make_extraction): Likewise.
5542 (try_widen_shift_mode): Take and return scalar_int_modes instead
5544 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5545 a scalar_int_mode instead of a machine_mode.
5546 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5547 (avr_addr_space_pointer_mode): Likewise.
5548 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5549 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5550 (msp430_unwind_word_mode): Likewise.
5551 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5552 (spu_addr_space_pointer_mode): Likewise.
5553 (spu_addr_space_address_mode): Likewise.
5554 (spu_libgcc_cmp_return_mode): Likewise.
5555 (spu_libgcc_shift_count_mode): Likewise.
5556 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5557 (rl78_addr_space_pointer_mode): Likewise.
5558 (fl78_unwind_word_mode): Likewise.
5559 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5561 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5562 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5563 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5564 (mips_valid_pointer_mode): Likewise.
5565 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5566 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5567 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5569 (ft32_addr_space_address_mode): Likewise.
5570 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5571 scalar_int_mode instead of a machine_mode.
5572 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5574 (m32c_addr_space_address_mode): Likewise.
5575 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5576 (rs6000_eh_return_filter_mode): Likewise.
5577 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5578 (rs6000_eh_return_filter_mode): Likewise.
5579 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5580 (s390_libgcc_shift_count_mode): Likewise.
5581 (s390_unwind_word_mode): Likewise.
5582 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5584 * target.def (mode_rep_extended): Likewise.
5585 (valid_pointer_mode): Likewise.
5586 (addr_space.valid_pointer_mode): Likewise.
5587 (eh_return_filter_mode): Return a scalar_int_mode rather than
5589 (libgcc_cmp_return_mode): Likewise.
5590 (libgcc_shift_count_mode): Likewise.
5591 (unwind_word_mode): Likewise.
5592 (addr_space.pointer_mode): Likewise.
5593 (addr_space.address_mode): Likewise.
5594 * doc/tm.texi: Regenerate.
5595 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5597 (do_jump): Use scalar_int_mode for local variables.
5598 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5599 rather than a machine_mode.
5600 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5601 (scompare_loc_descriptor_wide): Likewise.
5602 (scompare_loc_descriptor_narrow): Likewise.
5603 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5605 * except.c (sjlj_emit_dispatch_table): Likewise.
5606 (expand_builtin_eh_copy_values): Likewise.
5607 * explow.c (convert_memory_address_addr_space_1): Likewise.
5608 Take a scalar_int_mode rather than a machine_mode.
5609 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5610 than a machine_mode.
5611 (memory_address_addr_space): Use scalar_int_mode for local variables.
5612 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5613 rather than a machine_mode.
5614 * expmed.c (mask_rtx): Likewise.
5615 (init_expmed_one_conv): Likewise.
5616 (expand_mult_highpart_adjust): Likewise.
5617 (extract_high_half): Likewise.
5618 (expmed_mult_highpart_optab): Likewise.
5619 (expmed_mult_highpart): Likewise.
5620 (expand_smod_pow2): Likewise.
5621 (expand_sdiv_pow2): Likewise.
5622 (emit_store_flag_int): Likewise.
5623 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5625 (extract_low_bits): Likewise.
5626 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5628 * expr.c (pieces_addr::adjust): Likewise.
5629 (can_store_by_pieces): Likewise.
5630 (store_by_pieces): Likewise.
5631 (clear_by_pieces_1): Likewise.
5632 (expand_expr_addr_expr_1): Likewise.
5633 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5634 (expand_expr_real_1): Likewise.
5635 (try_casesi): Likewise.
5636 * final.c (shorten_branches): Likewise.
5637 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5638 type of "mode" to machine_mode.
5639 * internal-fn.c (expand_arith_overflow_result_store): Take a
5640 scalar_int_mode rather than a machine_mode.
5641 (expand_mul_overflow): Use scalar_int_mode for local variables.
5642 * loop-doloop.c (doloop_modify): Likewise.
5643 (doloop_optimize): Likewise.
5644 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5645 than a machine_mode.
5646 (expand_doubleword_shift_condmove): Likewise.
5647 (expand_doubleword_shift): Likewise.
5648 (expand_doubleword_clz): Likewise.
5649 (expand_doubleword_popcount): Likewise.
5650 (expand_doubleword_parity): Likewise.
5651 (expand_absneg_bit): Use scalar_int_mode for local variables.
5652 (prepare_float_lib_cmp): Likewise.
5653 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5654 rather than a machine_mode.
5655 (convert_memory_address_addr_space): Likewise.
5656 (get_mode_bounds): Likewise.
5657 (get_address_mode): Return a scalar_int_mode rather than a
5659 * rtlanal.c (get_address_mode): Likewise.
5660 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5661 than a machine_mode.
5662 * targhooks.c (default_mode_rep_extended): Likewise.
5663 (default_valid_pointer_mode): Likewise.
5664 (default_addr_space_valid_pointer_mode): Likewise.
5665 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5666 than a machine_mode.
5667 (default_libgcc_cmp_return_mode): Likewise.
5668 (default_libgcc_shift_count_mode): Likewise.
5669 (default_unwind_word_mode): Likewise.
5670 (default_addr_space_pointer_mode): Likewise.
5671 (default_addr_space_address_mode): Likewise.
5672 * targhooks.h (default_eh_return_filter_mode): Likewise.
5673 (default_libgcc_cmp_return_mode): Likewise.
5674 (default_libgcc_shift_count_mode): Likewise.
5675 (default_unwind_word_mode): Likewise.
5676 (default_addr_space_pointer_mode): Likewise.
5677 (default_addr_space_address_mode): Likewise.
5678 (default_mode_rep_extended): Take a scalar_int_mode rather than
5680 (default_valid_pointer_mode): Likewise.
5681 (default_addr_space_valid_pointer_mode): Likewise.
5682 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5684 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5685 rather than a machine_mode.
5686 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5687 for local variables.
5688 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5689 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5690 than a machine_mode.
5692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5693 Alan Hayward <alan.hayward@arm.com>
5694 David Sherwood <david.sherwood@arm.com>
5696 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5697 the mode argument to scalar_int_mode.
5698 (do_jump_by_parts_zero_rtx): Likewise.
5699 (do_jump_by_parts_equality_rtx): Likewise.
5700 (do_jump_by_parts_greater): Take a mode argument.
5701 (do_jump_by_parts_equality): Likewise.
5702 (do_jump_1): Update calls accordingly.
5704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5705 Alan Hayward <alan.hayward@arm.com>
5706 David Sherwood <david.sherwood@arm.com>
5708 * is-a.h (safe_dyn_cast): New function.
5709 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5710 (jump_table_for_label): Likewise.
5711 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5712 instead of an rtx_insn *.
5713 (shorten_branches): Use dyn_cast instead of LABEL_P and
5714 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5715 rtx_jump_table_data::get_data_mode.
5716 (final_scan_insn): Likewise.
5718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5719 Alan Hayward <alan.hayward@arm.com>
5720 David Sherwood <david.sherwood@arm.com>
5722 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5723 trying to combine a full-register integer set with a subreg
5726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5727 Alan Hayward <alan.hayward@arm.com>
5728 David Sherwood <david.sherwood@arm.com>
5730 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5731 that is always either address_mode or pointer_mode.
5733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5734 Alan Hayward <alan.hayward@arm.com>
5735 David Sherwood <david.sherwood@arm.com>
5737 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5738 when the two are known to be equal.
5740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5741 Alan Hayward <alan.hayward@arm.com>
5742 David Sherwood <david.sherwood@arm.com>
5744 * simplify-rtx.c (simplify_const_unary_operation): Use
5745 is_a <scalar_int_mode> instead of checking for a nonzero
5746 precision. Forcibly convert op_mode to a scalar_int_mode
5747 in that case. More clearly differentiate the operand and
5748 result modes and use the former when deciding what the value
5749 of a count-bits operation should be. Use is_int_mode instead
5750 of checking for a MODE_INT. Remove redundant check for whether
5751 this mode has a zero precision.
5753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5754 Alan Hayward <alan.hayward@arm.com>
5755 David Sherwood <david.sherwood@arm.com>
5757 * optabs.c (widen_leading): Change the type of the mode argument
5758 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5759 (widen_bswap): Likewise.
5760 (expand_parity): Likewise.
5761 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5762 (expand_ffs): Likewise.
5763 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5772 (expand_expr_real_1): Update call accordingly.
5774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5775 Alan Hayward <alan.hayward@arm.com>
5776 David Sherwood <david.sherwood@arm.com>
5778 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5779 value_mode arguments. Use scalar_int_mode internally.
5780 (store_bit_field_1): Rename the new integer mode from imode
5781 to op0_mode and use it instead of GET_MODE (op0). Update calls
5782 to store_split_bit_field, store_bit_field_using_insv and
5783 store_fixed_bit_field.
5784 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5785 Use scalar_int_mode internally. Use a bit count rather than a mode
5786 when calculating the largest bit size for get_best_mode.
5787 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5788 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5789 Remove assertion that OP0 has a scalar integer mode.
5790 (store_split_bit_field): Add op0_mode and value_mode arguments.
5791 Update calls to extract_fixed_bit_field.
5792 (extract_bit_field_using_extv): Add an op0_mode argument.
5793 Use scalar_int_mode internally.
5794 (extract_bit_field_1): Rename the new integer mode from imode to
5795 op0_mode and use it instead of GET_MODE (op0). Update calls to
5796 extract_split_bit_field, extract_bit_field_using_extv and
5797 extract_fixed_bit_field.
5798 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5799 to extract_split_bit_field and extract_fixed_bit_field_1.
5800 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5801 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5803 (extract_split_bit_field): Add an op0_mode argument. Update call
5804 to extract_fixed_bit_field.
5806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5807 Alan Hayward <alan.hayward@arm.com>
5808 David Sherwood <david.sherwood@arm.com>
5810 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5811 * explow.c (hard_function_value): Likewise.
5812 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5813 convert_to_mode call outside the loop.
5814 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5815 for the mode iterator. Require the mode specified by max_pieces
5817 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5819 (copy_blkmode_to_reg): Likewise.
5820 (set_storage_via_setmem): Likewise.
5821 * optabs.c (prepare_cmp_insn): Likewise.
5822 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5823 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5830 * expr.c (convert_move): Use them.
5831 (convert_modes): Likewise.
5832 (store_expr_with_bounds): Likewise.
5834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5835 Alan Hayward <alan.hayward@arm.com>
5836 David Sherwood <david.sherwood@arm.com>
5838 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5839 parameter for the mode of "x". Remove the "known_x", "known_mode"
5840 and "known_ret" arguments. Change the type of the mode argument
5842 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
5843 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
5844 (reg_num_sign_bit_copies_for_combine): Likewise.
5845 * rtlanal.c (nonzero_bits1): Likewise.
5846 (num_sign_bit_copies1): Likewise.
5847 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
5848 (reg_num_sign_bit_copies_general): Likewise.
5849 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
5850 (reg_nonzero_bits_general): Likewise.
5852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5853 Alan Hayward <alan.hayward@arm.com>
5854 David Sherwood <david.sherwood@arm.com>
5856 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
5857 than in subroutines. Return 1 for non-integer modes.
5858 (cached_num_sign_bit_copies): Change the type of the mode parameter
5860 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
5861 classes. Handle CONST_INT_P first and then check whether X also
5862 has a scalar integer mode. Check the same thing for inner registers
5863 of a SUBREG and for values that are being extended or truncated.
5865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5866 Alan Hayward <alan.hayward@arm.com>
5867 David Sherwood <david.sherwood@arm.com>
5869 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
5870 in subroutines. Return the mode mask for non-integer modes.
5871 (cached_nonzero_bits): Change the type of the mode parameter
5873 (nonzero_bits1): Likewise. Remove early exit for other mode
5874 classes. Handle CONST_INT_P first and then check whether X
5875 also has a scalar integer mode.
5877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5878 Alan Hayward <alan.hayward@arm.com>
5879 David Sherwood <david.sherwood@arm.com>
5881 * expr.c (widest_int_mode_for_size): Make the comment match the code.
5882 Return a scalar_int_mode and assert that the size is greater than
5884 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
5885 (op_by_pieces_d::op_by_pieces_d): Likewise.
5886 (op_by_pieces_d::run): Likewise.
5887 (can_store_by_pieces): Likewise.
5889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5890 Alan Hayward <alan.hayward@arm.com>
5891 David Sherwood <david.sherwood@arm.com>
5893 * combine.c (extract_left_shift): Add a mode argument and update
5895 (make_compound_operation_int): Change the type of the mode parameter
5896 to scalar_int_mode and update the call to extract_left_shift.
5898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5899 Alan Hayward <alan.hayward@arm.com>
5900 David Sherwood <david.sherwood@arm.com>
5902 * combine.c (simplify_and_const_int): Change the type of the mode
5903 parameter to scalar_int_mode.
5904 (simplify_and_const_int_1): Likewise. Update recursive call.
5906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5907 Alan Hayward <alan.hayward@arm.com>
5908 David Sherwood <david.sherwood@arm.com>
5910 * combine.c (simplify_compare_const): Check that the mode is a
5911 scalar_int_mode (rather than VOIDmode) before testing its
5913 (simplify_comparison): Move COMPARISON_P handling out of the
5914 loop and restrict the latter part of the loop to scalar_int_modes.
5915 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
5916 and when considering SUBREG_REGs. Use is_int_mode instead of
5917 checking GET_MODE_CLASS against MODE_INT.
5919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5920 Alan Hayward <alan.hayward@arm.com>
5921 David Sherwood <david.sherwood@arm.com>
5923 * combine.c (try_widen_shift_mode): Move check for equal modes to...
5924 (simplify_shift_const_1): ...here. Use scalar_int_mode for
5925 shift_unit_mode and for modes involved in scalar shifts.
5927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5928 Alan Hayward <alan.hayward@arm.com>
5929 David Sherwood <david.sherwood@arm.com>
5931 * combine.c (force_int_to_mode): New function, split out from...
5932 (force_to_mode): ...here. Keep xmode up-to-date and use it
5933 instead of GET_MODE (x).
5935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5936 Alan Hayward <alan.hayward@arm.com>
5937 David Sherwood <david.sherwood@arm.com>
5939 * optabs-query.h (extraction_insn::struct_mode): Change type to
5940 opt_scalar_int_mode and update comment.
5941 (extraction_insn::field_mode): Change type to scalar_int_mode.
5942 (extraction_insn::pos_mode): Likewise.
5943 * combine.c (make_extraction): Update accordingly.
5944 * optabs-query.c (get_traditional_extraction_insn): Likewise.
5945 (get_optab_extraction_insn): Likewise.
5946 * recog.c (simplify_while_replacing): Likewise.
5947 * expmed.c (narrow_bit_field_mem): Change the type of the mode
5948 parameter to opt_scalar_int_mode.
5950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5951 Alan Hayward <alan.hayward@arm.com>
5952 David Sherwood <david.sherwood@arm.com>
5954 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
5955 to a scalar_int_mode instead of a machine_mode.
5956 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
5957 (get_best_mode): Return a boolean and use a pointer argument to store
5958 the selected mode. Replace the limit mode parameter with a bit limit.
5959 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
5960 for the values returned by bit_field_mode_iterator::next_mode.
5961 (store_bit_field): Update call to get_best_mode.
5962 (store_fixed_bit_field): Likewise.
5963 (extract_fixed_bit_field): Likewise.
5964 * expr.c (optimize_bitfield_assignment_op): Likewise.
5965 * fold-const.c (optimize_bit_field_compare): Likewise.
5966 (fold_truth_andor_1): Likewise.
5967 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
5968 Update for new type of m_mode.
5969 (get_best_mode): As above.
5971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5972 Alan Hayward <alan.hayward@arm.com>
5973 David Sherwood <david.sherwood@arm.com>
5975 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
5976 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
5977 (store_bit_field): Check is_a <scalar_int_mode> before calling
5978 strict_volatile_bitfield_p.
5979 (extract_bit_field): Likewise.
5981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5982 Alan Hayward <alan.hayward@arm.com>
5983 David Sherwood <david.sherwood@arm.com>
5985 * target.def (cstore_mode): Return a scalar_int_mode.
5986 * doc/tm.texi: Regenerate.
5987 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
5988 * targhooks.h (default_cstore_mode): Likewise.
5989 * targhooks.c (default_cstore_mode): Likewise, using a forced
5991 * expmed.c (emit_cstore): Expect the target of the cstore to be
5994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5995 Alan Hayward <alan.hayward@arm.com>
5996 David Sherwood <david.sherwood@arm.com>
5998 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6000 (niter_desc): Likewise mode.
6001 (iv_analyze): Add a mode parameter.
6003 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6004 and change its type to scalar_int_mode.
6005 * loop-iv.c: Update commentary at head of file.
6006 (iv_constant): Pass the mode paraeter before the rtx it describes
6007 and change its type to scalar_int_mode. Remove VOIDmode handling.
6008 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6009 (iv_extend): Likewise.
6010 (shorten_into_mode): Likewise.
6011 (iv_add): Use scalar_int_mode.
6012 (iv_mult): Likewise.
6013 (iv_shift): Likewise.
6014 (canonicalize_iv_subregs): Likewise.
6015 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6016 it describes and change its mode to scalar_int_mode. Also change
6017 the type of the returned inner_mode to scalar_int_mode.
6018 (get_biv_step): Likewise, turning outer_mode from a pointer
6019 into a direct parameter. Update call to get_biv_step_1.
6020 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6021 iv_constant and get_biv_step.
6022 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6023 and change its type to scalar_int_mode. Don't initialise iv->mode
6024 to VOIDmode and remove later checks for its still being VOIDmode.
6025 Update calls to iv_analyze_op and iv_analyze_expr. Check
6026 is_a <scalar_int_mode> when changing the mode under consideration.
6027 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6028 Update call to iv_analyze_expr.
6029 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6030 inner register is not also a scalar_int_mode. Update call to
6032 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6033 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6034 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6035 separate mode class checks. Update calls to iv_analyze. Remove
6036 fix-up of VOIDmodes after iv_analyze_biv.
6037 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6038 don't have a scalar_int_mode. Update call to biv_p.
6040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6041 Alan Hayward <alan.hayward@arm.com>
6042 David Sherwood <david.sherwood@arm.com>
6044 * cfgexpand.c (convert_debug_memory_address): Use
6045 as_a <scalar_int_mode>.
6046 * combine.c (expand_compound_operation): Likewise.
6047 (make_extraction): Likewise.
6048 (change_zero_ext): Likewise.
6049 (simplify_comparison): Likewise.
6050 * cse.c (cse_insn): Likewise.
6051 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6052 (mem_loc_descriptor): Likewise.
6053 (loc_descriptor): Likewise.
6054 * expmed.c (init_expmed_one_mode): Likewise.
6055 (synth_mult): Likewise.
6056 (emit_store_flag_1): Likewise.
6057 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6058 of a comparison with size.
6059 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6060 (reduce_to_bit_field_precision): Likewise.
6061 * function.c (expand_function_end): Likewise.
6062 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6063 * loop-doloop.c (doloop_modify): Likewise.
6064 * optabs.c (expand_binop): Likewise.
6065 (expand_unop): Likewise.
6066 (expand_copysign_absneg): Likewise.
6067 (prepare_cmp_insn): Likewise.
6068 (maybe_legitimize_operand): Likewise.
6069 * recog.c (const_scalar_int_operand): Likewise.
6070 * rtlanal.c (get_address_mode): Likewise.
6071 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6072 (simplify_cond_clz_ctz): Likewise.
6073 * tree-nested.c (get_nl_goto_field): Likewise.
6074 * tree.c (build_vector_type_for_mode): Likewise.
6075 * var-tracking.c (use_narrower_mode): Likewise.
6077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6078 Alan Hayward <alan.hayward@arm.com>
6079 David Sherwood <david.sherwood@arm.com>
6081 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6082 * builtins.c (expand_builtin_signbit): Use it.
6083 * cfgexpand.c (expand_debug_expr): Likewise.
6084 * dojump.c (do_jump): Likewise.
6085 (do_compare_and_jump): Likewise.
6086 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6087 * expmed.c (make_tree): Likewise.
6088 * expr.c (expand_expr_real_2): Likewise.
6089 (expand_expr_real_1): Likewise.
6090 (try_casesi): Likewise.
6091 * fold-const-call.c (fold_const_call_ss): Likewise.
6092 * fold-const.c (unextend): Likewise.
6093 (extract_muldiv_1): Likewise.
6094 (fold_single_bit_test): Likewise.
6095 (native_encode_int): Likewise.
6096 (native_encode_string): Likewise.
6097 (native_interpret_int): Likewise.
6098 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6099 * internal-fn.c (expand_addsub_overflow): Likewise.
6100 (expand_neg_overflow): Likewise.
6101 (expand_mul_overflow): Likewise.
6102 (expand_arith_overflow): Likewise.
6103 * match.pd: Likewise.
6104 * stor-layout.c (layout_type): Likewise.
6105 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6106 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6107 * tree-ssanames.c (get_range_info): Likewise.
6108 * tree-switch-conversion.c (array_value_type) Likewise.
6109 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6110 (vect_recog_divmod_pattern): Likewise.
6111 (vect_recog_mixed_size_cond_pattern): Likewise.
6112 * tree-vrp.c (extract_range_basic): Likewise.
6113 (simplify_float_conversion_using_ranges): Likewise.
6114 * tree.c (int_fits_type_p): Likewise.
6115 * ubsan.c (instrument_bool_enum_load): Likewise.
6116 * varasm.c (mergeable_string_section): Likewise.
6117 (narrowing_initializer_constant_valid_p): Likewise.
6118 (output_constant): 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 (NARROWEST_INT_MODE): New macro.
6125 * expr.c (alignment_for_piecewise_move): Use it instead of
6126 GET_CLASS_NARROWEST_MODE (MODE_INT).
6127 (push_block): Likewise.
6128 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6130 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6133 Alan Hayward <alan.hayward@arm.com>
6134 David Sherwood <david.sherwood@arm.com>
6136 * postreload.c (move2add_valid_value_p): Change the type of the
6137 mode parameter to scalar_int_mode.
6138 (move2add_use_add2_insn): Add a mode parameter and use it instead
6140 (move2add_use_add3_insn): Likewise.
6141 (reload_cse_move2add): Update accordingly.
6143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6144 Alan Hayward <alan.hayward@arm.com>
6145 David Sherwood <david.sherwood@arm.com>
6147 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6149 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6150 * optabs.c (expand_unop): Likewise.
6152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6153 Alan Hayward <alan.hayward@arm.com>
6154 David Sherwood <david.sherwood@arm.com>
6156 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6157 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6158 (popcount_loc_descriptor): Likewise.
6159 (bswap_loc_descriptor): Likewise.
6160 (rotate_loc_descriptor): Likewise.
6161 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6162 calling the functions above.
6164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6165 Alan Hayward <alan.hayward@arm.com>
6166 David Sherwood <david.sherwood@arm.com>
6168 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6170 (try_combine): Likewise.
6171 (simplify_if_then_else): Likewise.
6172 * cse.c (cse_insn): Likewise.
6173 * dwarf2out.c (mem_loc_descriptor): Likewise.
6174 * emit-rtl.c (gen_lowpart_common): Likewise.
6175 * simplify-rtx.c (simplify_truncation): Likewise.
6176 (simplify_binary_operation_1): Likewise.
6177 (simplify_const_relational_operation): Likewise.
6178 (simplify_ternary_operation): Likewise.
6179 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6182 Alan Hayward <alan.hayward@arm.com>
6183 David Sherwood <david.sherwood@arm.com>
6185 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6186 * reload.c (push_reload): Likewise.
6187 (find_reloads): 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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6194 (make_compound_operation_int): Likewise.
6195 (change_zero_ext): Likewise.
6196 * expr.c (convert_move): Likewise.
6197 (convert_modes): Likewise.
6198 * fwprop.c (forward_propagate_subreg): Likewise.
6199 * loop-iv.c (get_biv_step_1): Likewise.
6200 * optabs.c (widen_operand): Likewise.
6201 * postreload.c (move2add_valid_value_p): Likewise.
6202 * recog.c (simplify_while_replacing): Likewise.
6203 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6204 (simplify_binary_operation_1): Likewise. Remove redundant
6205 mode equality check.
6207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6208 Alan Hayward <alan.hayward@arm.com>
6209 David Sherwood <david.sherwood@arm.com>
6211 * combine.c (combine_simplify_rtx): Add checks for
6212 is_a <scalar_int_mode>.
6213 (simplify_if_then_else): Likewise.
6214 (make_field_assignment): Likewise.
6215 (simplify_comparison): Likewise.
6216 * ifcvt.c (noce_try_bitop): Likewise.
6217 * loop-invariant.c (canonicalize_address_mult): Likewise.
6218 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6221 Alan Hayward <alan.hayward@arm.com>
6222 David Sherwood <david.sherwood@arm.com>
6224 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6225 is_a <scalar_int_mode> instead of != BLKmode.
6227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6228 Alan Hayward <alan.hayward@arm.com>
6229 David Sherwood <david.sherwood@arm.com>
6231 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6232 instead of != VOIDmode.
6233 * combine.c (if_then_else_cond): Likewise.
6234 (change_zero_ext): Likewise.
6235 * dwarf2out.c (mem_loc_descriptor): Likewise.
6236 (loc_descriptor): Likewise.
6237 * rtlanal.c (canonicalize_condition): Likewise.
6238 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6241 Alan Hayward <alan.hayward@arm.com>
6242 David Sherwood <david.sherwood@arm.com>
6244 * simplify-rtx.c (simplify_binary_operation_1): Use
6245 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6248 Alan Hayward <alan.hayward@arm.com>
6249 David Sherwood <david.sherwood@arm.com>
6251 * wide-int.h (int_traits<unsigned char>) New class.
6252 (int_traits<unsigned short>) Likewise.
6253 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6254 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6256 * combine.c (set_nonzero_bits_and_sign_copies): Use
6257 is_a <scalar_int_mode>.
6258 (find_split_point): Likewise.
6259 (combine_simplify_rtx): Likewise.
6260 (simplify_logical): Likewise.
6261 (expand_compound_operation): Likewise.
6262 (expand_field_assignment): Likewise.
6263 (make_compound_operation): Likewise.
6264 (extended_count): Likewise.
6265 (change_zero_ext): Likewise.
6266 (simplify_comparison): Likewise.
6267 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6268 (ucompare_loc_descriptor): Likewise.
6269 (minmax_loc_descriptor): Likewise.
6270 (mem_loc_descriptor): Likewise.
6271 (loc_descriptor): Likewise.
6272 * expmed.c (init_expmed_one_mode): Likewise.
6273 * lra-constraints.c (lra_constraint_offset): Likewise.
6274 * optabs.c (prepare_libcall_arg): Likewise.
6275 * postreload.c (move2add_note_store): Likewise.
6276 * reload.c (operands_match_p): Likewise.
6277 * rtl.h (load_extend_op): Likewise.
6278 * rtlhooks.c (gen_lowpart_general): Likewise.
6279 * simplify-rtx.c (simplify_truncation): Likewise.
6280 (simplify_unary_operation_1): Likewise.
6281 (simplify_binary_operation_1): Likewise.
6282 (simplify_const_binary_operation): Likewise.
6283 (simplify_const_relational_operation): Likewise.
6284 (simplify_subreg): Likewise.
6285 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6286 * var-tracking.c (adjust_mems): Likewise.
6287 (prepare_call_arguments): Likewise.
6289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6290 Alan Hayward <alan.hayward@arm.com>
6291 David Sherwood <david.sherwood@arm.com>
6293 * machmode.h (is_int_mode): New fuction.
6294 * combine.c (find_split_point): Use it.
6295 (combine_simplify_rtx): Likewise.
6296 (simplify_if_then_else): Likewise.
6297 (simplify_set): Likewise.
6298 (simplify_shift_const_1): Likewise.
6299 (simplify_comparison): Likewise.
6300 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6301 * cse.c (notreg_cost): Likewise.
6302 (cse_insn): Likewise.
6303 * cselib.c (cselib_lookup_1): Likewise.
6304 * dojump.c (do_jump_1): Likewise.
6305 (do_compare_rtx_and_jump): Likewise.
6306 * dse.c (get_call_args): Likewise.
6307 * dwarf2out.c (rtl_for_decl_init): Likewise.
6308 (native_encode_initializer): Likewise.
6309 * expmed.c (emit_store_flag_1): Likewise.
6310 (emit_store_flag): Likewise.
6311 * expr.c (convert_modes): Likewise.
6312 (store_field): Likewise.
6313 (expand_expr_real_1): Likewise.
6314 * fold-const.c (fold_read_from_constant_string): Likewise.
6315 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6316 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6317 * optabs.c (expand_binop): Likewise.
6318 (expand_unop): Likewise.
6319 (expand_abs_nojump): Likewise.
6320 (expand_one_cmpl_abs_nojump): Likewise.
6321 * simplify-rtx.c (mode_signbit_p): Likewise.
6322 (val_signbit_p): Likewise.
6323 (val_signbit_known_set_p): Likewise.
6324 (val_signbit_known_clear_p): Likewise.
6325 (simplify_relational_operation_1): Likewise.
6326 * tree.c (vector_type_mode): Likewise.
6328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6329 Alan Hayward <alan.hayward@arm.com>
6330 David Sherwood <david.sherwood@arm.com>
6332 * machmode.h (smallest_mode_for_size): Fix formatting.
6333 (smallest_int_mode_for_size): New function.
6334 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6335 instead of smallest_mode_for_size.
6336 * combine.c (make_extraction): Likewise.
6337 * config/arc/arc.c (arc_expand_movmem): Likewise.
6338 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6339 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6340 * config/s390/s390.c (s390_expand_insv): Likewise.
6341 * config/sparc/sparc.c (assign_int_registers): Likewise.
6342 * config/spu/spu.c (spu_function_value): Likewise.
6343 (spu_function_arg): Likewise.
6344 * coverage.c (get_gcov_type): Likewise.
6345 (get_gcov_unsigned_t): Likewise.
6346 * dse.c (find_shift_sequence): Likewise.
6347 * expmed.c (store_bit_field_1): Likewise.
6348 * expr.c (convert_move): Likewise.
6349 (store_field): Likewise.
6350 * internal-fn.c (expand_arith_overflow): Likewise.
6351 * optabs-query.c (get_best_extraction_insn): Likewise.
6352 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6353 * stor-layout.c (layout_type): Likewise.
6354 (initialize_sizetypes): Likewise.
6355 * targhooks.c (default_get_mask_mode): Likewise.
6356 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6359 Alan Hayward <alan.hayward@arm.com>
6360 David Sherwood <david.sherwood@arm.com>
6362 * machmode.h (opt_mode::else_blk): New function.
6363 (int_mode_for_mode): Declare.
6364 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6365 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6367 * cfgexpand.c (expand_debug_expr): Likewise.
6368 * combine.c (gen_lowpart_or_truncate): Likewise.
6369 (gen_lowpart_for_combine): Likewise.
6370 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6371 * config/avr/avr.c (avr_to_int_mode): Likewise.
6372 (avr_out_plus_1): Likewise.
6373 (avr_out_plus): Likewise.
6374 (avr_out_round): Likewise.
6375 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6376 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6377 (s390_expand_vcond): Likewise.
6378 * config/spu/spu.c (spu_split_immediate): Likewise.
6379 (spu_expand_mov): Likewise.
6380 * dse.c (get_stored_val): Likewise.
6381 * expmed.c (store_bit_field_1): Likewise.
6382 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6384 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6385 (extract_low_bits): Likewise.
6386 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6387 handling rather than repeating the check.
6388 (emit_group_store): Likewise.
6389 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6390 * optabs.c (expand_absneg_bit): Likewise.
6391 (expand_copysign_absneg): Likewise.
6392 (expand_copysign_bit): Likewise.
6393 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6394 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6395 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6396 * var-tracking.c (prepare_call_arguments): Likewise.
6397 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6398 int_mode_for_mode instead of mode_for_size.
6399 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6402 Alan Hayward <alan.hayward@arm.com>
6403 David Sherwood <david.sherwood@arm.com>
6405 * machmode.h (int_mode_for_size): New function.
6406 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6407 instead of mode_for_size.
6408 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6410 * combine.c (expand_field_assignment): Use int_mode_for_size
6411 instead of mode_for_size.
6412 (make_extraction): Likewise.
6413 (simplify_shift_const_1): Likewise.
6414 (simplify_comparison): Likewise.
6415 * dojump.c (do_jump): Likewise.
6416 * dwarf2out.c (mem_loc_descriptor): Likewise.
6417 * emit-rtl.c (init_derived_machine_modes): Likewise.
6418 * expmed.c (flip_storage_order): Likewise.
6419 (convert_extracted_bit_field): Likewise.
6420 * expr.c (copy_blkmode_from_reg): Likewise.
6421 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6422 * internal-fn.c (expand_mul_overflow): Likewise.
6423 * lower-subreg.c (simple_move): Likewise.
6424 * optabs-libfuncs.c (init_optabs): Likewise.
6425 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6426 * tree.c (vector_type_mode): Likewise.
6427 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6428 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6429 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6430 * tree-vect-stmts.c (vectorizable_load): Likewise.
6431 (vectorizable_store): Likewise.
6433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6434 Alan Hayward <alan.hayward@arm.com>
6435 David Sherwood <david.sherwood@arm.com>
6437 * coretypes.h (pod_mode): New type.
6438 (scalar_int_mode_pod): New typedef.
6439 * machmode.h (pod_mode): New class.
6440 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6441 * genmodes.c (emit_mode_int_n): Update accordingly.
6442 * lower-subreg.h (target_lower_subreg): Change type to
6443 scalar_int_mode_pod.
6444 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6445 scalar_int_mode_pod.
6447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6448 Alan Hayward <alan.hayward@arm.com>
6449 David Sherwood <david.sherwood@arm.com>
6451 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6452 machine_mode to scalar_int_mode.
6453 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6454 (rs6000_option_override_internal): Remove cast to int.
6455 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6456 machine_mode to scalar_int_mode.
6457 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6458 (rs6000_option_override_internal): Remove cast to int.
6459 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6460 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6462 * config/s390/s390.c (s390_expand_builtin): Likewise.
6463 * coretypes.h (scalar_int_mode): New type.
6464 (opt_scalar_int_mode): New typedef.
6465 * machmode.h (scalar_int_mode): New class.
6466 (scalar_int_mode::includes_p): New function.
6467 (byte_mode): Change type to scalar_int_mode.
6468 (word_mode): Likewise.
6469 (ptr_mode): Likewise.
6470 * emit-rtl.c (byte_mode): Likewise.
6471 (word_mode): Likewise.
6472 (ptr_mode): Likewise.
6473 (init_derived_machine_modes): Update accordingly.
6474 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6475 and MODE_PARTIAL_INT.
6476 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6477 opt_scalar_int_mode.
6479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6480 Alan Hayward <alan.hayward@arm.com>
6481 David Sherwood <david.sherwood@arm.com>
6483 * target.def (libgcc_floating_mode_supported_p): Take a
6485 * doc/tm.texi: Regenerate.
6486 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6488 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6489 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6493 Alan Hayward <alan.hayward@arm.com>
6494 David Sherwood <david.sherwood@arm.com>
6496 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6497 * doc/tm.texi: Regenerate.
6498 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6499 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6500 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6501 * targhooks.h (default_floatn_mode): Likewise.
6502 * targhooks.c (default_floatn_mode): Likewise.
6503 * tree.c (build_common_tree_nodes): Update accordingly.
6505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6506 Alan Hayward <alan.hayward@arm.com>
6507 David Sherwood <david.sherwood@arm.com>
6509 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6510 (mode_iterator::iterate_p): Likewise.
6511 (mode_iterator::get_wider): Likewise.
6512 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6515 Alan Hayward <alan.hayward@arm.com>
6516 David Sherwood <david.sherwood@arm.com>
6518 * coretypes.h (opt_scalar_float_mode): New typedef.
6519 * machmode.h (float_mode_for_size): New function.
6520 * emit-rtl.c (double_mode): Delete.
6521 (init_emit_once): Use float_mode_for_size.
6522 * stor-layout.c (layout_type): Likewise.
6523 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6526 Alan Hayward <alan.hayward@arm.com>
6527 David Sherwood <david.sherwood@arm.com>
6529 * output.h (assemble_real): Take a scalar_float_mode.
6530 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6531 * config/arm/arm.md (consttable_4): Likewise.
6532 (consttable_8): Likewise.
6533 (consttable_16): Likewise.
6534 * config/mips/mips.md (consttable_float): Likewise.
6535 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6536 * varasm.c (assemble_real): Take a scalar_float_mode.
6537 (output_constant_pool_2): Update accordingly.
6538 (output_constant): Likewise.
6540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6541 Alan Hayward <alan.hayward@arm.com>
6542 David Sherwood <david.sherwood@arm.com>
6544 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6545 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6546 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6547 (native_encode_real): Likewise.
6548 (native_interpret_real): Likewise.
6549 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6550 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6553 Alan Hayward <alan.hayward@arm.com>
6554 David Sherwood <david.sherwood@arm.com>
6556 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6557 <scalar_float_mode>. Simplify.
6558 (gen_extend_conv_libfunc): Likewise.
6560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6561 Alan Hayward <alan.hayward@arm.com>
6562 David Sherwood <david.sherwood@arm.com>
6564 * coretypes.h (scalar_float_mode): New type.
6565 * machmode.h (mode_traits::from_int): Use machine_mode if
6566 USE_ENUM_MODES is defined.
6567 (is_a): New function.
6569 (dyn_cast): Likewise.
6570 (scalar_float_mode): New class.
6571 (scalar_float_mode::includes_p): New function.
6572 (is_float_mode): Likewise.
6573 * gdbhooks.py (MachineModePrinter): New class.
6574 (build_pretty_printer): Use it for scalar_float_mode.
6575 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6576 (format_helper::format_helper): Turn into a template.
6577 * genmodes.c (get_mode_class): New function.
6578 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6579 or machine_mode if none.
6580 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6581 as_a <scalar_float_mode>.
6582 * dwarf2out.c (mem_loc_descriptor): Likewise.
6583 (insert_float): Likewise.
6584 (add_const_value_attribute): Likewise.
6585 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6586 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6587 (expand_unop): Update accordingly.
6588 (expand_abs_nojump): Likewise.
6589 (expand_copysign_absneg): Take a scalar_float_mode.
6590 (expand_copysign_bit): Likewise.
6591 (expand_copysign): Update accordingly.
6593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6594 Alan Hayward <alan.hayward@arm.com>
6595 David Sherwood <david.sherwood@arm.com>
6597 * coretypes.h (opt_mode): New class.
6598 * machmode.h (opt_mode): Likewise.
6599 (opt_mode::else_void): New function.
6600 (opt_mode::require): Likewise.
6601 (opt_mode::exists): Likewise.
6602 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6603 (GET_MODE_2XWIDER_MODE): Likewise.
6604 (mode_iterator::get_wider): Update accordingly.
6605 (mode_iterator::get_2xwider): Likewise.
6606 (mode_iterator::get_known_wider): Likewise, turning into a template.
6607 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6608 forcing a wider mode to exist.
6609 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6610 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6611 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6612 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6613 * lower-subreg.c (init_lower_subreg): Likewise.
6614 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6615 on the final iteration.
6616 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6617 a wider mode exists before asking for a move pattern.
6618 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6619 forcing a wider mode to exist.
6620 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6621 returning false if no such mode exists.
6622 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6623 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6624 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6625 Avoid checking for a MODE_INT if we already know the mode is not a
6627 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6628 forcing a wider mode to exist.
6629 (expmed_mult_highpart_optab): Likewise.
6630 (expmed_mult_highpart): Likewise.
6631 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6633 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6634 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6635 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6636 * internal-fn.c (expand_mul_overflow): Update use of
6637 GET_MODE_2XWIDER_MODE.
6638 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6639 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6640 GET_MODE_WIDER_MODE.
6641 (convert_plusminus_to_widen): Likewise.
6642 * tree-switch-conversion.c (array_value_type): Likewise.
6643 * var-tracking.c (emit_note_insn_var_location): Likewise.
6644 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6645 Return false inside rather than outside the loop if no wider mode
6647 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6648 and GET_MODE_2XWIDER_MODE
6649 (can_compare_p): Use else_void.
6650 * gdbhooks.py (OptMachineModePrinter): New class.
6651 (build_pretty_printer): Use it for opt_mode.
6653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6654 Alan Hayward <alan.hayward@arm.com>
6655 David Sherwood <david.sherwood@arm.com>
6657 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6658 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6661 Alan Hayward <alan.hayward@arm.com>
6662 David Sherwood <david.sherwood@arm.com>
6664 * machmode.h (mode_traits): New structure.
6665 (get_narrowest_mode): New function.
6666 (mode_iterator::start): Likewise.
6667 (mode_iterator::iterate_p): Likewise.
6668 (mode_iterator::get_wider): Likewise.
6669 (mode_iterator::get_known_wider): Likewise.
6670 (mode_iterator::get_2xwider): Likewise.
6671 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6672 (FOR_EACH_MODE): Likewise.
6673 (FOR_EACH_MODE_FROM): Likewise.
6674 (FOR_EACH_MODE_UNTIL): Likewise.
6675 (FOR_EACH_WIDER_MODE): Likewise.
6676 (FOR_EACH_2XWIDER_MODE): Likewise.
6677 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6678 * combine.c (simplify_comparison): Likewise
6679 * config/i386/i386.c (type_natural_mode): Likewise.
6680 * cse.c (cse_insn): Likewise.
6681 * dse.c (find_shift_sequence): Likewise.
6682 * emit-rtl.c (init_derived_machine_modes): Likewise.
6683 (init_emit_once): Likewise.
6684 * explow.c (hard_function_value): Likewise.
6685 * expmed.c (extract_fixed_bit_field_1): Likewise.
6686 (extract_bit_field_1): Likewise.
6687 (expand_divmod): Likewise.
6688 (emit_store_flag_1): Likewise.
6689 * expr.c (init_expr_target): Likewise.
6690 (convert_move): Likewise.
6691 (alignment_for_piecewise_move): Likewise.
6692 (widest_int_mode_for_size): Likewise.
6693 (emit_block_move_via_movmem): Likewise.
6694 (copy_blkmode_to_reg): Likewise.
6695 (set_storage_via_setmem): Likewise.
6696 (compress_float_constant): Likewise.
6697 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6698 * optabs-query.c (get_best_extraction_insn): Likewise.
6699 * optabs.c (expand_binop): Likewise.
6700 (expand_twoval_unop): Likewise.
6701 (expand_twoval_binop): Likewise.
6702 (widen_leading): Likewise.
6703 (widen_bswap): Likewise.
6704 (expand_parity): Likewise.
6705 (expand_unop): Likewise.
6706 (prepare_cmp_insn): Likewise.
6707 (prepare_float_lib_cmp): Likewise.
6708 (expand_float): Likewise.
6709 (expand_fix): Likewise.
6710 (expand_sfix_optab): Likewise.
6711 * postreload.c (move2add_use_add2_insn): Likewise.
6712 * reg-stack.c (reg_to_stack): Likewise.
6713 * reginfo.c (choose_hard_reg_mode): Likewise.
6714 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6715 * stor-layout.c (mode_for_size): Likewise.
6716 (smallest_mode_for_size): Likewise.
6717 (mode_for_vector): Likewise.
6718 (finish_bitfield_representative): Likewise.
6719 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6720 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6721 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6722 * var-tracking.c (prepare_call_arguments): Likewise.
6724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6725 Alan Hayward <alan.hayward@arm.com>
6726 David Sherwood <david.sherwood@arm.com>
6728 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6729 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6730 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6731 * machmode.h (mode_size): Move earlier in file.
6732 (mode_precision): Likewise.
6733 (mode_inner): Likewise.
6734 (mode_nunits): Likewise.
6735 (mode_unit_size): Likewise.
6736 (unit_unit_precision): Likewise.
6737 (mode_wider): Likewise.
6738 (mode_2xwider): Likewise.
6739 (machine_mode): New class.
6740 (mode_to_bytes): New function.
6741 (mode_to_bits): Likewise.
6742 (mode_to_precision): Likewise.
6743 (mode_to_inner): Likewise.
6744 (mode_to_unit_size): Likewise.
6745 (mode_to_unit_precision): Likewise.
6746 (mode_to_nunits): Likewise.
6747 (GET_MODE_SIZE): Use mode_to_bytes.
6748 (GET_MODE_BITSIZE): Use mode_to_bits.
6749 (GET_MODE_PRECISION): Use mode_to_precision.
6750 (GET_MODE_INNER): Use mode_to_inner.
6751 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6752 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6753 (GET_MODE_NUNITS): Use mode_to_nunits.
6754 * system.h (ALWAYS_INLINE): New macro.
6755 * config/powerpcspe/powerpcspe-c.c
6756 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6757 int for arg1_mode and arg2_mode.
6759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6760 Alan Hayward <alan.hayward@arm.com>
6761 David Sherwood <david.sherwood@arm.com>
6763 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6764 Prefix mode names with E_ in case statements.
6765 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6766 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6767 (aarch64_split_simd_move): Likewise.
6768 (aarch64_gen_storewb_pair): Likewise.
6769 (aarch64_gen_loadwb_pair): Likewise.
6770 (aarch64_gen_store_pair): Likewise.
6771 (aarch64_gen_load_pair): Likewise.
6772 (aarch64_get_condition_code_1): Likewise.
6773 (aarch64_constant_pool_reload_icode): Likewise.
6774 (get_rsqrte_type): Likewise.
6775 (get_rsqrts_type): Likewise.
6776 (get_recpe_type): Likewise.
6777 (get_recps_type): Likewise.
6778 (aarch64_gimplify_va_arg_expr): Likewise.
6779 (aarch64_simd_container_mode): Likewise.
6780 (aarch64_emit_load_exclusive): Likewise.
6781 (aarch64_emit_store_exclusive): Likewise.
6782 (aarch64_expand_compare_and_swap): Likewise.
6783 (aarch64_gen_atomic_cas): Likewise.
6784 (aarch64_emit_bic): Likewise.
6785 (aarch64_emit_atomic_swap): Likewise.
6786 (aarch64_emit_atomic_load_op): Likewise.
6787 (aarch64_evpc_trn): Likewise.
6788 (aarch64_evpc_uzp): Likewise.
6789 (aarch64_evpc_zip): Likewise.
6790 (aarch64_evpc_ext): Likewise.
6791 (aarch64_evpc_rev): Likewise.
6792 (aarch64_evpc_dup): Likewise.
6793 (aarch64_gen_ccmp_first): Likewise.
6794 (aarch64_gen_ccmp_next): Likewise.
6795 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6796 (alpha_emit_xfloating_libcall): Likewise.
6797 (emit_insxl): Likewise.
6798 (alpha_arg_type): Likewise.
6799 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6800 (arc_preferred_simd_mode): Likewise.
6801 (arc_secondary_reload): Likewise.
6802 (get_arc_condition_code): Likewise.
6803 (arc_print_operand): Likewise.
6804 (arc_legitimate_constant_p): Likewise.
6805 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6806 * config/arc/arc.md (casesi_load): Likewise.
6807 (casesi_compact_jump): Likewise.
6808 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6809 (cc_use_register): Likewise.
6810 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6811 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6812 (arm_init_iwmmxt_builtins): Likewise.
6813 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6814 (neon_expand_vector_init): Likewise.
6815 (arm_attr_length_move_neon): Likewise.
6816 (maybe_get_arm_condition_code): Likewise.
6817 (arm_emit_vector_const): Likewise.
6818 (arm_preferred_simd_mode): Likewise.
6819 (arm_output_iwmmxt_tinsr): Likewise.
6820 (thumb1_output_casesi): Likewise.
6821 (thumb2_output_casesi): Likewise.
6822 (arm_emit_load_exclusive): Likewise.
6823 (arm_emit_store_exclusive): Likewise.
6824 (arm_expand_compare_and_swap): Likewise.
6825 (arm_evpc_neon_vuzp): Likewise.
6826 (arm_evpc_neon_vzip): Likewise.
6827 (arm_evpc_neon_vrev): Likewise.
6828 (arm_evpc_neon_vtrn): Likewise.
6829 (arm_evpc_neon_vext): Likewise.
6830 (arm_validize_comparison): Likewise.
6831 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6832 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6833 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6834 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6835 (c6x_preferred_simd_mode): Likewise.
6836 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6837 (epiphany_rtx_costs): Likewise.
6838 * config/epiphany/predicates.md (proper_comparison_operator):
6840 * config/frv/frv.c (condexec_memory_operand): Likewise.
6841 (frv_emit_move): Likewise.
6842 (output_move_single): Likewise.
6843 (output_condmove_single): Likewise.
6844 (frv_hard_regno_mode_ok): Likewise.
6845 (frv_matching_accg_mode): Likewise.
6846 * config/h8300/h8300.c (split_adds_subs): Likewise.
6847 (h8300_rtx_costs): Likewise.
6848 (h8300_print_operand): Likewise.
6849 (compute_mov_length): Likewise.
6850 (output_logical_op): Likewise.
6851 (compute_logical_op_length): Likewise.
6852 (compute_logical_op_cc): Likewise.
6853 (h8300_shift_needs_scratch_p): Likewise.
6854 (output_a_shift): Likewise.
6855 (compute_a_shift_length): Likewise.
6856 (compute_a_shift_cc): Likewise.
6857 (expand_a_rotate): Likewise.
6858 (output_a_rotate): Likewise.
6859 * config/i386/i386.c (classify_argument): Likewise.
6860 (function_arg_advance_32): Likewise.
6861 (function_arg_32): Likewise.
6862 (function_arg_64): Likewise.
6863 (function_value_64): Likewise.
6864 (ix86_gimplify_va_arg): Likewise.
6865 (ix86_legitimate_constant_p): Likewise.
6866 (put_condition_code): Likewise.
6867 (split_double_mode): Likewise.
6868 (ix86_avx256_split_vector_move_misalign): Likewise.
6869 (ix86_expand_vector_logical_operator): Likewise.
6870 (ix86_split_idivmod): Likewise.
6871 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
6872 (ix86_build_const_vector): Likewise.
6873 (ix86_build_signbit_mask): Likewise.
6874 (ix86_match_ccmode): Likewise.
6875 (ix86_cc_modes_compatible): Likewise.
6876 (ix86_expand_branch): Likewise.
6877 (ix86_expand_sse_cmp): Likewise.
6878 (ix86_expand_sse_movcc): Likewise.
6879 (ix86_expand_int_sse_cmp): Likewise.
6880 (ix86_expand_vec_perm_vpermi2): Likewise.
6881 (ix86_expand_vec_perm): Likewise.
6882 (ix86_expand_sse_unpack): Likewise.
6883 (ix86_expand_int_addcc): Likewise.
6884 (ix86_split_to_parts): Likewise.
6885 (ix86_vectorize_builtin_gather): Likewise.
6886 (ix86_vectorize_builtin_scatter): Likewise.
6887 (avx_vpermilp_parallel): Likewise.
6888 (inline_memory_move_cost): Likewise.
6889 (ix86_tieable_integer_mode_p): Likewise.
6890 (x86_maybe_negate_const_int): Likewise.
6891 (ix86_expand_vector_init_duplicate): Likewise.
6892 (ix86_expand_vector_init_one_nonzero): Likewise.
6893 (ix86_expand_vector_init_one_var): Likewise.
6894 (ix86_expand_vector_init_concat): Likewise.
6895 (ix86_expand_vector_init_interleave): Likewise.
6896 (ix86_expand_vector_init_general): Likewise.
6897 (ix86_expand_vector_set): Likewise.
6898 (ix86_expand_vector_extract): Likewise.
6899 (emit_reduc_half): Likewise.
6900 (ix86_emit_i387_round): Likewise.
6901 (ix86_mangle_type): Likewise.
6902 (ix86_expand_round_sse4): Likewise.
6903 (expand_vec_perm_blend): Likewise.
6904 (canonicalize_vector_int_perm): Likewise.
6905 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
6906 (expand_vec_perm_1): Likewise.
6907 (expand_vec_perm_interleave3): Likewise.
6908 (expand_vec_perm_even_odd_pack): Likewise.
6909 (expand_vec_perm_even_odd_1): Likewise.
6910 (expand_vec_perm_broadcast_1): Likewise.
6911 (ix86_vectorize_vec_perm_const_ok): Likewise.
6912 (ix86_expand_vecop_qihi): Likewise.
6913 (ix86_expand_mul_widen_hilo): Likewise.
6914 (ix86_expand_sse2_abs): Likewise.
6915 (ix86_expand_pextr): Likewise.
6916 (ix86_expand_pinsr): Likewise.
6917 (ix86_preferred_simd_mode): Likewise.
6918 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
6919 * config/i386/sse.md (*andnot<mode>3): Likewise.
6920 (<mask_codefor><code><mode>3<mask_name>): Likewise.
6921 (*<code><mode>3): Likewise.
6922 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
6923 (ia64_expand_atomic_op): Likewise.
6924 (ia64_arg_type): Likewise.
6925 (ia64_mode_to_int): Likewise.
6926 (ia64_scalar_mode_supported_p): Likewise.
6927 (ia64_vector_mode_supported_p): Likewise.
6928 (expand_vec_perm_broadcast): Likewise.
6929 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
6930 (iq2000_function_arg_advance): Likewise.
6931 (iq2000_function_arg): Likewise.
6932 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
6933 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
6934 (m68k_libcall_value): Likewise.
6935 (m68k_function_value): Likewise.
6936 (sched_attr_op_type): Likewise.
6937 * config/mcore/mcore.c (mcore_output_move): Likewise.
6938 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
6940 (microblaze_function_arg): Likewise.
6941 * config/mips/mips.c (mips16_build_call_stub): Likewise.
6942 (mips_print_operand): Likewise.
6943 (mips_mode_ok_for_mov_fmt_p): Likewise.
6944 (mips_vector_mode_supported_p): Likewise.
6945 (mips_preferred_simd_mode): Likewise.
6946 (mips_expand_vpc_loongson_even_odd): Likewise.
6947 (mips_expand_vec_unpack): Likewise.
6948 (mips_expand_vi_broadcast): Likewise.
6949 (mips_expand_vector_init): Likewise.
6950 (mips_expand_vec_reduc): Likewise.
6951 (mips_expand_msa_cmp): Likewise.
6952 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
6953 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
6954 (cc_flags_for_mode): Likewise.
6955 * config/msp430/msp430.c (msp430_print_operand): Likewise.
6956 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
6957 (nds32_output_casesi_pc_relative): Likewise.
6958 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6959 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
6960 (nvptx_gen_unpack): Likewise.
6961 (nvptx_gen_pack): Likewise.
6962 (nvptx_gen_shuffle): Likewise.
6963 (nvptx_gen_wcast): Likewise.
6964 (nvptx_preferred_simd_mode): Likewise.
6965 * config/pa/pa.c (pa_secondary_reload): Likewise.
6966 * config/pa/predicates.md (base14_operand): Likewise.
6967 * config/powerpcspe/powerpcspe-c.c
6968 (altivec_resolve_overloaded_builtin): Likewise.
6969 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
6971 (rs6000_preferred_simd_mode): Likewise.
6972 (output_vec_const_move): Likewise.
6973 (rs6000_expand_vector_extract): Likewise.
6974 (rs6000_split_vec_extract_var): Likewise.
6975 (reg_offset_addressing_ok_p): Likewise.
6976 (rs6000_legitimate_offset_address_p): Likewise.
6977 (rs6000_legitimize_address): Likewise.
6978 (rs6000_emit_set_const): Likewise.
6979 (rs6000_const_vec): Likewise.
6980 (rs6000_emit_move): Likewise.
6981 (spe_build_register_parallel): Likewise.
6982 (rs6000_darwin64_record_arg_recurse): Likewise.
6983 (swap_selector_for_mode): Likewise.
6984 (spe_init_builtins): Likewise.
6985 (paired_init_builtins): Likewise.
6986 (altivec_init_builtins): Likewise.
6987 (do_load_for_compare): Likewise.
6988 (rs6000_generate_compare): Likewise.
6989 (rs6000_expand_float128_convert): Likewise.
6990 (emit_load_locked): Likewise.
6991 (emit_store_conditional): Likewise.
6992 (rs6000_output_function_epilogue): Likewise.
6993 (rs6000_handle_altivec_attribute): Likewise.
6994 (rs6000_function_value): Likewise.
6995 (emit_fusion_gpr_load): Likewise.
6996 (emit_fusion_p9_load): Likewise.
6997 (emit_fusion_p9_store): Likewise.
6998 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
6999 (fusion_gpr_mem_load): Likewise.
7000 (fusion_addis_mem_combo_load): Likewise.
7001 (fusion_addis_mem_combo_store): Likewise.
7002 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7003 (fusion_gpr_mem_load): Likewise.
7004 (fusion_addis_mem_combo_load): Likewise.
7005 (fusion_addis_mem_combo_store): Likewise.
7006 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7008 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7009 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7010 (rs6000_preferred_simd_mode): Likewise.
7011 (output_vec_const_move): Likewise.
7012 (rs6000_expand_vector_extract): Likewise.
7013 (rs6000_split_vec_extract_var): Likewise.
7014 (reg_offset_addressing_ok_p): Likewise.
7015 (rs6000_legitimate_offset_address_p): Likewise.
7016 (rs6000_legitimize_address): Likewise.
7017 (rs6000_emit_set_const): Likewise.
7018 (rs6000_const_vec): Likewise.
7019 (rs6000_emit_move): Likewise.
7020 (rs6000_darwin64_record_arg_recurse): Likewise.
7021 (swap_selector_for_mode): Likewise.
7022 (paired_init_builtins): Likewise.
7023 (altivec_init_builtins): Likewise.
7024 (rs6000_expand_float128_convert): Likewise.
7025 (emit_load_locked): Likewise.
7026 (emit_store_conditional): Likewise.
7027 (rs6000_output_function_epilogue): Likewise.
7028 (rs6000_handle_altivec_attribute): Likewise.
7029 (rs6000_function_value): Likewise.
7030 (emit_fusion_gpr_load): Likewise.
7031 (emit_fusion_p9_load): Likewise.
7032 (emit_fusion_p9_store): Likewise.
7033 * config/rx/rx.c (rx_gen_move_template): Likewise.
7034 (flags_from_mode): Likewise.
7035 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7036 (s390_slb_comparison): Likewise.
7037 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7038 (s390_vector_mode_supported_p): Likewise.
7039 (s390_cc_modes_compatible): Likewise.
7040 (s390_match_ccmode_set): Likewise.
7041 (s390_canonicalize_comparison): Likewise.
7042 (s390_emit_compare_and_swap): Likewise.
7043 (s390_branch_condition_mask): Likewise.
7044 (s390_rtx_costs): Likewise.
7045 (s390_secondary_reload): Likewise.
7046 (__SECONDARY_RELOAD_CASE): Likewise.
7047 (s390_expand_cs): Likewise.
7048 (s390_preferred_simd_mode): Likewise.
7049 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7050 * config/sh/sh.c (sh_print_operand): Likewise.
7051 (dump_table): Likewise.
7052 (sh_secondary_reload): Likewise.
7053 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7054 * config/sh/sh.md (casesi_worker_1): Likewise.
7055 (casesi_worker_2): Likewise.
7056 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7057 (fcc_comparison_operator): Likewise.
7058 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7059 (emit_soft_tfmode_cvt): Likewise.
7060 (sparc_preferred_simd_mode): Likewise.
7061 (output_cbranch): Likewise.
7062 (sparc_print_operand): Likewise.
7063 (sparc_expand_vec_perm_bmask): Likewise.
7064 (vector_init_bshuffle): Likewise.
7065 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7066 (spu_vector_mode_supported_p): Likewise.
7067 (spu_expand_insv): Likewise.
7068 (spu_emit_branch_or_set): Likewise.
7069 (spu_handle_vector_attribute): Likewise.
7070 (spu_builtin_splats): Likewise.
7071 (spu_builtin_extract): Likewise.
7072 (spu_builtin_promote): Likewise.
7073 (spu_expand_sign_extend): Likewise.
7074 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7075 (tilegx_simd_int): Likewise.
7076 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7077 (tilepro_simd_int): Likewise.
7078 * config/v850/v850.c (const_double_split): Likewise.
7079 (v850_print_operand): Likewise.
7080 (ep_memory_offset): Likewise.
7081 * config/vax/vax.c (vax_rtx_costs): Likewise.
7082 (vax_output_int_move): Likewise.
7083 (vax_output_int_add): Likewise.
7084 (vax_output_int_subtract): Likewise.
7085 * config/visium/predicates.md (visium_branch_operator): Likewise.
7086 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7087 (visium_print_operand_address): Likewise.
7088 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7089 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7090 (xtensa_expand_conditional_branch): Likewise.
7091 (xtensa_copy_incoming_a7): Likewise.
7092 (xtensa_output_literal): Likewise.
7093 * dfp.c (decimal_real_maxval): Likewise.
7094 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7097 Alan Hayward <alan.hayward@arm.com>
7098 David Sherwood <david.sherwood@arm.com>
7100 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7101 (mode_nunits_inline): Likewise.
7102 (mode_inner_inline): Likewise.
7103 (mode_unit_size_inline): Likewise.
7104 (mode_unit_precision_inline): Likewise.
7105 (emit_insn_modes_h): Likewise. Also emit a #define of the
7107 (emit_mode_wider): Add an E_ prefix to mode names.
7108 (emit_mode_complex): Likewise.
7109 (emit_mode_inner): Likewise.
7110 (emit_mode_adjustments): Likewise.
7111 (emit_mode_int_n): Likewise.
7112 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7113 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7114 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7115 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7116 (CRC32_BUILTIN, ENTRY): Likewise.
7117 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7118 (aarch64_pop_regs): Likewise.
7119 (aarch64_process_components): Likewise.
7120 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7121 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7122 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7123 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7124 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7125 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7127 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7128 (pcmpistr): Likewise.
7129 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7130 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7131 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7133 * config/rl78/rl78.c (mduc_regs): Likewise.
7134 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7135 (htm_expand_builtin): Likewise.
7136 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7137 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7138 * config/xtensa/xtensa.c (print_operand): Likewise.
7139 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7140 (NUM_MODE_VECTOR_INT): Likewise.
7141 * genoutput.c (null_operand): Likewise.
7142 (output_operand_data): Likewise.
7143 * genrecog.c (print_parameter_value): Likewise.
7144 * lra.c (debug_operand_data): Likewise.
7146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7148 * dwarf2out.c (scompare_loc_descriptor_wide)
7149 (scompare_loc_descriptor_narrow): New functions, split out from...
7150 (scompare_loc_descriptor): ...here.
7151 * expmed.c (emit_store_flag_int): New function, split out from...
7152 (emit_store_flag): ...here.
7154 2017-08-30 Richard Biener <rguenther@suse.de>
7156 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7157 (dwarf2out_early_finish): Move setting of AT_pubnames from
7158 early debug output to early finish.
7160 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7162 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7163 and -mdata-region to the assembler.
7165 2017-08-30 Richard Biener <rguenther@suse.de>
7167 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7169 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7170 (dwarf2out_early_global_decl): Only generate a DIE for the
7171 abstract origin if it doesn't already exist or is a declaration DIE.
7172 (resolve_addr): Do not add the linkage name twice when
7173 generating a stub DIE for the DW_TAG_GNU_call_site target.
7175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7177 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7178 Use machine_mode rather than int for arg1_mode.
7180 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7183 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7184 that the second argument of the built-in functions to unpack
7185 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7186 switch statement instead a lot of if statements.
7187 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7188 Allow 64-bit values to be in Altivec registers as well as
7189 traditional floating point registers.
7190 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7192 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7194 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7195 MAX_REGS_PER_ADDRESS == 1.
7197 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7199 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7200 * config/i386/i386.c (ix86_option_override_internal): Simplify
7201 setting of opts->x_flag_entry.
7203 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7204 Jakub Jelinek <jakub@redhat.com>
7205 Richard Biener <rguenther@suse.de>
7207 PR tree-optimization/81503
7208 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7209 folded constant fits in the target type; reorder tests for clarity.
7211 2017-08-29 Martin Liska <mliska@suse.cz>
7213 * passes.def: Include pass_lower_switch.
7214 * stmt.c (dump_case_nodes): Remove and move to
7215 tree-switch-conversion.
7216 (case_values_threshold): Likewise.
7217 (expand_switch_as_decision_tree_p): Likewise.
7218 (emit_case_decision_tree): Likewise.
7219 (expand_case): Likewise.
7220 (balance_case_nodes): Likewise.
7221 (node_has_low_bound): Likewise.
7222 (node_has_high_bound): Likewise.
7223 (node_is_bounded): Likewise.
7224 (emit_case_nodes): Likewise.
7225 (struct simple_case_node): New struct.
7226 (add_case_node): Remove.
7227 (emit_case_dispatch_table): Use vector instead of case_list.
7228 (reset_out_edges_aux): Remove.
7229 (compute_cases_per_edge): Likewise.
7230 (expand_case): Build list of simple_case_node.
7231 (expand_sjlj_dispatch_table): Use it.
7232 * tree-switch-conversion.c (struct case_node): Moved from
7233 stmt.c and adjusted.
7234 (emit_case_nodes): Likewise.
7235 (node_has_low_bound): Likewise.
7236 (node_has_high_bound): Likewise.
7237 (node_is_bounded): Likewise.
7238 (case_values_threshold): Likewise.
7239 (reset_out_edges_aux): Likewise.
7240 (compute_cases_per_edge): Likewise.
7241 (add_case_node): Likewise.
7242 (dump_case_nodes): Likewise.
7243 (balance_case_nodes): Likewise.
7244 (expand_switch_as_decision_tree_p): Likewise.
7245 (emit_jump): Likewise.
7246 (emit_case_decision_tree): Likewise.
7247 (try_switch_expansion): Likewise.
7248 (do_jump_if_equal): Likewise.
7249 (emit_cmp_and_jump_insns): Likewise.
7250 (fix_phi_operands_for_edge): New function.
7251 (record_phi_operand_mapping): Likewise.
7252 (class pass_lower_switch): New pass.
7253 (pass_lower_switch::execute): New function.
7254 (make_pass_lower_switch): Likewise.
7255 (conditional_probability):
7256 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7257 * tree-pass.h: Add make_pass_lower_switch.
7259 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7262 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7265 2017-08-29 Richard Biener <rguenther@suse.de>
7267 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7268 we do not add a DW_AT_inline attribute twice.
7269 (gen_subprogram_die): Remove code setting DW_AT_inline on
7270 DECL_ABSTRACT_P nodes.
7272 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7274 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7275 calls to internal functions.
7276 (gimplify_modify_expr): Likewise.
7277 * tree-call-cdce.c (use_internal_fn): Likewise.
7278 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7279 (convert_to_divmod): Set the nothrow flag.
7280 * tree-if-conv.c (predicate_mem_writes): Likewise.
7281 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7282 (vectorizable_call): Likewise.
7283 (vectorizable_store): Likewise.
7284 (vectorizable_load): Likewise.
7285 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7286 (vect_recog_mask_conversion_pattern): Likewise.
7288 2017-08-29 Martin Liska <mliska@suse.cz>
7291 * gcc.c (driver_handle_option): Add new argument.
7292 * opts-common.c (handle_option): Pass
7293 target_option_override_hook.
7294 * opts-global.c (lang_handle_option): Add new option.
7295 (set_default_handlers): Add new argument.
7296 (decode_options): Likewise.
7297 * opts.c (target_handle_option): Likewise.
7298 (common_handle_option): Call target_option_override_hook.
7299 * opts.h (struct cl_option_handler_func): Add hook for
7300 target option override.
7301 (struct cl_option_handlers): Likewise.
7302 (set_default_handlers): Add new argument.
7303 (decode_options): Likewise.
7304 (common_handle_option): Likewise.
7305 (target_handle_option): Likewise.
7306 * toplev.c (toplev::main): Pass targetm.target_option.override
7309 2017-08-29 Richard Biener <rguenther@suse.de>
7310 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7312 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7313 life to the active subtree.
7315 2017-08-28 Jeff Law <law@redhat.com>
7317 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7318 derive_equivalences.
7319 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7321 (edge_info::derive_equivalences): New private member function
7323 * tree-ssa-dom.c (class edge_info): Changed from a struct
7324 to a class. Add ctor/dtor, methods and data members.
7325 (edge_info::edge_info): Renamed from allocate_edge_info.
7326 Initialize additional members.
7327 (edge_info::~edge_info): New.
7328 (free_dom_edge_info): Delete the edge info.
7329 (record_edge_info): Use new class & associated member functions.
7330 Tighten forms for testing for edge equivalences.
7331 (record_temporary_equivalences): Iterate over the simple
7332 equivalences rather than assuming there's only one per edge.
7333 (cprop_into_successor_phis): Iterate over the simple
7334 equivalences rather than assuming there's only one per edge.
7335 (optimize_stmt): Use operand_equal_p rather than pointer
7336 equality for mini-DSE code.
7338 2017-08-28 Nathan Sidwell <nathan@acm.org>
7340 * gcc.c (execute): Fold SIGPIPE handling into switch
7341 statement. Adjust internal error message.
7343 2017-08-28 Richard Biener <rguenther@suse.de>
7346 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7347 Do nothing for removed DIEs.
7349 2017-08-28 Richard Biener <rguenther@suse.de>
7351 PR tree-optimization/81977
7352 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7355 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7358 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7359 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7360 using targetm.gen_mem_thread_fence.
7362 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7365 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7366 predicate to register_operand. Reorder operands.
7367 (*btr<mode>): Ditto.
7368 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7369 (*btr<mode>_mask): Ditto.
7371 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7373 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7374 * config/rs6000/xmmintrin.h: New file.
7375 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7377 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7380 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7381 parameter and_insn and return it.
7382 (recombine_lvx_pattern): Insert a copy to ensure availability of
7383 the base register of the copied masking operation at the point of
7384 the instruction replacement.
7385 (recombine_stvx_pattern): Likewise.
7387 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7389 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7390 undocumented switches.
7391 (-mpower9-dform-vector): Likewise.
7392 (-mpower9-dform): Likewise.
7393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7394 comments to delete references to -mpower9-dform* switches.
7395 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7396 Delete reference to -mpower9-dform* switches, test for
7397 -mpower9-vector instead.
7398 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7399 (OTHER_P9_VECTOR_MASKS): Likewise.
7400 (POWERPC_MASKS): Likewise.
7401 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7402 tests against -mpower9-dform* to -mpower9-vector. Delete code
7403 that checked for -mpower9-dform* consistancy with other options.
7404 Add test for -mpower9-misc to enable other power9 switches.
7405 (rs6000_init_hard_regno_mode_ok): Likewise.
7406 (rs6000_option_override_internal): Likewise.
7407 (rs6000_emit_prologue): Likewise.
7408 (rs6000_emit_epilogue): Likewise.
7409 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7410 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7411 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7413 (emit_fusion_p9_store): Likewise.
7414 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7415 resetting these macros if the assembler does not support ISA 3.0
7417 (TARGET_P9_DFORM_VECTOR): Likewise.
7418 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7419 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7421 2017-08-25 Alan Modra <amodra@gmail.com>
7423 PR rtl-optimization/81747
7424 * cse.c (cse_extended_basic_block): Don't attempt to record
7425 equivalences for degenerate conditional branches that branch
7426 to their fall-through.
7428 2017-08-24 Martin Sebor <msebor@redhat.com>
7431 * gimple-fold.c (size_must_be_zero_p): New function.
7432 (gimple_fold_builtin_memory_op): Call it.
7434 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7436 * config/rs6000/mm_malloc.h: New file.
7438 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7440 PR tree-optimization/81913
7441 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7442 analysis when either IVs in condition can wrap.
7444 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7446 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7447 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7449 2017-08-24 Richard Biener <rguenther@suse.de>
7452 * targhooks.c (default_target_can_inline_p): Properly
7453 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7454 is present and always compare.
7455 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7456 infer -mfpmath=sse from TARGET_SSE_P.
7457 (ix86_can_inline_p): Properly use target_option_default_node when
7458 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7460 2017-08-24 Richard Biener <rguenther@suse.de>
7463 * dwarf2out.c (output_die): Handle flag_generate_offload like
7465 (output_comp_unit): Likewise.
7466 (gen_array_type_die): Likewise.
7467 (dwarf2out_early_finish): Likewise.
7468 (note_variable_value_in_expr): Likewise.
7469 (dwarf2out_finish): Likewise. Adjust assert.
7470 * cgraphunit.c (symbol_table::compile): Move setting of
7471 flag_generate_offload earlier ...
7472 (symbol_table::finalize_compilation_unit): ... here, before
7473 early debug finalization.
7475 2017-08-24 Richard Biener <rguenther@suse.de>
7477 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7478 and ipa-fnsummary.h.
7479 (ix86_can_inline_p): When ix86_fpmath flags do not match
7480 check whether the callee uses FP math at all.
7482 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7485 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7486 instead of TYPE_PRECISION.
7488 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7489 Alan Hayward <alan.hayward@arm.com>
7490 David Sherwood <david.sherwood@arm.com>
7492 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7494 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7496 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7498 (ix86_frame::stack_realign_allocate): New field.
7499 (struct machine_frame_state): Modify comments.
7500 (machine_frame_state::sp_realigned_fp_end): New field.
7501 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7503 (sp_valid_at): Add assertion to assure no attempt to access invalid
7504 offset of a realigned stack.
7505 (fp_valid_at): Likewise.
7506 (choose_baseaddr): Modify comments.
7507 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7508 ix86_expand_prologue.
7509 (ix86_expand_prologue): Modify stack realignment and allocation.
7510 (ix86_expand_epilogue): Modify comments.
7511 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7512 avx2_runtime, avx512f, and avx512f_runtime.
7514 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7516 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7517 (mstackrealign): Do not init to -1.
7518 * config/i386/i386.c (ix86_option_override_internal):
7519 Check opts_set, not opts when setting default value of
7520 opts->x_ix86_force_align_arg_pointer.
7522 2017-08-23 Richard Biener <rguenther@suse.de>
7524 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7525 lang_hooks.decl_printable_name.
7526 * print-rtl-function.c (print_rtx_function): Likewise.
7527 * tree-pretty-print.c (dump_function_header): Likewise.
7529 2017-08-23 Richard Biener <rguenther@suse.de>
7532 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7533 -g0 at compile-time.
7535 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7538 * doc/sourcebuild.texi (Other hardware attributes):
7541 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7544 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7545 Fix single-use check.
7547 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7549 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7551 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7553 * doc/install.texi: Modify to add more details on running selected
7556 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7558 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7559 is combined with -mabi=ms.
7560 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7563 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7565 PR tree-optimization/81488
7566 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7567 and cached_basis fields.
7568 (MAX_SPREAD): New constant.
7569 (alloc_cand_and_find_basis): Initialize new fields.
7570 (clear_visited): New function.
7571 (create_phi_basis_1): Rename from create_phi_basis, set visited
7572 and cached_basis fields.
7573 (create_phi_basis): New wrapper function.
7574 (phi_add_costs_1): Rename from phi_add_costs, add spread
7575 parameter, set visited field, short-circuit when limits reached.
7576 (phi_add_costs): New wrapper function.
7577 (record_phi_increments_1): Rename from record_phi_increments, set
7579 (record_phi_increments): New wrapper function.
7580 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7581 (phi_incr_cost): New wrapper function.
7582 (all_phi_incrs_profitable_1): Rename from
7583 all_phi_incrs_profitable, set visited field.
7584 (all_phi_incrs_profitable): New wrapper function.
7586 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7587 Alan Hayward <alan.hayward@arm.com>
7588 David Sherwood <david.sherwood@arm.com>
7590 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7591 that takes the outer and inner modes.
7592 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7593 comparison as the canonical test for a paradoxical subreg.
7594 * combine.c (simplify_set): Use paradoxical_subreg_p.
7595 (make_extraction): Likewise.
7596 (force_to_mode): Likewise.
7597 (rtx_equal_for_field_assignment_p): Likewise.
7598 (gen_lowpart_for_combine): Likewise.
7599 (simplify_comparison): Likewise.
7600 * cse.c (equiv_constant): Likewise.
7601 * expmed.c (store_bit_field_1): Likewise.
7602 * final.c (alter_subreg): Likewise.
7603 * fwprop.c (propagate_rtx): Likewise.
7604 (forward_propagate_subreg): Likewise.
7605 * ira-conflicts.c (ira_build_conflicts): Likewise.
7606 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7607 * lra-constraints.c (curr_insn_transform): Likewise.
7608 (split_reg): Likewise.
7609 * lra-eliminations.c (move_plus_up): Likewise.
7610 (lra_eliminate_regs_1): Likewise.
7611 * recog.c (general_operand): Likewise.
7612 * ree.c (combine_reaching_defs): Likewise.
7613 * reload.c (push_reload): Likewise.
7614 (find_reloads): Likewise.
7615 * reload1.c (elimination_effects): Likewise.
7616 (compute_reload_subreg_offset): Likewise.
7617 (choose_reload_regs): Likewise.
7618 * rtlanal.c (subreg_lsb_1): Likewise.
7619 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7620 (simplify_subreg): Likewise.
7621 * var-tracking.c (track_loc_p): Likewise.
7622 * emit-rtl.c (byte_lowpart_offset): Likewise.
7623 (paradoxical_subreg_p): Delete out-of-line definition.
7625 2017-08-22 Jeff Law <law@redhat.com>
7627 PR tree-optimization/81741
7628 PR tree-optimization/71947
7629 * tree-ssa-dom.c: Include tree-inline.h.
7630 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7631 equivalences if one is more expensive to compute than the other.
7632 * tree-ssa-scopedtables.h (class const_or_copies): Make
7633 record_const_or_copy_raw method private.
7634 (class avail_exprs_stack): New method simplify_binary_operation.
7635 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7636 avail_exprs_stack::simplify_binary_operation as needed.
7637 (avail_exprs_stack::simplify_binary_operation): New function.
7639 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7641 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7642 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7643 (DOT_SYMBOLS): Likewise.
7644 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7645 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7646 (RS6000_ABI_NAME): Likewise.
7647 (TARGET_CMODEL): Likewise.
7648 (TOC_SECTION_ASM_OP): Likewise.
7649 (SET_CMODEL): New macro.
7650 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7652 2017-08-22 Richard Biener <rguenther@suse.de>
7654 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7655 to allow for free-lang-data replacements similar to verify_type_variant.
7657 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7658 Alan Hayward <alan.hayward@arm.com>
7659 David Sherwood <david.sherwood@arm.com>
7661 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7662 VOIDmode for the LABEL_REF.
7664 2017-08-22 Richard Biener <rguenther@suse.de>
7666 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7668 2017-08-22 Richard Biener <rguenther@suse.de>
7670 * common.opt (feliminate-dwarf2-dups): Ignore.
7671 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7672 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7673 same_die_p_wrap, compute_section_prefix,
7674 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7675 (comdat_symbol_id, comdat_symbol_number): Likewise.
7676 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7678 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7679 (output_die): Mark unreachable path unreachable.
7680 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7681 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7682 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7683 (dwarf2out_early_finish): Likewise.
7685 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7687 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7689 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7692 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7693 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7694 (avr_attribute_table) <io, io_low, address>: Initialize
7695 .decl_required with true.
7697 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7699 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7700 undocumented debugging options.
7701 (-mvsx-scalar-double): Likewise.
7702 (-mallow-df-permute): Likewise.
7703 (-mvectorize-builtins): Likewise.
7704 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7705 (rs6000_builtin_vectorized_function): Likewise.
7706 (rs6000_builtin_md_vectorized_function): Likewise.
7707 (rs6000_opt_vars): Likewise.
7709 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7712 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7713 (*btrq_imm): Rename from *btrq.
7714 (*btcq_imm): Rename from *btcq.
7715 (btsc): New code attribute.
7716 (*<btsc><mode>): New insn pattern.
7717 (*btr<mode>): Ditto.
7718 (*<btsc><mode>_mask): New insn_and_split pattern.
7719 (*btr<mode>_mask): Ditto.
7721 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7722 Alan Hayward <alan.hayward@arm.com>
7723 David Sherwood <david.sherwood@arm.com>
7725 * function.c (pad_below): Simplify padding calculation.
7727 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7728 Alan Hayward <alan.hayward@arm.com>
7729 David Sherwood <david.sherwood@arm.com>
7731 * target.def (function_prologue): Remove frame size argument.
7732 (function_epilogue): Likewise.
7733 * doc/tm.texi: Regenerate.
7734 * final.c (final_start_function): Update call to function_prologue.
7735 (final_end_function): Update call to function_epilogue.
7736 (default_function_pro_epilogue): Remove frame size argument.
7737 * output.h (default_function_pro_epilogue): Likewise.
7738 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7739 (arm_output_function_prologue): Likewise.
7740 * config/frv/frv.c (frv_function_prologue): Likewise.
7741 (frv_function_epilogue): Likewise.
7742 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7743 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7744 (ia64_output_function_epilogue): Likewise.
7745 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7746 (m32r_output_function_epilogue): Likewise.
7747 * config/microblaze/microblaze.c (microblaze_function_prologue)
7748 (microblaze_function_epilogue): Likewise.
7749 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7750 (mips_output_function_epilogue): Likewise.
7751 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7752 (mmix_target_asm_function_epilogue): Likewise.
7753 * config/msp430/msp430.c (msp430_start_function): Likewise.
7754 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7755 (nds32_asm_function_epilogue): Likewise.
7756 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7757 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7758 (pa_output_function_epilogue): Likewise.
7759 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7760 (rs6000_output_function_epilogue): Likewise.
7761 * config/rl78/rl78.c (rl78_start_function): Likewise.
7762 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7763 (rs6000_output_function_epilogue): Likewise.
7764 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7765 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7766 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7767 (sparc_asm_function_epilogue): Likewise.
7769 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7771 * tree.h (type_has_mode_precision_p): New function.
7772 * convert.c (convert_to_integer_1): Use it.
7773 * expr.c (expand_expr_real_2): Likewise.
7774 (expand_expr_real_1): Likewise.
7775 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7776 * match.pd: Likewise.
7777 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7778 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7779 * tree-tailcall.c (process_assignment): Likewise.
7780 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7781 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7782 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7783 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7784 (vectorizable_assignment): Likewise.
7785 (vectorizable_shift): Likewise.
7786 (vectorizable_operation): Likewise.
7787 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7789 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7791 * match.pd: Add pow (C, x) simplification.
7793 2017-08-21 Richard Biener <rguenther@suse.de>
7795 PR tree-optimization/81900
7796 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7797 for blocks with abnormal predecessors.
7798 (compute_antic): Do not set visited flag prematurely.
7800 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7803 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7805 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7807 * stor-layout.h (vector_type_mode): Move to...
7808 * tree.h (vector_type_mode): ...here.
7809 * stor-layout.c (vector_type_mode): Move to...
7810 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7812 2017-08-21 Richard Biener <rguenther@suse.de>
7814 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7815 register_external_die hooks.
7816 (debug_false_tree_charstarstar_uhwistar): Declare.
7817 (debug_nothing_tree_charstar_uhwi): Likewise.
7818 * debug.c (do_nothing_debug_hooks): Adjust.
7819 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7820 (debug_nothing_tree_charstar_uhwi): Likewise.
7821 * dbxout.c (dbx_debug_hooks): Adjust.
7822 (xcoff_debug_hooks): Likewise.
7823 * sdbout.c (sdb_debug_hooks): Likewise.
7824 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7825 * dwarf2out.c (macinfo_label_base): New global.
7826 (dwarf2out_register_external_die): New function for the
7827 register_external_die hook.
7828 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7829 (dwarf2_debug_hooks): Use them.
7830 (dwarf2_lineno_debug_hooks): Adjust.
7831 (struct die_struct): Add with_offset flag.
7832 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7833 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7834 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7835 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7836 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7837 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7838 defining section names for the early LTO debug variants.
7839 (reset_indirect_string): New helper.
7840 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
7841 (print_dw_val): Add support for offsetted symbol references.
7842 (get_ultimate_context): Split out from is_cxx.
7843 (is_cxx): Use get_ultimate_context.
7844 (is_fortran): Add decl overload.
7845 (compute_comp_unit_symbol): Split out worker from
7846 compute_section_prefix.
7847 (compute_section_prefix): Call compute_comp_unit_symbol and
7848 set comdat_type_p here.
7849 (output_die): Skip DIE symbol output for the LTO added one.
7850 Handle DIE symbol references with offset.
7851 (output_comp_unit): Guard section name mangling properly.
7852 For LTO debug sections emit a symbol at the section beginning
7853 which we use to refer to its DIEs.
7854 (add_abstract_origin_attribute): For DIEs registered via
7855 dwarf2out_register_external_die directly refer to the early
7856 DIE rather than indirectly through the shadow one we created.
7857 Remove obsolete call to dwarf2out_abstract_function for
7858 non-function/block origins.
7859 (gen_array_type_die): When generating early LTO debug do
7860 not emit DW_AT_string_length.
7861 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
7862 late when in LTO. As suggested place a gcc_unreachable for
7863 the DECL_ABSTRACT_P case.
7864 (gen_subprogram_die): Avoid another specification DIE
7865 for early built declarations/definitions for the late LTO case.
7866 (gen_variable_die): Add type references for late duplicated VLA dies
7868 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
7869 we have the abstract instance already.
7870 (process_scope_var): Adjust decl DIE contexts in LTO which
7871 first puts them in limbo.
7872 (gen_decl_die): Do not generate type DIEs late apart from
7873 types for VLAs or for decls we do not yet have a DIE. Do not
7874 call dwarf2out_abstract_function late.
7875 (dwarf2out_early_global_decl): Make sure to create DIEs
7876 for abstract instances of a decl first.
7877 (dwarf2out_late_global_decl): Adjust comment.
7878 (output_macinfo_op): With multiple macro sections use
7879 macinfo_label_base to distinguish labels.
7880 (output_macinfo): Likewise. Update macinfo_label_base.
7881 Pass in the line info label.
7882 (note_variable_value_in_expr): When generating LTO resolve
7883 all variable values here by generating DIEs as needed.
7884 (init_sections_and_labels): Add early LTO debug flag parameter
7885 and generate different sections and names if set. Add generation
7886 counter for the labels so we can have multiple of them.
7887 (reset_dies): Helper to allow DIEs to be output multiple times.
7888 (dwarf2out_finish): When outputting DIEs to the fat part of an
7889 LTO object first reset DIEs.
7890 (dwarf2out_early_finish): Output early DIEs when generating LTO.
7891 (modified_type_die): Check for decl_ultimate_origin being self
7893 (gen_type_die_with_usage): Likewise.
7894 (gen_typedef_die): Allow decl_ultimate_origin being self.
7895 (set_decl_abstract_flags): Remove.
7896 (set_block_abstract_flags): Likewise.
7897 (dwarf2out_abstract_function): Treat the early generated DIEs
7898 as the abstract copy and only add DW_AT_inline and
7899 DW_AT_artificial here and call set_decl_origin_self.
7900 If the DIE has an abstract origin don't do anything.
7901 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
7902 if we have none yet (Go fails to build one, PR78628).
7903 (variably_modified_type_p): Prevent endless recursion for Ada
7904 cyclic pointer types.
7905 * lto-streamer-in.c: Include debug.h.
7906 (dref_queue): New global.
7907 (lto_read_tree_1): Stream in DIE references.
7908 (lto_input_tree): Register DIE references.
7909 (input_function): Stream DECL_DEBUG_ARGS.
7910 * lto-streamer-out.c: Include debug.h.
7911 (lto_write_tree_1): Output DIE references.
7912 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
7913 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
7914 (output_function): Stream DECL_DEBUG_ARGS.
7915 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
7916 Stream DECL_ABSTRACT_ORIGIN.
7917 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
7918 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
7919 DECL_CONTEXT for file-scope decls.
7920 * lto-streamer.h (struct dref_entry): Declare.
7921 (dref_queue): Likewise.
7922 * cfgexpand.c (pass_expand::execute): Do not call the
7923 outlining_inline_function hook here.
7924 * lto-wrapper.c (debug_obj): New global.
7925 (tool_cleanup): Unlink it if required.
7926 (debug_objcopy): New function.
7927 (run_gcc): Handle early debug sections in the IL files by
7928 extracting them to separate files, partially linkin them and
7929 feeding the result back as result to the linker.
7930 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
7931 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
7932 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
7933 sections into a separate segment.
7934 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
7936 (darwin_asm_dwarf_section): Likewise.
7937 (darwin_asm_output_dwarf_offset): Likewise.
7938 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
7940 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7941 Alan Hayward <alan.hayward@arm.com>
7942 David Sherwood <david.sherwood@arm.com>
7944 * read-md.h (md_reader::record_potential_iterator_use): Replace
7945 pointer argument with an rtx and an index.
7946 * read-rtl.c (iterator_group::apply_iterator): Likewise.
7947 (apply_mode_iterator): Likewise.
7948 (apply_code_iterator): Likewise.
7949 (apply_int_iterator): Likewise.
7950 (apply_subst_iterator): Likewise.
7951 (record_iterator_use): Likewise.
7952 (record_attribute_use): Likewise.
7953 (md_reader::record_potential_iterator_use): Likewise. Update calls
7954 to record_iterator_use and apply_iterator.
7955 (iterator_use): Replace ptr with x and index.
7956 (attribute_use): Likewise.
7957 (apply_attribute_uses): Update calls to apply_iterator.
7958 (apply_iterators): Likewise. Update initialization of iterator_use.
7959 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
7960 and record_potential_iterator_use.
7961 (rtx_reader::read_rtx_operand): Likewise.
7963 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7964 Alan Hayward <alan.hayward@arm.com>
7965 David Sherwood <david.sherwood@arm.com>
7967 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
7970 2017-08-21 Richard Biener <rguenther@suse.de>
7973 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
7974 at struct end conservatively when comparing common bases.
7976 2017-08-21 Richard Biener <rguenther@suse.de>
7978 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
7979 (mem_ref_in_stmt): Remove.
7980 (determine_max_movement): Use ref index to get at the reference.
7981 (invariantness_dom_walker::before_dom_children): Deal with
7982 lim data already initialized.
7983 (gather_mem_refs_stmt): Initialize lim data and record ref index.
7985 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7987 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
7988 (TARGET_ISA_ROUND): Ditto.
7989 (TARGET_ROUND): Ditto.
7990 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
7991 * config/i386/i386.md: Ditto.
7992 * config/i386/sse.md: Ditto.
7993 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
7994 with OPTION_MASK_ISA_SSE4_1.
7996 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
7999 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8000 __builtin_ia32_lzcnt_u16.
8002 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8005 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8006 (rs6000_set_current_function): Rewrite function to use it.
8008 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8011 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8012 and TYPE_WARN_IF_NOT_ALIGN.
8013 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8014 (handle_warn_if_not_align): New.
8015 (place_union_field): Call handle_warn_if_not_align.
8016 (place_field): Call handle_warn_if_not_align.
8017 Copy TYPE_WARN_IF_NOT_ALIGN.
8018 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8019 (layout_type): Likewise.
8020 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8022 (tree_decl_common): Add warn_if_not_align.
8023 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8024 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8025 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8026 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8027 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8028 * doc/extend.texi: Document warn_if_not_aligned attribute.
8029 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8031 2017-08-17 Martin Liska <mliska@suse.cz>
8034 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8035 (get_data_dependence): Use it as pointer type.
8036 (distribute_loop): Likewise.
8038 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8040 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8041 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8042 (p8_vmrgow_<mode>_direct): New define_insn.
8043 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8044 handle endianness for vmrgew and vmrgow permute patterns.
8046 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8048 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8049 * config/rs6000/rs6000-cpus.def: Remove comment.
8050 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8051 (POWERPC_MASKS): Likewise.
8052 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8053 use of TARGET_VSX_TIMODE.
8054 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8055 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8056 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8057 (rs6000_option_override_internal): Remove dead code.
8058 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8059 (rs6000_legitimize_reload_address): Likewise.
8060 (rs6000_legitimate_address_p): Likewise.
8061 (rs6000_opt_masks): Delete "vsx-timode".
8062 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8063 from function comment.
8064 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8065 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8066 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8068 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8069 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8070 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8073 (define_peephole2): Likewise.
8075 2017-08-17 Martin Sebor <msebor@redhat.com>
8078 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8079 past the end of an array.
8080 (test_pp_format): Add test cases.
8082 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8084 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8085 missing ECF_NOTHROW flags.
8087 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8090 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8091 operands residing in integer registers.
8092 (*vsx_le_perm_load_<mode>): Likewise.
8093 (*vsx_le_perm_store_<mode>): Likewise.
8094 (define_peephole2): Add peepholes to optimize the above.
8096 2017-08-17 Marek Polacek <polacek@redhat.com>
8099 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8100 to mimic what shorten_compare did. Change the return type to bool.
8101 (fold_cond_expr_with_comparison): Update call to
8102 operand_equal_for_comparison_p.
8103 (fold_ternary_loc): Likewise.
8105 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8107 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8109 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8111 2017-08-17 Richard Biener <rguenther@suse.de>
8113 * tree-ssa-structalias.c (solve_graph): When propagating
8114 to successors update the graphs succ edges and avoid duplicate work.
8116 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8119 * config/i386/i386.c (ix86_option_override_internal): Save target
8120 specific options after ix86_stack_protector_guard_reg was changed.
8122 2017-08-17 Richard Biener <rguenther@suse.de>
8124 PR tree-optimization/81827
8125 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8126 (new_var_info): Initialize it conservatively.
8127 (get_call_vi): Mark register vars.
8128 (new_scalar_tmp_constraint_exp): Likewise.
8129 (handle_rhs_call): Likewise.
8130 (handle_const_call): Likewise.
8131 (create_function_info_for): Likewise.
8132 (solve_constraints): Sort varinfos to separate register from
8133 non-register vars to pack points-to solution bitmaps during
8136 2017-08-17 Marek Polacek <polacek@redhat.com>
8138 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8140 2017-08-17 Richard Biener <rguenther@suse.de>
8142 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8143 to true when overflow is undefined and we saturated the result.
8145 2017-08-17 Alan Modra <amodra@gmail.com>
8148 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8149 Don't use store multiple if only one reg needs saving.
8150 (interesting_frame_related_regno): New function.
8151 (rs6000_frame_related): Don't emit frame info for regs that
8153 (rs6000_emit_epilogue): Likewise.
8155 2017-08-16 Nathan Sidwell <nathan@acm.org>
8157 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8158 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8159 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8160 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8161 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8162 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8163 (verify_type): Adjust for TYPE_BINFO move.
8164 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8165 process TYPE_BINFO directly.
8166 (hash_tree): Likewise.
8167 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8169 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8172 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8174 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8176 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8179 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8180 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8181 Add ix86_binary_operator_ok to insn constraint.
8182 (*iordi_1_bts): Ditto.
8183 (*xordi_1_btc): Ditto.
8184 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8185 Update corresponding peephole2 pattern.
8189 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8191 PR tree-optimization/81832
8192 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8193 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8195 2017-08-16 Marek Polacek <polacek@redhat.com>
8198 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8199 regarding size_zero_node.
8201 2017-08-16 Martin Liska <mliska@suse.cz>
8204 * config.gcc: Respect previously set extra_objs in case
8207 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8209 PR tree-optimization/81835
8210 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8211 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8212 not depend on the phi.
8214 2017-08-16 Alan Modra <amodra@gmail.com>
8216 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8219 2017-08-16 Alan Modra <amodra@gmail.com>
8221 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8222 (save_reg_p): ..into this. Update all callers.
8223 (first_reg_to_save): Simplify.
8225 2017-08-16 Alan Modra <amodra@gmail.com>
8227 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8230 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8234 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8235 constant count if that count is less than 32.
8237 2017-08-15 Nathan Sidwell <nathan@acm.org>
8239 * gcc.c (execute): Emit friendlier message if inferior is killed
8240 by an external cause.
8242 2017-08-15 Richard Biener <rguenther@suse.de>
8244 PR tree-optimization/81790
8245 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8246 CONSTRUCTORs from simplifying and VN.
8248 2017-08-14 Martin Sebor <msebor@redhat.com>
8250 * builtin-attrs.def: Add comments.
8252 2017-08-14 Martin Sebor <msebor@redhat.com>
8255 * doc/extend.texi (attribute nonstring): Document new attribute.
8257 2017-08-14 Martin Sebor <msebor@redhat.com>
8260 * tree-diagnostic.c (default_tree_printer): Handle %G.
8261 * gimple-pretty-print.h (percent_G_format): Declare new function.
8262 * gimple-pretty-print.c (percent_G_format): Define.
8263 * tree-pretty-print.c (percent_K_format): Add argument.
8265 2017-08-14 Martin Sebor <msebor@redhat.com>
8267 PR translation/79998
8268 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8269 Remove a stray space.
8271 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8274 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8275 (*iordi_1_bts): Ditto.
8276 (*xordi_1_btc): Ditto.
8278 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8281 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8282 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8284 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8285 quoted strings, and make more translator-friendly.
8286 (darwin_rs6000_override_options): Likewise.
8287 (rs6000_option_override_internal): Likewise.
8288 (rs6000_return_in_memory): Fix overlong line.
8289 (init_cmulative_args): Use quoted strings, and make more
8290 translator-friendly.
8291 (rs6000_pass_by_reference): Fix overlong line.
8292 (def_builtin): Use quoted strings.
8293 (altivec_expand_predicate_builtin): Use quoted strings, and make
8294 more translator-friendly.
8295 (htm_expand_builtin): Use quoted strings.
8296 (cpu_expand_builtin): Use quoted strings, and make more
8297 translator-friendly.
8298 (altivec_expand_builtin): Likewise.
8299 (paired_expand_predicate_builtin): Likewise.
8300 (rs6000_invalid_builtin): Likewise.
8301 (builtin_function_type): Use quoted strings.
8302 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8303 more translator-friendly.
8304 (rs6000_trampoline_init): Likewise.
8305 (rs6000_handle_altivec_attribute): Likewise.
8306 (rs6000_inner_target_options): Use quoted strings.
8307 (rs6000_disable_incompatible_switches): Likewise.
8308 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8309 strings, and make more translator-friendly.
8310 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8312 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8314 PR tree-optimization/81799
8315 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8316 cond_expr to simple gimple operand.
8318 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8321 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8323 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8327 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8328 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8329 TARGET_GASISR_PROLOGUES.
8330 * config/avr/avr.c (avr_option_override): Same.
8331 (avr_pass_pre_proep::execute): Same.
8333 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8336 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8337 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8339 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8341 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8342 define_insn_and_split. Split to a memory load from 0 in
8343 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8344 using PTR mode iterator.
8345 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8346 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8347 (*add_tp_<mode>): Redefine as define_insn_and_split.
8348 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8349 address space. Merge with *add_tp_x32 using PTR mode iterator.
8350 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8351 Split to an add with a memory load from 0 in
8352 DEFAULT_TLS_SEG_REG address space.
8354 2017-08-12 Andrew Pinski <apinski@cavium.com>
8356 * config/aarch64/aarch64-option-extensions.def (rdma):
8357 Fix feature string to what Linux prints out in /proc/cpuinfo.
8359 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8362 * dwarf2out.c (modified_type_die): For C typedef types that have
8363 an ultimate origin, process the ultimate origin instead of the
8365 (gen_typedef_die): Assert that input DECLs have no ultimate
8367 (gen_type_die_with_usage): For typedef variants that have an
8368 ultimate origin, just call gen_decl_die on the original DECL.
8369 (process_scope_var): Avoid creating DIEs for local typedefs and
8370 concrete static variables.
8372 2017-08-12 Alan Modra <amodra@gmail.com>
8376 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8377 match gnu-user.h startfile.
8378 (ENDFILE_LINUX_SPEC): Similarly.
8380 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8383 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8385 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8387 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8388 * config/aarch64/aarch64.md (mov<mode>): Change.
8389 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8390 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8391 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8393 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8395 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8396 for storage order barriers.
8398 2017-08-11 Martin Liska <mliska@suse.cz>
8401 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8402 variables of void type.
8404 2017-08-11 Martin Liska <mliska@suse.cz>
8406 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8407 TARGET_SUPPORTS_ALIASES.
8408 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8409 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8410 (optimize_weakref): Likewise.
8411 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8412 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8413 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8415 2017-08-11 Martin Liska <mliska@suse.cz>
8418 * config/i386/i386.c (make_resolver_func): Do complete
8419 refactoring of the function.
8421 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8424 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8425 * config/i386/i386.c (ix86_stack_protect_guard): Use
8426 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8427 * doc/invoke.texi (x86 Options): Document
8428 -mstack-protector-guard-symbol= option.
8430 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8432 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8433 * config/i386/i386.c (ix86_split_stack_guard): New function.
8434 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8435 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8436 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8437 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8438 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8439 (split_stack_space_check): Call ix86_split_stack_guard.
8441 2017-08-10 Martin Sebor <msebor@redhat.com>
8443 * print-tree.c (print_node): Print location using the established
8445 Replace spaces with colons.
8446 (debug_raw, debug): Ditto.
8448 2017-08-10 Martin Sebor <msebor@redhat.com>
8451 * pretty-print.c (pp_format): Correct the handling of %s precision.
8453 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8456 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8458 (ix86_finalize_stack_frame_flags): This. Also clear
8459 frame_pointer_needed if -fno-omit-frame-pointer is used without
8461 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8462 with ix86_finalize_stack_frame_flags.
8463 (ix86_expand_epilogue): Likewise.
8464 (ix86_expand_split_stack_prologue): Likewise.
8465 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8467 2017-08-10 Martin Liska <mliska@suse.cz>
8470 * c-attribs.c (handle_target_attribute):
8471 Report warning for an empty string argument of target attribute.
8473 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8476 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8478 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8479 or DECL_NONLOCAL labels.
8480 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8481 or DECL_NONLOCAL labels here.
8483 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8485 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8486 to indicate when early gimple folding has been disabled.
8487 (rs6000_gimple_fold_builtin): Add debug content.
8488 (rs6000_invalid_builtin): Fix whitespace.
8489 (rs6000_expand_builtin): Fix whitespace.
8490 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8492 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8495 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8496 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8498 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8500 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8501 (qdf24xx): Likewise.
8502 * config/aarch64/aarch64.md: Include falkor.md.
8503 * config/aarch64/falkor.md: New.
8505 2017-08-09 Marek Polacek <polacek@redhat.com>
8508 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8509 * diagnostic.c (emit_diagnostic): Add a comment.
8510 (emit_diagnostic_valist): New function.
8512 2017-08-09 Marek Polacek <polacek@redhat.com>
8515 * input.c (make_location): New overload.
8516 * input.h (make_location): Declare.
8518 2017-08-08 Alan Modra <amodra@gmail.com>
8519 H.J. Lu <hongjiu.lu@intel.com>
8522 * gcc.c (NO_PIE_SPEC): Delete.
8523 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8525 (LINK_PIE_SPEC): ..to here.
8526 (LINK_COMMAND_SPEC): Support -no-pie.
8527 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8528 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8530 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8531 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8532 (ENDFILE_CRTEND_SPEC): Similarly.
8534 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8537 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8538 (mstack-protector-guard-offset=): Ditto.
8539 * config/i386/i386.c (ix86_option_override): Handle
8540 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8542 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8543 ix86_stack_protect_guard_offset variables.
8544 (TARGET_STACK_PROTECT_GUARD): Always define.
8545 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8546 and -mstack-protector-guard-offset= options.
8548 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8550 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8551 boundary case for the last candidate.
8553 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8555 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8556 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8558 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8561 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8562 * config/aarch64/aarch64-builtins.c
8563 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8564 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8565 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8567 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8568 Andrew Pinski <pinskia@gmail.com>
8571 * internal-fn.def (XORSIGN): New.
8572 * optabs.def (xorsign_optab): New.
8573 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8574 (convert_expand_mult_copysign): New.
8575 (pass_optimize_widening_mul::execute): Call
8576 convert_expand_mult_copysign.
8578 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8580 PR tree-optimization/81354
8581 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8582 Insert on edges rather than explicitly creating landing pads.
8583 (analyze_candidates_and_replace): Commit edge inserts.
8585 2017-08-08 Richard Biener <rguenther@suse.de>
8588 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8589 (expand_simple_operations): Also look through ADDR_EXPRs with
8590 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8592 2017-08-08 Richard Biener <rguenther@suse.de>
8594 PR tree-optimization/81723
8595 * tree-vect-slp.c (struct bst_traits): New hash traits.
8596 (bst_fail): New global.
8597 (vect_build_slp_tree_2): New worker, split out from ...
8598 (vect_build_slp_tree): ... this now wrapping it with using
8599 bst_fail set to cache SLP tree build fails. Properly handle
8601 (vect_analyze_slp_instance): Allocate and free bst_fail.
8603 2017-08-08 Martin Liska <mliska@suse.cz>
8606 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8607 LABEL_DECLs that can be from a different function.
8609 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8611 PR tree-optimization/81744
8612 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8613 loop's number of iterations.
8615 2017-08-08 Martin Liska <mliska@suse.cz>
8617 * asan.c: Include header files.
8618 * attribs.c (build_decl_attribute_variant): New function moved
8620 (build_type_attribute_qual_variant): Likewise.
8621 (cmp_attrib_identifiers): Likewise.
8622 (simple_cst_list_equal): Likewise.
8623 (omp_declare_simd_clauses_equal): Likewise.
8624 (attribute_value_equal): Likewise.
8625 (comp_type_attributes): Likewise.
8626 (build_type_attribute_variant): Likewise.
8627 (lookup_ident_attribute): Likewise.
8628 (remove_attribute): Likewise.
8629 (merge_attributes): Likewise.
8630 (merge_type_attributes): Likewise.
8631 (merge_decl_attributes): Likewise.
8632 (merge_dllimport_decl_attributes): Likewise.
8633 (handle_dll_attribute): Likewise.
8634 (attribute_list_equal): Likewise.
8635 (attribute_list_contained): Likewise.
8636 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8637 (lookup_attribute_by_prefix): Likewise.
8638 * bb-reorder.c: Include header files.
8639 * builtins.c: Likewise.
8640 * calls.c: Likewise.
8641 * cfgexpand.c: Likewise.
8642 * cgraph.c: Likewise.
8643 * cgraphunit.c: Likewise.
8644 * convert.c: Likewise.
8645 * dwarf2out.c: Likewise.
8646 * final.c: Likewise.
8647 * fold-const.c: Likewise.
8648 * function.c: Likewise.
8649 * gimple-expr.c: Likewise.
8650 * gimple-fold.c: Likewise.
8651 * gimple-pretty-print.c: Likewise.
8652 * gimple.c: Likewise.
8653 * gimplify.c: Likewise.
8654 * hsa-common.c: Likewise.
8655 * hsa-gen.c: Likewise.
8656 * internal-fn.c: Likewise.
8657 * ipa-chkp.c: Likewise.
8658 * ipa-cp.c: Likewise.
8659 * ipa-devirt.c: Likewise.
8660 * ipa-fnsummary.c: Likewise.
8661 * ipa-inline.c: Likewise.
8662 * ipa-visibility.c: Likewise.
8664 * lto-cgraph.c: Likewise.
8665 * omp-expand.c: Likewise.
8666 * omp-general.c: Likewise.
8667 * omp-low.c: Likewise.
8668 * omp-offload.c: Likewise.
8669 * omp-simd-clone.c: Likewise.
8670 * opts-global.c: Likewise.
8671 * passes.c: Likewise.
8672 * predict.c: Likewise.
8673 * sancov.c: Likewise.
8674 * sanopt.c: Likewise.
8675 * symtab.c: Likewise.
8676 * toplev.c: Likewise.
8677 * trans-mem.c: Likewise.
8678 * tree-chkp.c: Likewise.
8679 * tree-eh.c: Likewise.
8680 * tree-into-ssa.c: Likewise.
8681 * tree-object-size.c: Likewise.
8682 * tree-parloops.c: Likewise.
8683 * tree-profile.c: Likewise.
8684 * tree-ssa-ccp.c: Likewise.
8685 * tree-ssa-live.c: Likewise.
8686 * tree-ssa-loop.c: Likewise.
8687 * tree-ssa-sccvn.c: Likewise.
8688 * tree-ssa-structalias.c: Likewise.
8689 * tree-ssa.c: Likewise.
8690 * tree-streamer-in.c: Likewise.
8691 * tree-vectorizer.c: Likewise.
8692 * tree-vrp.c: Likewise.
8694 * ubsan.c: Likewise.
8695 * varasm.c: Likewise.
8696 * varpool.c: Likewise.
8697 * tree.c: Remove functions moved to attribs.[ch].
8699 * config/aarch64/aarch64.c: Add attrs.h header file.
8700 * config/alpha/alpha.c: Likewise.
8701 * config/arc/arc.c: Likewise.
8702 * config/arm/arm.c: Likewise.
8703 * config/avr/avr.c: Likewise.
8704 * config/bfin/bfin.c: Likewise.
8705 * config/c6x/c6x.c: Likewise.
8706 * config/cr16/cr16.c: Likewise.
8707 * config/cris/cris.c: Likewise.
8708 * config/darwin.c: Likewise.
8709 * config/epiphany/epiphany.c: Likewise.
8710 * config/fr30/fr30.c: Likewise.
8711 * config/frv/frv.c: Likewise.
8712 * config/ft32/ft32.c: Likewise.
8713 * config/h8300/h8300.c: Likewise.
8714 * config/i386/winnt.c: Likewise.
8715 * config/ia64/ia64.c: Likewise.
8716 * config/iq2000/iq2000.c: Likewise.
8717 * config/lm32/lm32.c: Likewise.
8718 * config/m32c/m32c.c: Likewise.
8719 * config/m32r/m32r.c: Likewise.
8720 * config/m68k/m68k.c: Likewise.
8721 * config/mcore/mcore.c: Likewise.
8722 * config/microblaze/microblaze.c: Likewise.
8723 * config/mips/mips.c: Likewise.
8724 * config/mmix/mmix.c: Likewise.
8725 * config/mn10300/mn10300.c: Likewise.
8726 * config/moxie/moxie.c: Likewise.
8727 * config/msp430/msp430.c: Likewise.
8728 * config/nds32/nds32-isr.c: Likewise.
8729 * config/nds32/nds32.c: Likewise.
8730 * config/nios2/nios2.c: Likewise.
8731 * config/nvptx/nvptx.c: Likewise.
8732 * config/pa/pa.c: Likewise.
8733 * config/pdp11/pdp11.c: Likewise.
8734 * config/powerpcspe/powerpcspe.c: Likewise.
8735 * config/riscv/riscv.c: Likewise.
8736 * config/rl78/rl78.c: Likewise.
8737 * config/rx/rx.c: Likewise.
8738 * config/s390/s390.c: Likewise.
8739 * config/sh/sh.c: Likewise.
8740 * config/sol2.c: Likewise.
8741 * config/sparc/sparc.c: Likewise.
8742 * config/spu/spu.c: Likewise.
8743 * config/stormy16/stormy16.c: Likewise.
8744 * config/tilegx/tilegx.c: Likewise.
8745 * config/tilepro/tilepro.c: Likewise.
8746 * config/v850/v850.c: Likewise.
8747 * config/vax/vax.c: Likewise.
8748 * config/visium/visium.c: Likewise.
8749 * config/xtensa/xtensa.c: Likewise.
8751 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8754 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8755 constraints since the -mupper-regs-* switches have been
8757 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8758 into a vector from a double word element that was extracted from
8759 another vector, and eliminate extra XXPERMDI instructions.
8760 (vsx_concat_<mode>_2): Likewise.
8761 (vsx_concat_<mode>_3): Likewise.
8762 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8763 concat to allow optimizing inserts from previous extracts.
8765 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8767 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8768 memory reference to a SSP offset in TLS address space.
8769 (ix86_print_operand) <case '@'>: Remove.
8770 (ix86_print_operand_punct_valid_p): Remove '@' code.
8771 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8773 (stack_tls_protect_set_<mode>): Remove.
8774 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8775 (stack_tls_protect_test_<mode>): Remove.
8776 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8778 2017-08-07 Olivier Hainque <hainque@adacore.com>
8781 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8783 2017-08-07 Douglas Rupp <rupp@adacore.com>
8785 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8786 variable was referenced as multidir in command.
8788 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8791 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8794 2017-08-07 Martin Liska <mliska@suse.cz>
8796 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8797 * config/rl78/rl78.c: Add include of attribs.h.
8798 * config/sh/sh.c: Likewise.
8799 * config/v850/v850.c: Likewise.
8801 2017-08-07 Tom de Vries <tom@codesourcery.com>
8804 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8806 2017-08-07 Martin Liska <mliska@suse.cz>
8808 * config/mips/mips.c: Include attribs.h.
8810 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8813 * doc/invoke.texi: Document change in behvaior for -Ofast for
8816 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8818 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8820 (aarch64_pop_regs): Likewise.
8821 (aarch64_gen_load_pair): Likewise.
8822 (aarch64_save_callee_saves): Likewise.
8823 (aarch64_restore_callee_saves): Likewise.
8825 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8827 * config/i386/i386.c: Revert the last change.
8829 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8832 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8834 (ix86_finalize_stack_frame_flags): This. Also clear
8835 frame_pointer_needed if -fno-omit-frame-pointer is used without
8837 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8838 with ix86_finalize_stack_frame_flags.
8839 (ix86_expand_epilogue): Likewise.
8840 (ix86_expand_split_stack_prologue): Likewise.
8842 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8845 * config/i386/i386.c (get_builtin_code_for_version): Set priority
8846 to P_AES for Westmere.
8848 2017-08-07 Jonathan Yong <10walls@gmail.com>
8850 * config/i386/mingw.opt (fset-stack-executable): Removed.
8851 * config/i386/cygming.opt (fset-stack-executable): Moved
8853 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
8855 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
8857 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
8859 2017-08-07 Marek Polacek <polacek@redhat.com>
8862 * fold-const.c (fold_indirect_ref_1): Check type_domain.
8864 2017-08-07 Martin Liska <mliska@suse.cz>
8866 * attribs.h (canonicalize_attr_name): New function.
8867 (cmp_attribs): Move from c-format.c and adjusted.
8868 (is_attribute_p): Moved from tree.h.
8869 * tree-inline.c: Add new includes.
8870 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
8871 (private_is_attribute_p): Remove.
8872 (private_lookup_attribute): Likewise.
8873 (private_lookup_attribute_by_prefix): Simplify.
8874 (remove_attribute): Use is_attribute_p.
8875 * tree.h: Remove removed declarations.
8877 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8880 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
8881 instead of computing it in the function. Formatting fix.
8882 (expand_case): Don't rely on default_edge being the first edge,
8883 clear it if removing it, pass default_edge to
8884 emit_case_dispatch_table.
8885 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
8888 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
8890 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
8891 insn in the function, emit NOP after the insn.
8893 2017-08-06 Tom de Vries <tom@codesourcery.com>
8895 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
8898 2017-08-06 Tom de Vries <tom@codesourcery.com>
8900 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
8903 2017-08-04 Yury Gribov <tetra2005@gmail.com>
8905 PR tree-optimization/57371
8906 * match.pd: New pattern.
8908 2017-08-04 Marek Polacek <polacek@redhat.com>
8911 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
8912 perform the computation in offset_int.
8914 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8916 PR tree-optimization/81136
8917 * tree-vectorizer.h: Include tree-hash-traits.h.
8918 (vec_base_alignments): New typedef.
8919 (vec_info): Add a base_alignments field.
8920 (vect_record_base_alignments): Declare.
8921 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
8923 (DR_IS_CONDITIONAL_IN_STMT): New macro.
8924 (create_data_ref): Add an is_conditional_in_stmt argument.
8925 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
8926 the is_conditional_in_stmt field.
8927 (data_ref_loc): Add an is_conditional_in_stmt field.
8928 (get_references_in_stmt): Set the is_conditional_in_stmt field.
8929 (find_data_references_in_stmt): Update call to create_data_ref.
8930 (graphite_find_data_references_in_stmt): Likewise.
8931 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
8932 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
8933 (vect_record_base_alignment): New function.
8934 (vect_record_base_alignments): Likewise.
8935 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
8936 for nested statements even if we fail to compute a misalignment.
8937 Use pooled base alignments for unconditional references.
8938 (vect_find_same_alignment_drs): Compare base addresses instead
8940 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
8941 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
8943 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8945 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
8946 Add an explicit name for the enum. Use auto_vec for slp_instances
8948 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
8950 (_bb_vec_info): Add a constructor and destructor.
8951 (vinfo_for_stmt): Return NULL for uids of -1 as well.
8952 (destroy_loop_vec_info): Delete.
8953 (vect_destroy_datarefs): Likewise.
8954 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
8955 (vec_info::vec_info): New function.
8956 (vec_info::~vec_info): Likewise.
8957 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
8958 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
8959 destroy_loop_vec_info.
8960 * tree-vect-loop.c (new_loop_vec_info): Replace with...
8961 (_loop_vec_info::_loop_vec_info): ...this.
8962 (destroy_loop_vec_info): Replace with...
8963 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
8964 the stmt_vec_infos. Leave handling of vec_info information to its
8965 destructor. Remove explicit vector releases.
8966 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
8967 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
8968 * tree-vect-slp.c (new_bb_vec_info): Replace with...
8969 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
8970 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
8971 (destroy_bb_vec_info): Replace with...
8972 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
8973 information to its destructor.
8974 (vect_slp_analyze_bb_1): Use new and delete instead of
8975 new_bb_vec_info and destroy_bb_vec_info.
8976 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
8979 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
8981 * tree-data-ref.h (subscript): Add access_fn field.
8982 (data_dependence_relation): Add could_be_independent_p.
8983 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
8984 (same_access_functions): Move to tree-data-ref.c.
8985 * tree-data-ref.c (ref_contains_union_access_p): New function.
8986 (access_fn_component_p): Likewise.
8987 (access_fn_components_comparable_p): Likewise.
8988 (dr_analyze_indices): Add a reference to access_fn_component_p.
8989 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
8991 (constant_access_functions): Likewise.
8992 (add_other_self_distances): Likewise.
8993 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
8994 (initialize_data_dependence_relation): Use XCNEW and remove
8995 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
8996 of access functions that have the same type. Allow the
8997 subsequence to end with different bases in some circumstances.
8998 Record the chosen access functions in SUB_ACCESS_FN.
8999 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9000 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9001 (subscript_dependence_tester_1): Likewise dra and drb.
9002 (build_classic_dist_vector): Update calls accordingly.
9003 (subscript_dependence_tester): Likewise.
9004 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9005 DDR_COULD_BE_INDEPENDENT_P.
9006 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9007 comp_alias_ddrs instead of may_alias_ddrs.
9008 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9010 (vect_analyze_data_ref_dependence): Use it if
9011 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9012 distance vectors if that fails.
9013 (dependence_distance_ge_vf): New function.
9014 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9015 LOOP_VINFO_MAY_ALIAS_DDRS.
9017 2017-08-04 Richard Biener <rguenther@suse.de>
9020 * fold-const.c (fold_binary_loc): Properly restrict
9021 minus_var0 && minus_var1 case when associating undefined overflow
9024 2017-08-04 Tom de Vries <tom@codesourcery.com>
9026 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9028 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9030 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9031 Don't start diagnostic messages with a capital letter.
9032 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9034 (rs6000_invalid_builtin): Likewise.
9035 (rs6000_trampoline_init): Likewise.
9037 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9040 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9041 after setting changeable df flags.
9043 2017-08-03 Richard Biener <rguenther@suse.de>
9045 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9046 up if the use is in USE - X.
9048 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9050 * toplev.c (dumpfile.h): New include.
9051 (internal_error_reentered): New static function. Use it...
9052 (internal_error_function): ...here to handle reentered internal_error.
9054 2017-08-03 Richard Biener <rguenther@suse.de>
9057 * fold-const.c (split_tree): Add minus_var and minus_con
9058 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9059 here but always use minus_*.
9060 (associate_trees): Assert we never associate with MINUS_EXPR
9061 and NULL first operand. Do not recurse for PLUS_EXPR operands
9062 when associating as MINUS_EXPR either.
9063 (fold_binary_loc): Track minus_var and minus_con.
9065 2017-08-03 Tom de Vries <tom@codesourcery.com>
9068 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9069 ACCEL_COMPILER, apply finish_options on
9070 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9072 2017-08-03 Tom de Vries <tom@codesourcery.com>
9075 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9076 function_entry_patch_area_size > 0.
9078 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9081 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9082 instead of 10??LU, perform unit multiplication in wide_int,
9083 don't change alloc_object_size_limit if the limit is larger
9086 PR tree-optimization/81655
9087 PR tree-optimization/81588
9088 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9089 the case when ranges[i].low and high are 1 for unsigned type with
9093 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9094 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9096 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9098 * tree-vrp.h: Add include guard.
9100 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9103 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9104 (ud2): New insn pattern.
9105 * config/i386/i386.c (ix86_expand_epilogue):
9106 For naked functions, generate ud2 instead of trap insn.
9108 2017-08-02 Marek Polacek <polacek@redhat.com>
9111 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9113 2017-08-02 Tom de Vries <tom@codesourcery.com>
9114 Cesar Philippidis <cesar@codesourcery.com>
9116 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9117 Add missing edge probabilities.
9119 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9121 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9124 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9127 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9128 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9129 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9130 prologue_seq sequences - if any.
9132 2017-08-02 Richard Biener <rguenther@suse.de>
9134 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9135 via vectors if supported, integer extracts via punning if supported
9136 or otherwise vector extracts.
9138 2017-08-02 Richard Biener <rguenther@suse.de>
9140 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9142 (bitmap_insert_into_set): ... this.
9144 2017-08-02 Richard Biener <rguenther@suse.de>
9146 PR tree-optimization/81633
9148 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9150 PR tree-optimization/71752
9151 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9153 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9155 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9156 (machine_function::call_ms2sysv_pad_out): Remove field.
9157 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9158 (ix86_compute_frame_layout): Likewise.
9160 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9163 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9164 attribute with interrupt attribute.
9166 2017-08-01 Andrew Pinski <apinski@cavium.com>
9168 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9169 BIT_INSERT_EXPR's operand 1
9170 to see if the types precision matches.
9172 2017-08-01 Martin Liska <mliska@suse.cz>
9175 * builtins.c (expand_builtin_memcpy_args): Remove.
9176 (expand_builtin_memcpy): Call newly added function
9177 expand_builtin_memory_copy_args.
9178 (expand_builtin_memcpy_with_bounds): Likewise.
9179 (expand_builtin_mempcpy): Remove last argument.
9180 (expand_builtin_mempcpy_with_bounds): Likewise.
9181 (expand_builtin_memory_copy_args): New function created from
9182 expand_builtin_mempcpy_args with small modifications.
9183 (expand_builtin_mempcpy_args): Remove.
9184 (expand_builtin_stpcpy): Remove unused argument.
9185 (expand_builtin): Likewise.
9186 (expand_builtin_with_bounds): Likewise.
9188 2017-08-01 Martin Liska <mliska@suse.cz>
9191 Make mempcpy more optimal (PR middle-end/70140).
9193 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9196 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9197 __builtin_vec_cmpne verify both arguments are compatible vectors
9198 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9199 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9200 move computation of aligned to after checking the argument types.
9204 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9205 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9208 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9209 Doug Rupp <rupp@adacore.com>
9210 Olivier Hainque <hainque@adacore.com>
9212 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9213 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9215 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9216 for TARGET_OS_CPP_BUILTIN.
9217 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9218 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9220 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9221 passing required abi options to the assembler for EABI configurations.
9222 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9223 of .text.hot and .text.unlikely sections for kernel modules when
9224 using ARM style exceptions.
9225 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9226 options. Add EXTRA_CC1_SPEC.
9227 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9229 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9231 (ARM_TARGET2_DWARF_FORMAT): Define.
9232 * config/arm/t-vxworks: Adjust multilib control to removal of the
9233 Diab command line options.
9235 2017-08-01 Martin Liska <mliska@suse.cz>
9237 PR gcov-profile/81561
9238 * gcov.c (unblock): Make unblocking safe as we need to preserve
9239 index correspondence of blocks and block_lists.
9241 2017-08-01 Richard Biener <rguenther@suse.de>
9243 PR tree-optimization/81181
9244 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9245 (compute_antic): ... end of iteration here.
9247 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9249 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9250 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9251 (ftree-slp-vectorize): Likewise.
9252 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9253 can no longer be set independent of flag_tree_loop_vectorize.
9254 * omp-general.c (emp_max_vf): Likewise.
9255 * opts.c (enable_fdo_optimizations): Remove references to
9256 flag_tree_vectorize, these are now implicit.
9257 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9258 and leave it for the options machinery.
9260 2017-08-01 Martin Liska <mliska@suse.cz>
9263 * builtins.c (expand_builtin_memcpy_args): Remove.
9264 (expand_builtin_memcpy): Call newly added function
9265 expand_builtin_memory_copy_args.
9266 (expand_builtin_memcpy_with_bounds): Likewise.
9267 (expand_builtin_mempcpy): Remove last argument.
9268 (expand_builtin_mempcpy_with_bounds): Likewise.
9269 (expand_builtin_memory_copy_args): New function created from
9270 expand_builtin_mempcpy_args with small modifications.
9271 (expand_builtin_mempcpy_args): Remove.
9272 (expand_builtin_stpcpy): Remove unused argument.
9273 (expand_builtin): Likewise.
9274 (expand_builtin_with_bounds): Likewise.
9276 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9279 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9280 print "ds:" only for immediates in generic address space.
9282 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9285 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9286 (ix86_function_ok_for_sibcall): Return false for naked functions.
9288 2017-08-01 Richard Biener <rguenther@suse.de>
9290 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9291 (compute_antic): Seed worklist with exit block predecessors.
9292 * cfganal.c (dfs_find_deadend): For a cycle return the source
9293 of the edge closing it.
9295 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9297 * config/aarch64/aarch64.c
9298 (aarch64_can_const_movi_rtx_p): Move 0 check.
9300 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9302 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9303 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9305 * match.pd: Ditto in address comparison pattern.
9307 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9309 PR tree-optimization/81627
9310 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9311 closed ssa form for store-store chain.
9313 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9315 PR tree-optimization/81620
9316 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9317 for store-store chain.
9319 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9321 PR tree-optimization/81588
9322 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9323 ranges[i].in_p, invert comparison code ccode. For >/>=,
9324 swap rhs1 and rhs2 and comparison code unconditionally,
9325 for </<= don't do that. Don't swap rhs1/rhs2 again if
9326 ranges[i].in_p, instead invert comparison code ccode if
9327 opcode or oe->rank is BIT_IOR_EXPR.
9330 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9331 a direct optab to conversion optab.
9332 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9333 with GET_MODE_INNER as last argument instead of optab_handler.
9334 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9335 vector extraction if possible and optab is available.
9336 * expr.c (store_constructor): Use convert_optab_handler instead
9337 of optab_handler. Use vector initialization from smaller
9338 vectors if possible and optab is available.
9339 * tree-vect-stmts.c (vectorizable_load): Likewise.
9340 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9342 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9343 of vec_init from half-sized vectors with the same element mode.
9344 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9345 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9346 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9347 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9348 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9349 after mode in gen_vec_extract* calls.
9350 (vec_extract<mode>): Renamed to ...
9351 (vec_extract<mode><ssescalarmodelower>): ... this.
9352 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9353 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9354 element mode after mode in gen_vec_init* calls.
9355 (VEC_INIT_HALF_MODE): New mode iterator.
9356 (vec_init<mode>): Renamed to ...
9357 (vec_init<mode><ssescalarmodelower>): ... this.
9358 (vec_init<mode><ssehalfvecmodelower>): New expander.
9359 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9360 (vec_extractv2sfsf): ... this.
9361 (vec_initv2sf): Renamed to ...
9362 (vec_initv2sfsf): ... this.
9363 (vec_extractv2si): Renamed to ...
9364 (vec_extractv2sisi): ... this.
9365 (vec_initv2si): Renamed to ...
9366 (vec_initv2sisi): ... this.
9367 (vec_extractv4hi): Renamed to ...
9368 (vec_extractv4hihi): ... this.
9369 (vec_initv4hi): Renamed to ...
9370 (vec_initv4hihi): ... this.
9371 (vec_extractv8qi): Renamed to ...
9372 (vec_extractv8qiqi): ... this.
9373 (vec_initv8qi): Renamed to ...
9374 (vec_initv8qiqi): ... this.
9375 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9376 (vec_init<mode>): Renamed to ...
9377 (vec_init<mode><VEC_base_l>): ... this.
9378 (vec_extract<mode>): Renamed to ...
9379 (vec_extract<mode><VEC_base_l>): ... this.
9380 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9381 (vec_initv2sfsf): ... this.
9382 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9383 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9384 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9385 element mode after mode in gen_vec_init* calls.
9386 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9387 (vec_init<mode><Vel>): ... this.
9388 (vec_extract<mode>): Renamed to ...
9389 (vec_extract<mode><Vel>): ... this.
9390 * config/aarch64/iterators.md (Vel): New mode attribute.
9391 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9392 Add element mode after mode in gen_vec_extract* calls.
9393 * config/s390/vector.md (non_vec_l): New mode attribute.
9394 (vec_extract<mode>): Renamed to ...
9395 (vec_extract<mode><non_vec_l>): ... this.
9396 (vec_init<mode>): Renamed to ...
9397 (vec_init<mode><non_vec_l>): ... this.
9398 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9399 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9401 * config/arm/iterators.md (V_elem_l): New mode attribute.
9402 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9403 (vec_extract<mode><V_elem_l>): ... this.
9404 (vec_extractv2di): Renamed to ...
9405 (vec_extractv2didi): ... this.
9406 (vec_init<mode>): Renamed to ...
9407 (vec_init<mode><V_elem_l>): ... this.
9408 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9409 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9410 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9411 Add element mode after gen_vec_extract* calls.
9412 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9413 (vec_init<mode><unitmode>): ... this.
9414 (vec_extract<mode>): Renamed to ...
9415 (vec_extract<mode><unitmode>): ... this.
9416 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9417 (vec_init<mode><unitmode>): ... this.
9418 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9419 (vec_initv2sfsf): ... this.
9420 (vec_extractv2sf): Renamed to ...
9421 (vec_extractv2sfsf): ... this.
9422 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9423 Add element mode after gen_vec_extract* calls.
9424 * config/mips/mips.md (unitmode): New mode iterator.
9425 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9426 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9427 * config/spu/spu.md (inner_l): New mode attribute.
9428 (vec_init<mode>): Renamed to ...
9429 (vec_init<mode><inner_l>): ... this.
9430 (vec_extract<mode>): Renamed to ...
9431 (vec_extract<mode><inner_l>): ... this.
9432 * config/sparc/sparc.md (veltmode): New mode iterator.
9433 (vec_init<VMALL:mode>): Renamed to ...
9434 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9435 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9436 (vec_initv2sisi): ... this.
9437 (vec_initv2sf): Renamed to ...
9438 (vec_initv2sfsf): ... this.
9439 (vec_extractv2sf): Renamed to ...
9440 (vec_extractv2sfsf): ... this.
9441 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9442 (vec_init<mode>): Renamed to ...
9443 (vec_init<mode><VEC_base_l>): ... this.
9444 (vec_extract<mode>): Renamed to ...
9445 (vec_extract<mode><VEC_base_l>): ... this.
9446 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9447 (vec_initv2sfsf): ... this.
9448 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9449 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9450 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9451 gen_vec_init* calls.
9453 2017-08-01 Richard Biener <rguenther@suse.de>
9455 PR tree-optimization/81297
9456 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9457 TREE_OVERFLOW from INTEGER_CSTs.
9459 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9461 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9463 2017-07-31 Carl Love <cel@us.ibm.com>
9465 * config/rs6000/rs6000-c: Add support for built-in functions
9466 vector signed char vec_xl_be (signed long long, signed char *);
9467 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9468 vector signed int vec_xl_be (signed long long, signed int *);
9469 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9470 vector signed long long vec_xl_be (signed long long, signed long long *);
9471 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9472 vector signed short vec_xl_be (signed long long, signed short *);
9473 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9474 vector double vec_xl_be (signed long long, double *);
9475 vector float vec_xl_be (signed long long, float *);
9476 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9477 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9478 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9480 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9481 (altivec_expand_builtin): Add switch statement to call
9482 altivec_expand_xl_be for each builtin.
9483 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9484 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9485 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9486 __builtin_vsx_le_be_v16qi.
9487 * doc/extend.texi: Update the built-in documentation file for the
9488 new built-in functions.
9490 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9493 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9495 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9497 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9499 * config.gcc: Add z14.
9500 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9501 CPU model numbers for z13s and z14.
9502 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9504 * config/s390/s390-opts.h (enum processor_type): Rename
9505 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9506 * config/s390/s390.c (processor_table): Add field for CPU name to
9507 be passed to Binutils.
9508 (s390_asm_output_machine_for_arch): Use the new field in
9509 processor_table for Binutils.
9510 (s390_expand_builtin): Replace arch12 with z14.
9511 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9512 (s390_get_sched_attrmask): Likewise.
9513 (s390_get_unit_mask): Likewise.
9514 * config/s390/s390.opt: Add z14 to processor_type enum.
9516 2017-07-31 Martin Jambor <mjambor@suse.cz>
9519 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9520 regardless of optimization level.
9522 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9523 Martin Liska <mliska@suse.cz>
9525 * predict.def: Remove old comment and adjust probability.
9526 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9529 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9532 * config/i386/i386.c (ix86_function_naked): New function.
9533 (ix86_can_use_return_insn_p): Return false for naked functions.
9534 (ix86_expand_prologue): Skip prologue for naked functions.
9535 (ix86_expand_epilogue): Skip epilogue for naked functions
9536 and emit trap instruction.
9537 (ix86_warn_func_return): New function.
9538 (ix86_attribute_table): Add "naked" attribute specification.
9539 (TARGET_WARN_FUNC_RETURN): Define.
9540 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9542 2017-07-31 Martin Liska <mliska@suse.cz>
9544 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9545 (dump_gimple_bb_header): Always dump BB info.
9546 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9548 2017-07-31 Martin Liska <mliska@suse.cz>
9551 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9552 also with current_function_decl non-null equality.
9554 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9557 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9558 change type to the element type, instead add eltype variable and
9559 use it where we are interested in the element type.
9561 PR tree-optimization/81603
9562 * ipa-polymorphic-call.c
9563 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9564 offset arithmetic in offset_int, bail out if the resulting bit offset
9565 doesn't fit into shwi.
9567 2017-07-31 Martin Liska <mliska@suse.cz>
9569 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9570 (gimplify_save_expr): Fix comment.
9572 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9575 * config/i386/i386.c (ix86_function_arg): Update arguments for
9577 (ix86_compute_frame_layout): Set the initial stack offset to
9578 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9579 INCOMING_FRAME_SP_OFFSET.
9580 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9581 stack before exception handler returns.
9582 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9583 the 'ERROR_CODE' for exception handler.
9585 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9587 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9588 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9589 (ASM_OUTPUT_REG_POP): Ditto.
9590 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9591 instead of asm_fprintf to output pure string.
9593 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9595 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9596 to imported_module_or_decl hook.
9597 (debug_nothing_tree_tree_tree_bool): Remove.
9598 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9599 * debug.c (do_nothing_debug_hooks): Use
9600 debug_nothing_tree_tree_tree_bool_bool instead of
9601 debug_nothing_tree_tree_tree_bool.
9602 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9603 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9604 * sdbout.c (sdb_debug_hooks): Likewise.
9605 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9606 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9608 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9609 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9610 attribute, don't add anything.
9612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9614 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9615 (fold_build2_stat_loc): Likewise.
9616 (fold_build3_stat_loc): Likewise.
9617 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9618 (fold_build1_loc): Remove macro.
9619 (fold_build2_loc): Likewise.
9620 (fold_build3_loc): Likewise.
9622 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9624 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9625 (gimple_build_debug_bind_source_stat): Likewise.
9626 * gimple.h (gimple_build_debug_bind): Remove macro.
9627 (gimple_build_debug_bind_source): Likewise.
9629 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9631 * bitmap.c (bitmap_alloc): Adjust.
9632 (bitmap_gc_alloc): Likewise.
9633 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9635 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9637 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9638 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9639 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9640 (bitmap_gc_alloc_stat): Likewise.
9641 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9645 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9646 * rtl.h (shallow_copy_rtx): Remove macro.
9648 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9650 * emit-rtl.c (gen_raw_REG): Adjust.
9651 * gengenrtl.c (gendef): Likewise.
9652 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9653 * rtl.h (rtx_alloc): Remove macro.
9655 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9657 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9658 (build_tree_list_stat): Likewise.
9659 * tree.h (build_tree_list): Remove macro.
9660 (build_tree_list_vec): Likewise.
9662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9664 * tree.c (make_vector_stat): Remove _stat from name.
9665 (build_vector_stat): Likewise.
9666 * tree.h (make_vector_stat): Remove macro.
9667 (build_vector_stat): Likewise.
9669 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9671 * tree.h (build_var_debug_value): Remove prototype.
9673 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9675 * tree.c (tree_cons_stat): Remove _stat from name.
9676 * tree.h (tree_cons): Remove macro.
9678 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9680 * tree.c (build_vl_exp_stat): Remove _stat from name.
9681 * tree.h (build_vl_exp): Remove macro.
9683 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9685 * tree.c (build_decl_stat): Remove _stat from name.
9686 * tree.h (build_decl): Remove macro.
9688 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9690 * gimple.c (gimple_build_with_ops_stat): Adjust.
9691 (gimple_alloc_stat): Remove _stat from name.
9692 * gimple.h (gimple_alloc): Remove macro.
9694 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9696 * tree.c (make_tree_vec_stat): Remove _stat from name.
9697 (grow_tree_vec_stat): Likewise.
9698 * tree.h (make_tree_vec_stat): Adjust prototype.
9699 (grow_tree_vec_stat): Likewise.
9700 (make_tree_vec): Remove macro.
9701 (grow_tree_vec): Likewise.
9703 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9705 * fold-const.c (fold_build1_stat_loc): Adjust.
9706 (fold_build2_stat_loc): Likewise.
9707 (fold_build3_stat_loc): Likewise.
9708 * tree.c (build0_stat): Remove _stat from name.
9709 (build1_stat): Likewise.
9710 (build2_stat): Likewise.
9711 (build3_stat): Likewise.
9712 (build4_stat): Likewise.
9713 (build5_stat): Likewise.
9714 * tree.h (build1_loc): Remove macro, and rename _stat function
9716 (build2_loc): Likewise.
9717 (build3_loc): Likewise.
9718 (build4_loc): Likewise.
9719 (build5_loc): Likewise.
9721 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9723 * tree.c (make_int_cst_stat): Remove _stat from name.
9724 * tree.h (make_int_cst_stat): Adjust prototype.
9725 (make_int_cst): Remove macro.
9727 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9729 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9730 * tree.h (make_tree_binfo_stat): Adjust prototype.
9731 (make_tree_binfo): Remove.
9733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9735 * tree.c (copy_node_stat): Rename to copy_node.
9736 (build_distinct_type_copy): Adjust.
9737 * tree.h (copy_node_stat): Adjust prototype.
9738 (copy_node): Remove macro.
9740 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9742 * tree.c (make_node_stat): rename to make_node.
9743 (build_tree_list_stat): Adjust.
9744 (build0_stat): Likewise.
9745 (build2_stat): Likewise.
9746 (build3_stat): Likewise.
9747 (build4_stat): Likewise.
9748 (build5_stat): Likewise.
9749 (build_decl_stat): Likewise.
9750 * tree.h (make_node_stat): Adjust prototype.
9751 (make_node): remove macro.
9753 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9755 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9756 (PPC_FEATURE2_SCV): Likewise.
9757 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9759 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9761 * config/aarch64/aarch64.c
9762 (aarch64_internal_mov_immediate): Add new special pattern.
9763 * config/aarch64/aarch64.md (*movdi_aarch64):
9764 Add reg/32bit const mov case.
9766 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9767 Richard Sandiford <richard.sandiford@linaro.org>
9769 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9770 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9771 Add integer and movi cases.
9772 (movi-split-hf-df-sf split, fp16): New.
9773 (enabled): Added TARGET_FP_F16INST.
9774 * config/aarch64/iterators.md (GPF_HF): New.
9775 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9777 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9779 * config/aarch64/aarch64.c
9780 (aarch64_simd_container_mode): Add prototype.
9781 (aarch64_expand_mov_immediate): Add HI support.
9782 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9783 (aarch64_can_const_movi_rtx_p): New.
9784 (aarch64_preferred_reload_class):
9785 Remove restrictions of using FP registers for certain SIMD operations.
9786 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9787 (aarch64_valid_floating_const): Add integer move validation.
9788 (aarch64_simd_imm_scalar_p): Remove.
9789 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9790 (aarch64_legitimate_constant_p): Expand list of supported cases.
9791 * config/aarch64/aarch64-protos.h
9792 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9793 (aarch64_reinterpret_float_as_int): New.
9794 (aarch64_simd_imm_scalar_p): Remove.
9795 * config/aarch64/constraints.md (Uvi): New.
9796 (Dd): Split into Ds and new Dd.
9797 * config/aarch64/aarch64.md (*movsi_aarch64):
9799 (*movdi_aarch64): Add SIMD mov case.
9801 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9803 * tree-predcom.c: (struct chain): Handle store-store chain in which
9804 stores for elimination only store loop invariant values.
9805 (execute_pred_commoning_chain): Ditto.
9806 (prepare_initializers_chain_store_elim): Ditto.
9807 (prepare_finalizers): Ditto.
9808 (is_inv_store_elimination_chain): New function.
9809 (initialize_root_vars_store_elim_1): New function.
9811 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9813 * tree-predcom.c: Revise general description of the pass.
9814 (enum chain_type): New enum type for store elimination.
9815 (struct chain): New field supporting store elimination.
9816 (struct component): Ditto.
9817 (dump_chain): Dump store-stores chain.
9818 (release_chain): Release resources.
9819 (split_data_refs_to_components): Compute and create component
9820 contains only stores for elimination.
9821 (get_chain_last_ref_at): New function.
9822 (make_invariant_chain): Initialization.
9823 (make_rooted_chain): Specify chain type in parameter and record it.
9824 (add_looparound_copies): Skip for store-stores chain.
9825 (determine_roots_comp): Compute type of chain and pass it to
9827 (initialize_root_vars_store_elim_2): New function.
9828 (finalize_eliminated_stores): New function.
9829 (remove_stmt): Handle store for elimination.
9830 (execute_pred_commoning_chain): Execute predictive commoning on
9832 (determine_unroll_factor): Skip unroll for store-stores chain.
9833 (prepare_initializers_chain_store_elim): New function.
9834 (prepare_initializers_chain): Hanlde store-store chain.
9835 (prepare_finalizers_chain, prepare_finalizers): New function.
9836 (tree_predictive_commoning_loop): Return integer value indicating
9837 if loop is unrolled or lcssa form is corrupted.
9838 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9840 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9842 * tree-predcom.c (initialize_root): Delete.
9843 (execute_pred_commoning_chain): Initialize root vars and replace
9844 reference of non-combined chain directly, rather than call above
9847 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9849 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
9850 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
9852 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9854 * tree-predcom.c (struct chain): New field init_seq.
9855 (release_chain): Release init_seq.
9856 (prepare_initializers_chain): Record intialization stmts in above
9858 (insert_init_seqs): New function.
9859 (tree_predictive_commoning_loop): Call insert_init_seqs.
9861 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9863 * tree-predcom.c (determine_roots_comp): Skip trivial components.
9865 2017-07-28 Richard Biener <rguenther@suse.de>
9867 * match.pd: Remove superfluous :c.
9868 * genmatch.c (simplify::id): Add member.
9869 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
9871 (current_id): New global.
9872 (dt_node::parent): Move from ...
9873 (dt_operand::parent): ... here. Add for_id member.
9874 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
9875 (decision_tree::find_node): Relax order requirement when
9876 merging DT_TRUE nodes to ones inbetween the current simplify
9877 and the one we try to merge with. Add diagnostic whenever
9878 we need to enforce pattern order by not merging.
9879 (decision_tree::insert): Set current_id.
9880 (decision_tree::print_node): Dump parent node and for_id.
9881 (parser::last_id): Add member.
9882 (parser::push_simplify): Assign unique id.
9883 (parser::parser): Initialize last_id.
9885 2017-07-28 Martin Liska <mliska@suse.cz>
9888 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
9889 gimple_build_debug_bind.
9891 2017-07-28 Richard Biener <rguenther@suse.de>
9893 PR tree-optimization/81502
9894 * match.pd: Add pattern combining BIT_INSERT_EXPR with
9896 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
9898 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
9899 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
9900 for BIT_FIELD_REF args.
9901 * fold-const.c (make_bit_field_ref): Likewise.
9902 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9904 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9907 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
9908 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
9909 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
9910 Or it into SANITIZER_UNDEFINED.
9911 * ubsan.c: Include gimple-fold.h and varasm.h.
9912 (ubsan_expand_ptr_ifn): New function.
9913 (instrument_pointer_overflow): New function.
9914 (maybe_instrument_pointer_overflow): New function.
9915 (instrument_object_size): Formatting fix.
9916 (pass_ubsan::execute): Call instrument_pointer_overflow
9917 and maybe_instrument_pointer_overflow.
9918 * internal-fn.c (expand_UBSAN_PTR): New function.
9919 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
9920 * sanitizer.def (__ubsan_handle_pointer_overflow,
9921 __ubsan_handle_pointer_overflow_abort): New builtins.
9922 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
9923 * internal-fn.def (UBSAN_PTR): New internal function.
9924 * opts.c (sanitizer_opts): Add pointer-overflow.
9925 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
9926 * fold-const.c (build_range_check): Compute pointer range check in
9927 integral type if pointer arithmetics would be needed. Formatting
9930 2017-07-28 Martin Liska <mliska@suse.cz>
9933 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
9934 parameters that are of a variable-length.
9936 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
9938 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
9940 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
9941 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
9942 (CRT_CALL_STATIC_FUNCTION): Likewise.
9943 (ASM_DEFAULT_SPEC): New define.
9944 (ASM_SPEC32): Likewise.
9945 (ASM_SPEC64): Likewise.
9946 (ASM_SPEC_COMMON): Likewise.
9947 (ASM_SPEC): Likewise.
9948 (INVALID_64BIT): Likewise.
9949 (LINK_OS_DEFAULT_SPEC): Likewise.
9950 (LINK_OS_SPEC32): Likewise.
9951 (LINK_OS_SPEC64): Likewise.
9952 (POWERPC_LINUX): Likewise.
9953 (PTRDIFF_TYPE): Likewise.
9954 (RESTORE_FP_PREFIX): Likewise.
9955 (RESTORE_FP_SUFFIX): Likewise.
9956 (SAVE_FP_PREFIX): Likewise.
9957 (SAVE_FP_SUFFIX): Likewise.
9958 (SIZE_TYPE): Likewise.
9959 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
9960 (TARGET_64BIT): Likewise.
9961 (TARGET_64BIT): Likewise.
9962 (TARGET_AIX): Likewise.
9963 (WCHAR_TYPE_SIZE): Likewise.
9964 (WCHAR_TYPE): Undefine.
9965 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
9966 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
9967 (CPP_OS_RTEMS_SPEC): Delete.
9968 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
9969 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
9971 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
9973 2017-07-28 Jakub Jelinek <jakub@redhat.com>
9975 PR tree-optimization/81578
9976 * tree-parloops.c (build_new_reduction): Bail out if
9977 reduction_code isn't one of the standard OpenMP reductions.
9978 Move the details printing after that decision.
9980 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
9982 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
9983 related to reload_in_progress.
9984 (splat_input_operand): Likewise.
9985 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
9987 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
9989 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
9990 (TARGET_INSTANTIATE_DECLS): Likewise.
9991 (legitimate_indexed_address_p): Delete reload_in_progress code.
9992 (rs6000_debug_legitimate_address_p): Likewise.
9993 (rs6000_eliminate_indexed_memrefs): Likewise.
9994 (rs6000_emit_le_vsx_store): Likewise.
9995 (rs6000_emit_move_si_sf_subreg): Likewise.
9996 (rs6000_emit_move): Likewise.
9997 (register_to_reg_type): Likewise.
9998 (rs6000_pre_atomic_barrier): Likewise.
9999 (rs6000_machopic_legitimize_pic_address): Likewise.
10000 (rs6000_allocate_stack_temp): Likewise.
10001 (rs6000_address_for_fpconvert): Likewise.
10002 (rs6000_address_for_altivec): Likewise.
10003 (rs6000_secondary_memory_needed_rtx): Delete function.
10004 (rs6000_check_sdmode): Likewise.
10005 (rs6000_alloc_sdmode_stack_slot): Likewise.
10006 (rs6000_instantiate_decls): Likewise.
10007 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10008 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10009 Delete reload_in_progress.
10010 (*vec_reload_and_plus_<mptrsize>): Likewise.
10011 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10012 (vsx_div_v2di): Likewise.
10013 (vsx_udiv_v2di): Likewise.
10015 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10017 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10018 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10019 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10020 (rs6000_debug_reg_global): Delete print of LRA status.
10021 (rs6000_option_override_internal): Delete dead LRA related code.
10022 (rs6000_lra_p): Delete function.
10023 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10025 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10027 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10028 * config/riscv/rtems.h: New file.
10030 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10031 Sudakshina Das <sudi.das@arm.com>
10033 * config/aarch64/aarch64.md
10034 (define_split for and<mode>3nr_compare): Move
10035 non aarch64_logical_operand to a register.
10036 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10037 register immediate operand to a register.
10038 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10040 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10042 PR middle-end/81564
10043 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10045 2017-07-27 Richard Biener <rguenther@suse.de>
10047 PR tree-optimization/81573
10048 PR tree-optimization/81494
10049 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10050 multi defuse cycle case.
10052 2017-07-27 Richard Biener <rguenther@suse.de>
10054 PR tree-optimization/81571
10055 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10058 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10060 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10061 earlier and only if MASK_FPU is set. Adjust formatting.
10063 2017-07-27 Martin Liska <mliska@suse.cz>
10065 * opt-functions.awk: Add validation of value of Init.
10066 * optc-gen.awk: Pass new argument.
10068 2017-07-27 Martin Liska <mliska@suse.cz>
10070 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10071 Fix wrong condition.
10073 2017-07-27 Martin Liska <mliska@suse.cz>
10075 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10076 BBs and edges seen by autoFDO.
10078 2017-07-27 Richard Biener <rguenther@suse.de>
10080 PR tree-optimization/81502
10081 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10082 with incompatible but same sized type.
10083 (execute_update_addresses_taken): Likewise.
10085 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10087 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10088 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10090 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10093 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10094 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10095 Change s_operand to memory_operand.
10097 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10099 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10100 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10101 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10102 Emit instructions rather than returning an expression. Handle TFmode
10103 and KFmode by casting to TImode.
10104 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10105 (rs6000_emit_le_vsx_store): Likewise.
10106 * config/rs6000/vsx.md (VSX_TI): New iterator.
10107 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10108 (*vsx_le_undo_permute_<mode>): Likewise.
10109 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10110 emit the split sequence.
10111 (*vsx_le_perm_store_<mode>): Likewise.
10113 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10115 PR tree-optimization/81555
10116 PR tree-optimization/81556
10117 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10118 if true, force CHANGED for the recursive invocation.
10119 (reassociate_bb): Remember original length of ops array, pass
10120 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10122 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10123 attributes for noipa attribute. For naked attribute use
10124 lookup_attribute first before lookup_attribute_spec.
10125 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10127 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10128 for functions with noipa attribute.
10129 (cgraph_externally_visible_p): Return true for functions with noipa
10131 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10132 for functions with noipa attribute.
10133 * doc/extend.texi: Document noipa function attribute.
10134 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10135 also for functions with noipa attribute.
10136 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10138 2017-07-26 Andrew Pinski <apinski@cavium.com>
10140 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10141 vec_unalign_load_cost and vec_unalign_store_cost.
10143 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10145 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10146 -mvsx-small-integer option.
10147 (ISA_3_0_MASKS_IEEE): Likewise.
10148 (OTHER_VSX_VECTOR_MASKS): Likewise.
10149 (POWERPC_MASKS): Likewise.
10150 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10151 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10152 code, only testing for DImode being allowed in non-VSX floating
10154 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10155 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10157 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10158 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10159 TARGET_P8_VECTOR test.
10160 (rs6000_secondary_reload_simple_move): Likewise.
10161 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10162 since TARGET_P9_VECTOR was already tested.
10163 (rs6000_opt_masks): Remove -mvsx-small-integer.
10164 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10165 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10167 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10168 test for TARGET_VEXTRACTUB was used, and that uses
10170 (p9 extract splitter): Likewise.
10171 (vsx_extract_<mode>_di_p9): Likewise.
10172 (vsx_extract_<mode>_store_p9): Likewise.
10173 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10174 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10175 the elimination of TARGET_VSX_SMALL_INTEGER.
10176 (vsx_extract_<mode>_p8): Likewise.
10177 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10178 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10179 (vsx_set_<mode>_p9): Likewise.
10180 (vsx_set_v4sf_p9): Likewise.
10181 (vsx_set_v4sf_p9_zero): Likewise.
10182 (vsx_insert_extract_v4sf_p9): Likewise.
10183 (vsx_insert_extract_v4sf_p9_2): Likewise.
10184 * config/rs6000/rs6000.md (sign extend splitter): Change
10185 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10186 (floatsi<mode>2_lfiwax_mem): Likewise.
10187 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10188 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10189 since a test for TARGET_P9_VECTOR was used.
10190 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10191 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10192 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10193 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10194 TARGET_P8_VECTOR test.
10195 (fix_trunc<mode>si2_stfiwx): Likewise.
10196 (fix_trunc<mode>si2_internal): Likewise.
10197 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10198 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10200 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10201 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10202 TARGET_P8_VECTOR test.
10203 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10204 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10205 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10207 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10208 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10209 since a test for TARGET_P9_VECTOR was used.
10210 (splitter for loading small constants): Likewise.
10212 2017-07-26 Andrew Pinski <apinski@cavium.com>
10214 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10217 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10220 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10221 (fp_valid_at): Likewise.
10223 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10225 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10226 (qdf24xx_addrcost_table): Likewise.
10227 (cortexa57_tunings): Update to use generic_branch_cost.
10228 (cortexa72_tunings): Likewise.
10229 (cortexa73_tunings): Likewise.
10230 (qdf24xx_tunings): Likewise.
10232 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10234 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10235 (thunderx2t99_branch_cost): Likewise.
10236 (cortexa35_tunings): Update to use generic_branch_cost.
10237 (cortexa53_tunings): Likewise.
10238 (cortexa57_tunings): Likewise.
10239 (cortexa72_tunings): Likewise.
10240 (cortexa73_tunings): Likewise.
10241 (thunderx2t99_tunings): Likewise.
10243 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10245 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10246 (sparc_option_override): Honour MASK_FSMULD.
10247 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10248 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10249 * config/sparc/sparc.opt (mfsmuld): New option.
10250 * doc/invoke.texi (mfsmuld): Document option.
10252 2017-07-26 Marek Polacek <polacek@redhat.com>
10254 PR middle-end/70992
10255 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10257 2017-07-26 Richard Biener <rguenther@suse.de>
10259 * gimple-match-head.c (do_valueize): Return OP if valueize
10261 (get_def): New helper to get at the def stmt of a SSA name
10262 if valueize allows.
10263 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10264 do_valueize to get at the def stmt.
10265 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10267 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10269 PR middle-end/46932
10270 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10272 2017-07-26 Martin Liska <mliska@suse.cz>
10275 * function.c (expand_function_start): Make expansion of
10276 nonlocal_goto_save_area after parm_birth_insn.
10278 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10280 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10281 from all CPU target flags enable members.
10283 2017-07-26 Richard Biener <rguenther@suse.de>
10285 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10286 (decision_tree::gen): Make 'type' const.
10287 (write_predicate): Likewise.
10289 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10291 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10292 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10293 (rs6000_option_override_internal): Likewise.
10294 (rs6000_expand_vector_set): Likewise.
10295 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10296 (TARGET_UPPER_REGS_SF): Likewise.
10297 (TARGET_UPPER_REGS_DI): Likewise.
10298 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10299 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10300 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10301 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10302 (Splitters for DI constants in Altivec registers): Likewise.
10303 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10304 (vsx_set_v4sf_p9): Likewise.
10305 (vsx_set_v4sf_p9_zero): Likewise.
10306 (vsx_insert_extract_v4sf_p9): Likewise.
10307 (vsx_insert_extract_v4sf_p9_2): Likewise.
10309 2017-07-25 Carl Love <cel@us.ibm.com>
10311 * doc/extend.texi: Update the built-in documentation file for the
10312 existing built-in functions
10313 vector signed char vec_cnttz (vector signed char);
10314 vector unsigned char vec_cnttz (vector unsigned char);
10315 vector signed short vec_cnttz (vector signed short);
10316 vector unsigned short vec_cnttz (vector unsigned short);
10317 vector signed int vec_cnttz (vector signed int);
10318 vector unsigned int vec_cnttz (vector unsigned int);
10319 vector signed long long vec_cnttz (vector signed long long);
10320 vector unsigned long long vec_cnttz (vector unsigned long long);
10322 2017-07-25 Andrew Pinski <apinski@cavium.com>
10324 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10325 accesses where the use is for the first operand of a BIT_INSERT.
10327 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10330 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10331 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10333 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10335 * config/i386/gstabs.h: Delete.
10336 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10338 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10340 * config/i386/i386.c (ix86_decompose_address): Do not check for
10341 register RTX when looking at index_reg or base_reg.
10342 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10344 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10346 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10347 to update EH info here.
10349 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10351 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10353 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10355 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10357 2017-07-25 Torsten Duwe <duwe@suse.de>
10359 * common.opt: Introduce -fpatchable-function-entry
10360 command line option, and its variables function_entry_patch_area_size
10361 and function_entry_patch_area_start.
10362 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10363 including a two-value parser.
10364 * target.def (print_patchable_function_entry): New target hook.
10365 * targhooks.h (default_print_patchable_function_entry): New function.
10366 * targhooks.c (default_print_patchable_function_entry): Likewise.
10367 * toplev.c (process_options): Switch off IPA-RA if
10368 patchable function entries are being generated.
10369 * varasm.c (assemble_start_function): Look at the
10370 patchable-function-entry command line switch and current
10371 function attributes and maybe generate NOP instructions by
10372 calling the print_patchable_function_entry hook.
10373 * doc/extend.texi: Document patchable_function_entry attribute.
10374 * doc/invoke.texi: Document -fpatchable_function_entry
10375 command line option.
10376 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10378 * doc/tm.texi: Re-generate.
10380 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10383 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10384 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10386 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10388 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10390 2017-07-25 Richard Biener <rguenther@suse.de>
10392 PR tree-optimization/81455
10393 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10394 not walk in cycles when looking for guards.
10396 2017-07-25 Richard Biener <rguenther@suse.de>
10398 PR tree-optimization/81529
10399 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10400 when optimizing backedge uses.
10402 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10404 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10406 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10407 to dl_section_ref. On AIX, append an expression to subtract
10408 the size of the section length to dl_section_ref.
10410 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10412 * configure.ac: If any of the config.* scripts fail, exit 1.
10413 * configure: Regenerate.
10415 2017-07-25 Richard Biener <rguenther@suse.de>
10417 PR middle-end/81546
10418 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10419 of immediate uses, be more verbose on errors.
10421 2017-07-25 Richard Biener <rguenther@suse.de>
10423 PR tree-optimization/81510
10424 * tree-vect-loop.c (vect_is_simple_reduction): When the
10425 reduction stmt is not inside the loop bail out.
10427 2017-07-25 Richard Biener <rguenther@suse.de>
10429 PR tree-optimization/81303
10430 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10431 profitability check against LOOP_VINFO_NITERSM1.
10433 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10435 * domwalk.c (cmp_bb_postorder): Simplify.
10436 (sort_bbs_postorder): New function. Use it...
10437 (dom_walker::walk): ...here to optimize common cases.
10439 2017-07-25 Martin Liska <mliska@suse.cz>
10442 * ipa-visibility.c (function_and_variable_visibility): Make the
10443 redirection just on target that supports aliasing.
10444 Fix GNU coding style.
10446 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10449 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10451 * config/arm/rtems.h: Likewise.
10452 * config/bfin/rtems.h: Likewise.
10453 * config/i386/rtemself.h: Likewise.
10454 * config/lm32/rtems.h: Likewise.
10455 * config/m32c/rtems.h: Likewise.
10456 * config/m68k/rtemself.h: Likewise.
10457 * config/microblaze/rtems.h: Likewise.
10458 * config/mips/rtems.h: Likewise.
10459 * config/moxie/rtems.h: Likewise.
10460 * config/nios2/rtems.h: Likewise.
10461 * config/powerpcspe/rtems.h: Likewise.
10462 * config/rs6000/rtems.h: Likewise.
10463 * config/rtems.h: Likewise.
10464 * config/sh/rtems.h: Likewise.
10465 * config/sh/rtemself.h: Likewise.
10466 * config/sparc/rtemself.h: Likewise.
10468 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10471 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10472 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10473 * tree-ssa-structalias.c (alias_get_name): Same.
10475 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10478 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10479 instructions if no du chain is found.
10481 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10483 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10485 2017-07-25 Richard Biener <rguenther@suse.de>
10487 PR middle-end/81505
10488 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10491 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10493 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10494 upper-regs options.
10495 (ISA_2_7_MASKS_SERVER): Likewise.
10496 (ISA_3_0_MASKS_IEEE): Likewise.
10497 (OTHER_P8_VECTOR_MASKS): Likewise.
10498 (OTHER_VSX_VECTOR_MASKS): Likewise.
10499 (POWERPC_MASKS): Likewise.
10500 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10501 duplicate list of options.
10502 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10503 explicit -mupper-regs options.
10504 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10505 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10506 alias for -mupper-regs-df.
10507 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10508 (rs6000_init_hard_regno_mode_ok): Likewise.
10509 (rs6000_option_override_internal): Likewise.
10510 (rs6000_opt_masks): Likewise.
10511 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10512 options in terms of whether -mvsx or -mpower8-vector was used.
10513 (TARGET_UPPER_REGS_DI): Likewise.
10514 (TARGET_UPPER_REGS_SF): Likewise.
10515 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10516 -mupper-regs-* options.
10518 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10520 * passes.c (emergency_dump_function): Print some empty lines and a
10521 header before the RTL dump.
10523 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10525 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10527 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10530 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10531 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10533 2017-07-24 Carl Love <cel@us.ibm.com>
10535 * config/rs6000/rs6000-c.c: Add support for built-in functions
10536 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10537 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10538 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10539 vec_extract_fp_from_shortl): Add defines for the two builtins.
10540 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10541 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10543 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10544 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10545 * doc/extend.texi: Update the built-in documentation file for the
10546 new built-in function.
10548 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10551 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10553 * doc/generic.texi: Likewise.
10554 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10555 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10557 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10559 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10560 (aarch64_mls_elt_merge<mode>): Likewise.
10562 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10564 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10565 having __cxa_atexit.
10567 2017-07-23 Michael Collison <michael.collison@arm.com>
10569 * config/arm/arm.c (arm_option_override): Deprecate
10570 use of -mstructure-size-boundary.
10571 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10572 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10574 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10577 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10578 Reduce cost estimate for direct moves.
10580 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10583 * config/i386/i386.c (ix86_option_override_internal): Disable
10584 BMI, BMI2 and TBM instructions for -m16.
10586 2017-07-21 Carl Love <cel@us.ibm.com>
10588 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10589 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10590 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10591 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10592 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10593 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10594 VMULOSW): New enum "unspec" values.
10595 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10596 altivec_vmulosw): New patterns.
10597 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10598 VMULOSW): Add definitions.
10600 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10602 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10603 (qdf24xx): Likewise.
10604 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10605 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10606 (AARCH64_FL_V8_1): Renumber.
10607 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10608 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10609 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10610 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10611 rdma to feature modifiers list.
10613 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10615 PR middle-end/56727
10616 * ipa-visibility (function_and_variable_visibility): Convert
10617 recursive PLT call to direct call if appropriate.
10619 2017-07-21 Andrew Pinski <apinski@cavium.com>
10621 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10622 operand 1 to see if the types precision matches.
10623 * fold-const.c (operand_equal_p): Likewise.
10625 2017-07-21 Richard Biener <rguenther@suse.de>
10627 PR tree-optimization/81303
10628 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10629 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10630 (vect_peeling_hash_get_lowest_cost): Adjust.
10631 (vect_enhance_data_refs_alignment): Likewise. Use
10632 vect_get_peeling_costs_all_drs to compute the penalty for no
10633 peeling to match up costs.
10635 2017-07-21 Richard Biener <rguenther@suse.de>
10637 PR tree-optimization/81500
10638 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10639 we didn't identify a reduction path.
10641 2017-07-21 Tom de Vries <tom@codesourcery.com>
10642 Cesar Philippidis <cesar@codesourcery.com>
10644 PR gcov-profile/81442
10645 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10648 2017-07-21 Tom de Vries <tom@codesourcery.com>
10651 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10653 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10654 nvptx_override_options_after_change.
10656 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10658 * dwarf2out.c (output_file_names): Avoid double testing for
10659 dwarf_version >= 5.
10661 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10663 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10665 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10667 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10669 (try_crossjump_to_edge): Do not punt on partitioned functions.
10671 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10673 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10674 Put all BBs reachable only via paths crossing cold region to cold
10676 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10678 2016-07-21 Richard Biener <rguenther@suse.de>
10680 PR tree-optimization/81303
10681 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10682 into account prologue and epilogue iterations when raising
10683 min_profitable_iters to sth at least covering one vector iteration.
10685 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10687 * config/arm/arm.c (arm_test_cpu_arch_dat):
10690 2017-07-20 Nathan Sidwell <nathan@acm.org>
10692 Remove TYPE_METHODS.
10693 * tree.h (TYPE_METHODS): Delete.
10694 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10695 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10696 (dbxout_type_methods): Scan TYPE_FIELDS.
10697 (dbxout_type): Don't check TYPE_METHODS here.
10698 * function.c (use_register_for_decl): Always ignore register for
10699 class types when not optimizing.
10700 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10701 * tree.c (free_lang_data_in_type): Stitch out member functions and
10702 templates from TYPE_FIELDS.
10703 (build_distinct_type_copy, verify_type_variant,
10704 verify_type): Member fns are on TYPE_FIELDS.
10705 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10706 * tree-pretty-print.c (dump_generic_node): Likewise.
10708 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10711 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10712 V2TImode and V4TImode.
10713 (ix86_expand_vector_extract): Likewise.
10714 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10715 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10716 (ssescalarmode): Handle V4TImode and V2TImode.
10717 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10718 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10719 (VEXTRACTI128_MODE): New mode iterator.
10720 (splitter for *vec_extractv?ti first element): New.
10721 (VEC_INIT_MODE): New mode iterator.
10722 (vec_init<mode>): Consolidate 3 expanders into one using
10723 VEC_INIT_MODE mode iterator.
10725 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10727 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10728 non_spilled_static_chain_regno_p.
10730 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10732 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10734 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10736 * bb-reorder.c (connect_traces): Allow copying of blocks within
10739 2017-07-20 Richard Biener <rguenther@suse.de>
10741 * gimple.h (gimple_phi_result): Add gphi * overload.
10742 (gimple_phi_result_ptr): Likewise.
10743 (gimple_phi_arg): Likewise. Adjust index assert to only
10744 allow actual argument accesses rather than all slots available
10746 (gimple_phi_arg_def): Add gphi * overload.
10747 * tree-phinodes.c (make_phi_node): Initialize only actual
10749 (resize_phi_node): Clear memory not covered by old node,
10750 do not initialize excess argument slots.
10751 (reserve_phi_args_for_new_edge): Initialize new argument slot
10754 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10756 PR tree-optimization/81388
10758 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10760 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10761 by removing computation of may_be_zero.
10763 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10764 Tom de Vries <tom@codesourcery.com>
10766 PR middle-end/81030
10767 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10768 when gimple level profile disagrees with what RTL expander did.
10770 2017-07-20 Richard Biener <rguenther@suse.de>
10772 PR tree-optimization/61171
10773 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10774 (vect_analyze_stmt): Add slp instance parameter.
10775 (vectorizable_reduction): Likewise.
10776 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10777 (vect_is_simple_reduction): Deal with chains not detected
10778 as SLP reduction chain, specifically not properly associated
10779 chains containing a mix of plus/minus.
10780 (get_reduction_op): Remove.
10781 (get_initial_defs_for_reduction): Simplify, pass in whether
10782 this is a reduction chain, pass in the SLP node for the PHIs.
10783 (vect_create_epilog_for_reduction): Get the SLP instance as
10785 (vectorizable_reduction): Get the SLP instance as arg.
10786 During analysis remember the SLP node with the PHIs in the
10787 instance. Simplify getting at the vectorized reduction PHIs.
10788 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10789 through SLP instance.
10790 (vect_slp_analyze_operations): Likewise.
10791 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10792 (vect_transform_stmt): Likewise.
10794 2017-07-20 Tom de Vries <tom@codesourcery.com>
10796 PR tree-optimization/81489
10797 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10798 read of phi arg location to before loop that modifies phi.
10800 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10802 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10805 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10807 PR middle-end/81331
10808 * except.c (execute): Fix ordering issue.
10810 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10812 PR rtl-optimization/81423
10813 * combine.c (make_compound_operation_int): Don't try to optimize
10814 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10816 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10818 PR rtl-optimization/81423
10819 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10820 with a constant that is -1 in the truncated to mode.
10822 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10824 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10825 (determine_unlikely_bbs): ... here.
10826 * predict.h (propagate_unlikely_bbs_forward): Declare.
10827 * cfgexpand.c (pass_expand::execute): Use it.
10828 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10830 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10831 propagate_unlikely_bbs_forward.
10833 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10835 PR middle-end/81331
10836 * except.c (maybe_add_nop_after_section_switch): New function.
10839 2017-07-19 Tom de Vries <tom@codesourcery.com>
10841 * gimple.h (gimple_phi_set_arg): Make assert more strict.
10843 2017-07-19 Tom de Vries <tom@codesourcery.com>
10845 * gimple.h (gimple_phi_arg): Make assert more strict.
10847 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
10849 * config.gcc (powerpc*-*-*): Add mmintrin.h.
10850 * config/rs6000/mmintrin.h: New file.
10851 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
10853 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10855 PR tree-optimization/81346
10856 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
10858 2017-07-19 Tom de Vries <tom@codesourcery.com>
10860 * config/nvptx/nvptx.md (VECIM): Add V2DI.
10862 2017-07-19 Tom de Vries <tom@codesourcery.com>
10864 * config/nvptx/nvptx-modes.def: Add V2DImode.
10865 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
10866 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
10867 (nvptx_output_mov_insn): Handle lack of mov.b128.
10868 (nvptx_print_operand): Handle 'H' and 'L' codes.
10869 (nvptx_vector_mode_supported): Allow V2DImode.
10870 (nvptx_preferred_simd_mode): New function.
10871 (nvptx_data_alignment): New function.
10872 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
10873 nvptx_preferred_simd_mode.
10874 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
10876 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
10878 2017-07-19 Tom de Vries <tom@codesourcery.com>
10880 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
10881 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
10882 (nvptx_vector_mode_supported): New function. Allow V2SImode.
10883 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
10884 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
10885 (mov<VECIM>_insn): New define_insn.
10886 (define_expand "mov<VECIM>): New define_expand.
10888 2017-07-19 Tom de Vries <tom@codesourcery.com>
10890 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
10892 2017-07-19 Jakub Jelinek <jakub@redhat.com>
10894 PR tree-optimization/81346
10895 * fold-const.h (fold_div_compare, range_check_type): Declare.
10896 * fold-const.c (range_check_type): New function.
10897 (build_range_check): Use range_check_type.
10898 (fold_div_compare): No longer static, rewritten into
10899 a match.pd helper function.
10900 (fold_comparison): Don't call fold_div_compare here.
10901 * match.pd (X / C1 op C2): New optimization using fold_div_compare
10902 as helper function.
10904 2017-07-19 Nathan Sidwell <nathan@acm.org>
10906 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
10907 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
10908 * tree.c (find_decls_types_r, verify_type): Use
10909 TYPE_{MIN,MAX}_VALUE_RAW.
10910 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
10911 (hash_tree): Likewise.
10912 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10914 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10917 2017-07-18 Tom de Vries <tom@codesourcery.com>
10919 PR middle-end/81464
10920 * omp-expand.c (expand_omp_for_static_chunk): Handle
10921 equal-argument loop exit phi.
10923 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
10926 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
10927 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
10928 operand 2 predicate.
10929 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
10930 operand 2 predicate.
10931 (ror,rol -> rorx splitters): Use const_int_operand as
10932 operand 2 predicate.
10934 2017-06-18 Richard Biener <rguenther@suse.de>
10936 PR tree-optimization/81410
10937 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
10938 the gap in the ! slp_perm SLP case after each group.
10940 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10942 PR middle-end/81463
10943 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
10946 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10948 PR middle-end/81462
10949 * predict.c (set_even_probabilities): Cleanup; do not affect
10950 probabilities that are already known.
10951 (combine_predictions_for_bb): Call even when count is set.
10953 2017-07-18 Nathan Sidwell <nathan@acm.org>
10955 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
10958 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10961 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
10962 optimization for loop niter analysis.
10964 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
10967 * config/avr/avr.c (avr_optimize_casesi): Don't use
10968 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
10970 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10972 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
10973 body_cost_vec from _vect_peel_extended_info.
10974 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
10975 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
10978 2017-07-18 Bin Cheng <bin.cheng@arm.com>
10980 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
10982 2017-07-18 Richard Biener <rguenther@suse.de>
10984 PR tree-optimization/80620
10985 PR tree-optimization/81403
10986 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
10987 info when re-using a VN table entry.
10989 2017-07-18 Richard Biener <rguenther@suse.de>
10991 PR tree-optimization/81418
10992 * tree-vect-loop.c (vectorizable_reduction): Properly compute
10993 vectype_in. Verify that with lane-reducing reduction operations
10994 we have a single def-use cycle.
10996 2017-07-17 Carl Love <cel@us.ibm.com>
10998 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11000 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11001 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11002 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11003 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11004 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11005 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11006 VMULOSW): New enum "unspec" values.
11007 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11008 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11009 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11010 altivec_vmulosw): New patterns.
11011 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11012 VMULOSW): Add definitions.
11013 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11015 * config/alpha/alpha.c: Include predict.h.
11017 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11019 * tree-vrp.c (compare_assert_loc): Fix comparison function
11020 to return predictable results.
11022 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11024 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11026 (subdi3): Likewise.
11027 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11028 * doc/invoke.texi (mexpand-adddi): Update text.
11030 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11032 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11033 that also clobbers the CC register. The old expand code is moved
11035 (*arc_clzsi2): ... here.
11036 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11037 the CC register. The old expand code is moved to ...
11038 (arc_ctzsi2): ... here.
11040 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11042 * config/arc/arc.opt (mindexed-loads): Use initial value
11043 TARGET_INDEXED_LOADS_DEFAULT.
11044 (mauto-modify-reg): Use initial value
11045 TARGET_AUTO_MODIFY_REG_DEFAULT.
11046 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11047 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11048 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11049 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11051 2017-07-17 Martin Liska <mliska@suse.cz>
11054 * opts.c (finish_options): Do not allow -fgnu-tm
11055 w/ -fsanitize={kernel-,}address. Say sorry.
11057 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11060 * tree-loop-distribution.c (classify_partition): Only assert on
11061 numer of iterations.
11062 (merge_dep_scc_partitions): Delete prameter. Update function call.
11063 (distribute_loop): Remove code handling loop with unknown niters.
11064 (pass_loop_distribution::execute): Skip loop with unknown niters.
11066 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11069 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11070 function sort_partitions_by_post_order.
11072 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11074 PR tree-optimization/81374
11075 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11076 the max index of basic blocks, rather than number of basic blocks.
11078 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11080 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11082 (arc_legitimate_pic_operand_p): Likewise.
11083 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11085 (arc_needs_pcl_p): Likewise.
11086 (arc_legitimate_pc_offset_p): Likewise.
11087 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11088 function is also used in constrains.md.
11089 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11090 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11091 PLUS. Only return true/false in known cases, otherwise assert.
11092 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11093 is already called in arc_legitimate_constant_p.
11094 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11096 (LEGITIMATE_PIC_OPERAND_P): Use
11097 arc_raw_symbolic_reference_mentioned_p function.
11098 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11103 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11104 Andrew Burgess <andrew.burgess@embecosm.com>
11106 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11107 (arc_return_address_register): New function.
11108 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11109 (arc_handle_fndecl_attribute): Add naked attribute.
11110 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11111 (TARGET_WARN_FUNC_RETURN): Likewise.
11112 (arc_allocate_stack_slots_for_args): New function.
11113 (arc_warn_func_return): Likewise.
11114 (machine_function): Change type fn_type.
11115 (arc_compute_function_type): Consider new naked function type,
11116 change function return type.
11117 (arc_must_save_register): Adapt to handle new
11118 arc_compute_function_type's return type.
11119 (arc_expand_prologue): Likewise.
11120 (arc_expand_epilogue): Likewise.
11121 (arc_return_address_regs): Delete.
11122 (arc_return_address_register): New function.
11123 (arc_epilogue_uses): Use above function.
11124 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11125 (arc_function_type): Change encoding, add naked type.
11126 (ARC_INTERRUPT_P): Change to handle the new encoding.
11127 (ARC_FAST_INTERRUPT_P): Likewise.
11128 (ARC_NORMAL_P): Define.
11129 (ARC_NAKED_P): Likewise.
11130 (arc_compute_function_type): Delete prototype.
11131 * config/arc/arc.md (in_ret_delay_slot): Use
11132 arc_return_address_register function.
11133 (simple_return): Likewise.
11134 (p_return_i): Likewise.
11136 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11138 PR tree-optimization/81428
11139 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11140 can't be built for those types.
11142 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11144 Remove stuff dead since r239246.
11146 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11147 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11148 (avr_inform_devices): Remove dead stuff.
11150 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11152 * config/arm/arm_neon.h: Fix softp typo.
11154 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11156 PR tree-optimization/81365
11157 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11158 aggregate moves onto bb predecessor edges, make sure there are no
11159 loads that could alias the lhs in between the start of bb and the
11162 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11165 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11166 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11167 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11169 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11171 PR tree-optimization/81396
11172 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11173 (init_symbolic_number): Initialize it to 1.
11174 (perform_symbolic_merge): Add n_ops from both operands into the new
11176 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11177 without base_addr as useless if they need more than one operation.
11178 (bswap_replace): Handle !bswap case for NULL base_addr.
11180 2017-07-17 Tom de Vries <tom@codesourcery.com>
11183 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11186 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11188 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11189 conditional builtin define __FIX_LEON3FT_B2BST.
11191 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11193 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11194 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11197 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11199 PR rtl-optimization/81424
11200 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11201 to remove potential trapping from operands if -fnon-call-exceptions.
11203 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11205 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11206 profile_proability for scalling.
11207 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11209 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11211 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11213 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11215 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11216 fixpoint arithmetics.
11218 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11220 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11221 fixpoint arithmetics.
11223 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11225 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11226 fixpoint arithmetics.
11228 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11230 * profile-count.h (profile_probability::from_reg_br_prob_note,
11231 profile_probability::to_reg_br_prob_note): New functions.
11232 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11233 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11234 * predict.c (probability_reliable_p): Update.
11235 (edge_probability_reliable_p): Update.
11236 (br_prob_note_reliable_p): Update.
11237 (invert_br_probabilities): Update.
11238 (add_reg_br_prob_note): New function.
11239 (combine_predictions_for_insn): Update.
11240 * asan.c (asan_clear_shadow): Update.
11241 * cfgbuild.c (compute_outgoing_frequencies): Update.
11242 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11243 (update_br_prob_note): Update.
11244 (rtl_verify_edges): Update.
11245 (purge_dead_edges): Update.
11246 (fixup_reorder_chain): Update.
11247 * emit-rtl.c (try_split): Update.
11248 * ifcvt.c (cond_exec_process_insns): Update.
11249 (cond_exec_process_if_block): Update.
11250 (dead_or_predicable): Update.
11251 * internal-fn.c (expand_addsub_overflow): Update.
11252 (expand_neg_overflow): Update.
11253 (expand_mul_overflow): Update.
11254 * loop-doloop.c (doloop_modify): Update.
11255 * loop-unroll.c (compare_and_jump_seq): Update.
11256 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11257 * predict.h: Update.
11258 * reorg.c (mostly_true_jump): Update.
11260 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11261 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11262 * config/arc/arc.c: (emit_unlikely_jump): Update.
11263 * config/arm/arm.c: (emit_unlikely_jump): Update.
11264 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11265 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11266 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11267 (ix86_print_operand): Update.
11268 (ix86_split_fp_branch): Update.
11269 (predict_jump): Update.
11270 * config/ia64/ia64.c (ia64_print_operand): Update.
11271 * config/mmix/mmix.c (mmix_print_operand): Update.
11272 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11273 (rs6000_expand_split_stack_prologue): Update.
11274 * config/rs6000/rs6000.c: Update.
11275 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11276 (s390_expand_vec_movstr): Update.
11277 (s390_expand_cs_tdsi): Update.
11278 (s390_expand_split_stack_prologue): Update.
11279 * config/sh/sh.c (sh_print_operand): Update.
11280 (expand_cbranchsi4): Update.
11281 (expand_cbranchdi4): Update.
11282 * config/sparc/sparc.c (output_v9branch): Update.
11283 * config/spu/spu.c (get_branch_target): Update.
11284 (ea_load_store_inline): Update.
11285 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11286 * config/tilepro/tilepro.c: Update.
11288 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11290 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11291 (gimplify_save_expr): Likewise.
11293 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11295 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11297 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11299 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11301 * timevar.def (TV_IPA_FNSUMMARY): Define.
11303 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11305 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11306 to back store errata sensitive sequence from being generated.
11307 (sqrtdf2_fix): Likewise.
11309 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11311 * tree-ssa-threadupdate.c (compute_path_counts,
11312 update_joiner_offpath_counts): Use profile_probability.
11314 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11317 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11319 * config/arm/arm-c.c (arm_cpu_builtins): Define
11320 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11322 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11324 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11325 array entries to represent __ieee128 versions of the
11326 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11327 scalar_extract_sig, and scalar_insert_exp built-in functions.
11328 (altivec_resolve_overloaded_builtin): Add special case handling
11329 for the __builtin_scalar_insert_exp function, as represented by
11330 the P9V_BUILTIN_VEC_VSIEDP constant.
11331 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11332 exponent support for __ieee128 argument.
11333 (VSESQP): Add scalar extract signature support for __ieee128
11335 (VSTDCNQP): Add scalar test negative support for __ieee128
11337 (VSIEQP): Add scalar insert exponent support for __int128 argument
11338 with __ieee128 result.
11339 (VSIEQPF): Add scalar insert exponent support for __ieee128
11340 argument with __ieee128 result.
11341 (VSTDCQP): Add scalar test data class support for __ieee128
11343 (VSTDCNQP): Add overload support for scalar test negative with
11344 __ieee128 argument.
11345 (VSTDCQP): Add overload support for scalar test data class
11346 __ieee128 argument.
11347 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11348 UNSPEC_VSX_SXSIGDP.
11349 (UNSPEC_VSX_SIEXPQP): New constant.
11350 (xsxexpqp): New insn for VSX scalar extract exponent quad
11352 (xsxsigqp): New insn for VSX scalar extract significand quad
11354 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11355 precision with floating point argument.
11356 (xststdcqp): New expand for VSX scalar test data class quad
11358 (xststdcnegqp): New expand for VSX scalar test negative quad
11360 (xststdcqp): New insn to match expansions for VSX scalar test data
11361 class quad precision and VSX scalar test negative quad precision.
11362 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11363 special case operand checking to enforce that second operand of
11364 VSX scalar test data class with quad precision argument is a 7-bit
11366 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11367 prototypes and descriptions of __ieee128 versions of
11368 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11369 scalar_test_data_class, and scalar_test_neg built-in functions.
11371 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11373 PR tree-optimization/81162
11374 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11375 replace a negate with an add.
11377 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11379 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11381 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11383 * config/arm/arm-c.c (arm_cpu_builtins): Define
11384 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11386 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11388 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11389 (armv8-r): Set ARM Cortex-R52 as default CPU.
11390 * config/arm/arm-tables.opt: Regenerate.
11391 * config/arm/arm-tune.md: Regenerate.
11392 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11394 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11395 extension for -mcpu=cortex-r52.
11397 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11399 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11400 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11401 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11402 (fp-armv8): Define it as FP_ARMv8 only.
11403 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11404 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11406 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11407 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11408 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11409 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11410 than TARGET_FPU_ARMV8.
11411 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11412 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11413 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11415 * config/arm/neon.md (neon_vrint): Likewise.
11416 (neon_vcvt): Likewise.
11417 (neon_<fmaxmin_op><mode>): Likewise.
11418 (<fmaxmin><mode>3): Likewise.
11419 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11420 * config/arm/predicates.md (arm_cond_move_operator): Check against
11421 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11423 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11425 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11426 to top of function.
11428 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11430 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11431 loop in comment with memset.
11433 2017-07-14 Martin Liska <mliska@suse.cz>
11435 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11436 * dwarf2out.c (is_java): Remove the function.
11437 (output_pubname): Remove usage of the function.
11438 (lower_bound_default): Remove usage of DW_LANG_Java.
11439 (gen_compile_unit_die): Likewise.
11440 * gcc.c: Remove compiler defaults for .java and .zip files.
11441 * gimple-expr.c (remove_suffix): Change as there's no longer
11442 extension than 4-letter one.
11443 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11444 (gimplify_save_expr): Likewise.
11445 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11446 as it's possible even for other languages than Java.
11447 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11448 * lto-opts.c (lto_write_options): Remove reference to Java.
11449 * opts.c (strip_off_ending): Update file extension handling.
11450 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11451 * tree-eh.c (lower_resx): Likewise.
11452 * tree.c (free_lang_data_in_type): Remove dead code.
11453 (find_decls_types_r): Likewise.
11454 (build_common_builtin_nodes): Remove Java from a comment.
11455 (verify_type): Remove dead code.
11456 * varasm.c (assemble_external): Remove Java from a comment.
11458 2017-07-14 Martin Liska <mliska@suse.cz>
11460 * opts.c (finish_options): Add quotes.
11461 (common_handle_option): Likewise.
11463 2017-07-14 Martin Liska <mliska@suse.cz>
11465 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11466 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11467 Remove N_SO_PASCAL.
11468 * dwarf2out.c (lower_bound_default): Do not handle
11470 (gen_compile_unit_die): Likewise.
11471 * gcc.c: Remove default extension binding for GNU Pascal.
11472 * stmt.c: Remove Pascal language from a comment.
11473 * xcoffout.c: Likewise.
11475 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11478 * diagnostic-show-locus.c (fixit_cmp): New function.
11479 (layout::layout): Sort m_fixit_hints.
11480 (column_range::column_range): Assert that the values are valid.
11481 (struct char_span): New struct.
11482 (correction::overwrite): New method.
11483 (struct source_line): New struct.
11484 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11485 calls in terms of classes source_line and char_span, and
11486 correction::overwrite.
11487 (selftest::test_overlapped_fixit_printing_2): New function.
11488 (selftest::diagnostic_show_locus_c_tests): Call it.
11490 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11493 early if there is no lhs.
11495 2017-07-13 Martin Liska <mliska@suse.cz>
11497 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11498 (gen_reference_type_die): Likewise.
11499 * stor-layout.c: Remove Pascal-related comment.
11501 2017-07-13 Martin Liska <mliska@suse.cz>
11503 * opts.c (finish_options): Add quotes to error messages.
11504 (parse_sanitizer_options): Likewise.
11506 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11508 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11510 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11512 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11514 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11516 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11518 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11520 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11523 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11524 provides the hardware capability bits, define the macro
11525 __BUILTIN_CPU_SUPPORTS__.
11526 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11527 if GLIBC does not provide the hardware capability bits. Add a
11528 gcc_unreachable call if the built-in cpu function is neither
11529 __builtin_cpu_is nor __builtin_cpu_supports.
11530 (rs6000_get_function_versions_dispatcher): Change the warning
11531 that an old GLIBC is used which does not export the capability
11532 bits to be an error.
11533 * doc/extend.texi (target_clones attribute): Document the
11534 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11535 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11536 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11537 the macros defined by GCC if the newer GLIBC is available.
11539 2017-07-12 Jeff Law <law@redhat.com>
11541 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11542 remaining includes slightly.
11543 * config/riscv/riscv-builtins.c: Include profile-count.h.
11545 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11548 * config/avr/avr.c (avr_set_current_function): In diagnostic
11549 messages: Quote keywords and (parts of) identifiers.
11550 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11553 2017-07-12 Carl Love <cel@us.ibm.com>
11555 * config/rs6000/rs6000-c.c: Add support for built-in functions
11556 vector bool char vec_revb (vector bool char);
11557 vector bool short vec_revb (vector short char);
11558 vector bool int vec_revb (vector bool int);
11559 vector bool long long vec_revb (vector bool long long);
11560 * doc/extend.texi: Update the built-in documentation file for the
11561 new built-in functions.
11563 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11565 * config/s390/s390.md: Remove movcc splitter.
11567 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11569 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11570 load/store on condition.
11572 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11575 * config/avr/avr.c (avr_encode_section_info)
11576 [progmem && !TREE_READONLY]: Error if progmem object needs
11579 2017-07-11 Michael Collison <michael.collison@arm.com>
11581 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11584 2017-07-11 Carl Love <cel@us.ibm.com>
11586 * config/rs6000/rs6000-c.c: Add support for builtins
11587 vector unsigned int vec_parity_lsbb (vector signed int);
11588 vector unsigned int vec_parity_lsbb (vector unsigned int);
11589 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11590 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11591 vector unsigned long long vec_parity_lsbb (vector signed long long);
11592 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11593 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11594 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11595 * doc/extend.texi: Update the built-in documentation file for the
11596 new built-in functions.
11598 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11600 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11601 (layout::m_primary_loc): New field.
11602 (layout::layout): Initialize new field. Move location filtering
11603 logic from here to...
11604 (layout::maybe_add_location_range): ...this new method. Add
11605 support for filtering to just the lines already specified by other
11607 (layout::will_show_line_p): New method.
11608 (gcc_rich_location::add_location_if_nearby): New method.
11609 (selftest::test_add_location_if_nearby): New test function.
11610 (selftest::diagnostic_show_locus_c_tests): Call it.
11611 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11614 2017-07-11 Tom de Vries <tom@codesourcery.com>
11616 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11617 (bb_first_real_insn): New function.
11618 (nvptx_single): Add extra initialization of broadcasted condition
11621 2017-07-11 Nathan Sidwell <nathan@acm.org>
11623 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11625 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11627 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11628 Binutils doc as TEXI will mess them up.
11629 * doc/invoke.texi (AVR Options): Same here.
11631 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11633 * config/sparc/sparc.opt (mfix-ut700): New option.
11634 (mfix-gr712rc): Likewise.
11635 (sparc_fix_b2bst): New variable.
11636 * doc/invoke.texi (SPARC options): Document them.
11637 (ARM options): Fix warnings.
11638 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11639 instructions to prevent sequences that can trigger the store-store
11640 errata for certain LEON3FT processors.
11641 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11642 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11643 * config/sparc/sparc.md (fix_b2bst): New attribute.
11644 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11646 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11649 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11651 (*rsqrtsf2_sse): Ditto.
11653 (div<mode>3): Macroize insn from divdf3 and divsf3
11654 using MODEF mode iterator.
11656 2017-07-10 Martin Sebor <msebor@redhat.com>
11658 PR tree-optimization/80397
11659 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11660 instead of testing for equality to INTEGER_TYPE.
11662 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11664 * config.gcc: Remove uclibc from arc target spec.
11666 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11668 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11670 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11673 * lto-wrapper.c (remove_option): New function.
11674 (merge_and_complain): Merge PIC/PIE options more realistically.
11676 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11678 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11682 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11683 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11684 * config.in: Regenerate.
11685 * configure: Regenerate.
11686 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11687 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11688 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11689 (TARGET_GASISR_PROLOGUES): ...target mask.
11690 * common/config/avr/avr-common.c
11691 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11692 Set -mgas-isr-prologues.
11693 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11694 INSERT_PASS_BEFORE for it.
11695 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11696 * config/avr/avr.c (avr_option_override)
11697 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11698 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11699 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11700 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11701 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11703 (make_avr_pass_pre_proep): New function.
11704 (emit_push_sfr) <treg>: Add argument to function and use it
11705 instead of TMP_REG.
11706 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11707 and set machine->gasisr.yes.
11708 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11709 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11710 __gcc_isr.n_pushed to .L__stack_usage.
11711 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11712 (avr_asm_final_postscan_insn): ...this new static function.
11713 * config/avr/avr.h (machine_function)
11714 <is_no_gccisr, use_L__stack_usage>: New fields.
11715 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11716 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11717 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11718 (gasisr, *gasisr): New expander and insn.
11719 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11720 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11721 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11723 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11725 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11728 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11730 Move jump-tables out of .text again.
11733 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11734 (ASM_OUTPUT_ADDR_VEC): New function.
11735 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11736 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11737 INSN_ADDRESSes as asm comment.
11738 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11739 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11740 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11741 * config/avr/avr.md (*tablejump): Adjust comment.
11742 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11743 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11745 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11746 (avr_output_addr_vec): New proto.
11747 (avr_log_t) <insn_addresses>: New field.
11749 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11752 * config/i386/i386.c (ix86_function_arg_advance): Set
11753 outgoing_args_on_stack to true if there are outgoing arguments
11755 (ix86_function_arg): Likewise.
11756 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11757 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11758 * config/i386/i386.h (machine_function): Add
11759 outgoing_args_on_stack.
11761 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11763 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11764 supporting pthreds.
11765 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11767 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11769 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11770 (REAL_H): Remove $(MACHMODE_H).
11771 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11773 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11774 $(MACHMODE_H) and double-int.h.
11775 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11777 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11780 2017-07-07 Andrew Pinski <apinski@cavium.com>
11782 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11783 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11785 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11787 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11788 Add warning if GCC was not configured to link against a GLIBC that
11789 exports the hardware capability bits.
11790 (make_resolver_func): Make resolver function private and not a
11791 COMDAT function. Create the name with clone_function_name instead
11792 of make_unique_name.
11795 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11796 correct operand in doing the split.
11798 2017-07-07 Carl Love <cel@us.ibm.com>
11800 * config/rs6000/rs6000-c: Add support for built-in function
11801 vector unsigned short vec_pack_to_short_fp32 (vector float,
11803 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11804 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11805 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11806 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11807 (convert_4f32_8i16): Add define_expand.
11808 * doc/extend.texi: Update the built-in documentation file for the
11809 new built-in function.
11811 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11813 * config/sparc/m8.md: New file.
11814 * config/sparc/sparc.md: Include m8.md.
11816 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11818 * config/sparc/sparc.opt: New option -mvis4b.
11819 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11820 (sparc_option_override): Handle VIS4B.
11821 (enum sparc_builtins): Define
11822 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11823 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11824 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11825 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11826 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11827 (check_constant_argument): New function.
11828 (sparc_vis_init_builtins): Define builtins
11829 __builtin_vis_dictunpack{8,16,32},
11830 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11831 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11832 __builtin_vis_fpcmpde{8,16,32}shl and
11833 __builtin_vis_fpcmpur{8,16,32}shl.
11834 (sparc_expand_builtin): Check that the constant operands to
11835 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11836 constant and in range.
11837 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11839 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11840 (SPARC_IMM5_P): Likewise.
11841 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
11842 (enabled): Handle vis4b.
11843 (UNSPEC_DICTUNPACK): New unspec.
11844 (UNSPEC_FPCMPSHL): Likewise.
11845 (UNSPEC_FPUCMPSHL): Likewise.
11846 (UNSPEC_FPCMPDESHL): Likewise.
11847 (UNSPEC_FPCMPURSHL): Likewise.
11848 (cpu_feature): New CPU feature `vis4b'.
11849 (dictunpack{8,16,32}): New insns.
11850 (FPCSMODE): New mode iterator.
11851 (fpcscond): New code iterator.
11852 (fpcsucond): Likewise.
11853 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
11854 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
11855 (fpcmpde{8,16,32}{si,di}shl): Likewise.
11856 (fpcmpur{8,16,32}{si,di}shl): Likewise.
11857 * config/sparc/constraints.md: Define constraints `q' for unsigned
11858 2-bit integer constants and `t' for unsigned 5-bit integer
11860 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
11862 (imm5_operand_dictunpack16): Likewise.
11863 (imm5_operand_dictunpack32): Likewise.
11864 (imm2_operand): Likewise.
11865 * doc/invoke.texi (SPARC Options): Document -mvis4b.
11866 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
11867 ditunpack* and fpcmp*shl builtins.
11869 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11871 * config.gcc: Handle m8 in --with-{cpu,tune} options.
11872 * config.in: Add HAVE_AS_SPARC6 define.
11873 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
11875 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
11877 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
11878 (CPP_CPU_SPEC): Handle m8.
11879 (ASM_CPU_SPEC): Likewise.
11880 * config/sparc/sparc-opts.h (enum processor_type): Add
11882 * config/sparc/sparc.c (m8_costs): New struct.
11883 (sparc_option_override): Handle TARGET_CPU_m8.
11884 (sparc32_initialize_trampoline): Likewise.
11885 (sparc64_initialize_trampoline): Likewise.
11886 (sparc_issue_rate): Likewise.
11887 (sparc_register_move_cost): Likewise.
11888 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
11889 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
11890 (ASM_CPU64_DEFAULT_SPEC): Likewise.
11891 (CPP_CPU_SPEC): Handle M8.
11892 (ASM_CPU_SPEC): Likewise.
11893 (AS_M8_FLAG): Define.
11894 * config/sparc/sparc.md: Add m8 to the cpu attribute.
11895 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
11896 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
11898 * configure: Regenerate.
11899 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
11902 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11904 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
11906 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
11907 ("*movdi_insn_sp32"): Do not set v3pipe.
11908 ("*movsi_insn"): Likewise.
11909 ("*movdi_insn_sp64"): Likewise.
11910 ("*movsf_insn"): Likewise.
11911 ("*movdf_insn_sp32"): Likewise.
11912 ("*movdf_insn_sp64"): Likewise.
11913 ("*zero_extendsidi2_insn_sp64"): Likewise.
11914 ("*sign_extendsidi2_insn"): Likewise.
11915 ("*mov<VM32:mode>_insn"): Likewise.
11916 ("*mov<VM64:mode>_insn_sp64"): Likewise.
11917 ("*mov<VM64:mode>_insn_sp32"): Likewise.
11918 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11919 ("<vlop:code><VL:mode>3"): Likewise.
11920 ("*not_<vlop:code><VL:mode>3"): Likewise.
11921 ("*nand<VL:mode>_vis"): Likewise.
11922 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
11923 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
11924 ("one_cmpl<VL:mode>2"): Likewise.
11925 ("faligndata<VM64:mode>_vis"): Likewise.
11926 ("alignaddrsi_vis"): Likewise.
11927 ("alignaddrdi_vis"): Likweise.
11928 ("alignaddrlsi_vis"): Likewise.
11929 ("alignaddrldi_vis"): Likewise.
11930 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
11931 ("bmaskdi_vis"): Likewise.
11932 ("bmasksi_vis"): Likewise.
11933 ("bshuffle<VM64:mode>_vis"): Likewise.
11934 ("cmask8<P:mode>_vis"): Likewise.
11935 ("cmask16<P:mode>_vis"): Likewise.
11936 ("cmask32<P:mode>_vis"): Likewise.
11937 ("pdistn<P:mode>_vis"): Likewise.
11938 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11940 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11942 * config/sparc/sparc.md ("subtype"): New insn attribute.
11943 ("*wrgsr_sp64"): Set insn subtype.
11944 ("*rdgsr_sp64"): Likewise.
11945 ("alignaddrsi_vis"): Likewise.
11946 ("alignaddrdi_vis"): Likewise.
11947 ("alignaddrlsi_vis"): Likewise.
11948 ("alignaddrldi_vis"): Likewise.
11949 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
11950 ("fexpand_vis"): Likewise.
11951 ("fpmerge_vis"): Likewise.
11952 ("faligndata<VM64:mode>_vis"): Likewise.
11953 ("bshuffle<VM64:mode>_vis"): Likewise.
11954 ("cmask8<P:mode>_vis"): Likewise.
11955 ("cmask16<P:mode>_vis"): Likewise.
11956 ("cmask32<P:mode>_vis"): Likewise.
11957 ("fchksm16_vis"): Likewise.
11958 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
11959 ("fmean16_vis"): Likewise.
11960 ("fp<plusminus_insn>64_vis"): Likewise.
11961 ("<plusminus_insn>v8qi3"): Likewise.
11962 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
11963 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
11964 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
11965 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
11966 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
11967 ("*movqi_insn"): Likewise.
11968 ("*movhi_insn"): Likewise.
11969 ("*movsi_insn"): Likewise.
11970 ("movsi_pic_gotdata_op"): Likewise.
11971 ("*movdi_insn_sp32"): Likewise.
11972 ("*movdi_insn_sp64"): Likewise.
11973 ("movdi_pic_gotdata_op"): Likewise.
11974 ("*movsf_insn"): Likewise.
11975 ("*movdf_insn_sp32"): Likewise.
11976 ("*movdf_insn_sp64"): Likewise.
11977 ("*zero_extendhisi2_insn"): Likewise.
11978 ("*zero_extendqihi2_insn"): Likewise.
11979 ("*zero_extendqisi2_insn"): Likewise.
11980 ("*zero_extendqidi2_insn"): Likewise.
11981 ("*zero_extendhidi2_insn"): Likewise.
11982 ("*zero_extendsidi2_insn_sp64"): Likewise.
11983 ("ldfsr"): Likewise.
11984 ("prefetch_64"): Likewise.
11985 ("prefetch_32"): Likewise.
11986 ("tie_ld32"): Likewise.
11987 ("tie_ld64"): Likewise.
11988 ("*tldo_ldub_sp32"): Likewise.
11989 ("*tldo_ldub1_sp32"): Likewise.
11990 ("*tldo_ldub2_sp32"): Likewise.
11991 ("*tldo_ldub_sp64"): Likewise.
11992 ("*tldo_ldub1_sp64"): Likewise.
11993 ("*tldo_ldub2_sp64"): Likewise.
11994 ("*tldo_ldub3_sp64"): Likewise.
11995 ("*tldo_lduh_sp32"): Likewise.
11996 ("*tldo_lduh1_sp32"): Likewise.
11997 ("*tldo_lduh_sp64"): Likewise.
11998 ("*tldo_lduh1_sp64"): Likewise.
11999 ("*tldo_lduh2_sp64"): Likewise.
12000 ("*tldo_lduw_sp32"): Likewise.
12001 ("*tldo_lduw_sp64"): Likewise.
12002 ("*tldo_lduw1_sp64"): Likewise.
12003 ("*tldo_ldx_sp64"): Likewise.
12004 ("*mov<VM32:mode>_insn"): Likewise.
12005 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12006 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12008 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12010 * config/sparc/sparc.md ("type"): New insn type viscmp.
12011 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12013 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12014 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12015 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12016 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12018 ("n7_vis_logical_11cycle"): Likewise.
12019 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12020 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12021 * config/sparc/niagara.md ("niag_vis"): Likewise.
12022 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12023 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12025 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12027 * config/sparc/sparc.md: New instruction type `bmask'.
12028 (bmaskdi_vis): Use the `bmask' type.
12029 (bmasksi_vis): Likewise.
12030 * config/sparc/ultra3.md (us3_array): Likewise.
12031 * config/sparc/niagara7.md (n7_array): Likewise.
12032 * config/sparc/niagara4.md (n4_array): Likewise.
12033 * config/sparc/niagara2.md (niag2_vis): Likewise.
12034 (niag3_vis): Likewise.
12035 * config/sparc/niagara.md (niag_vis): Likewise.
12037 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12039 * ipa-comdats.c: Remove optimize check from gate.
12040 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12041 for functions not optimized.
12042 (ipa_fn_summary_read): Skip optimize check.
12043 (ipa_fn_summary_write): Likewise.
12044 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12046 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12048 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12049 (check_callers): Give up on caller which is not optimized.
12050 (inline_small_functions): Likewise.
12051 (ipa_inline): Do not give up when not optimizing.
12052 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12053 away unoptimizes cdtors.
12054 (whole_program_function_and_variable_visibility): Do
12055 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12056 * ipa.c (process_references): Do not check optimize.
12057 (symbol_table::remove_unreachable_nodes): Update optimize check.
12058 (set_writeonly_bit): Update optimize check.
12059 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12060 (pass_ipa_single_use::gate): Remove.
12062 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12064 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12065 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12066 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12067 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12068 permute_load, permute_store, adjust_extract, adjust_splat,
12069 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12070 replace_swap_with_copy, dump_swap_insn_table,
12071 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12072 recombine_lvx_pattern, recombine_stvx_pattern,
12073 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12074 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12075 to file rs6000-p8swap.c.
12076 * config/rs6000/rs6000-p8swap.c: New file.
12077 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12078 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12079 and rs6000*-*-* targets.
12081 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12083 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12085 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12087 * lto-wrapper.c (merge_and_complain): Do not merge
12088 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12089 fsigned_zeros, ftrapping_math, fwrapv.
12090 (append_compiler_options): Do not track these options.
12091 (append_linker_options): Likewie
12093 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12095 * cgraphunit.c (cgraph_node::finalize_function): When
12096 !flag_toplevel_reorde set no_reorder flag.
12097 (varpool_node::finalize_decl): Likewise.
12098 (symbol_table::compile): Drop no toplevel reorder path.
12100 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12102 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12103 edges; zero probability is not better than uninitialized.
12105 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12107 * asan.h (asan_sanitize_allocas_p): Declare.
12108 * asan.c (asan_sanitize_allocas_p): New function.
12109 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12110 (handle_builtin_alloca): Likewise.
12111 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12112 if !asan_sanitize_allocas_p.
12113 * params.def (asan-instrument-allocas): Add new option.
12114 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12115 * opts.c (common_handle_option): Disable allocas sanitization for
12118 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12120 * asan.c: Include gimple-fold.h.
12121 (get_last_alloca_addr): New function.
12122 (handle_builtin_stackrestore): Likewise.
12123 (handle_builtin_alloca): Likewise.
12124 (asan_emit_allocas_unpoison): Likewise.
12125 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12126 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12127 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12128 (instrument_builtin_call): Pass gimple iterator to
12129 get_mem_refs_of_builtin_call.
12130 (last_alloca_addr): New global.
12131 * asan.h (asan_emit_allocas_unpoison): Declare.
12132 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12133 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12134 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12135 if function calls alloca.
12136 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12137 * gimple-fold.h (replace_call_with_value): Declare.
12138 * internal-fn.c: Include asan.h.
12139 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12140 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12142 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12144 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12145 (C_SELFTEST_FLAGS): New.
12146 (CPP_SELFTEST_FLAGS): New.
12147 (SELFTEST_DEPS): New, from deps of s-selftest.
12148 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12149 (CPP_SELFTEST_DEPS): New.
12150 (selftest): Add dependency on s-selftest-c++.
12151 (s-selftest): Rename to...
12152 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12153 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12154 than SELFTEST_FLAGS.
12155 (selftest-gdb): Rename to...
12156 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12158 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12159 (selftest-valgrind): Rename to...
12160 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12162 (selftest-valgrind): Reintroduce as an alias for
12163 selftest-c-valgrind.
12164 (s-selftest-c++): New.
12165 (selftest-c++-gdb): New.
12166 (selftest-c++-valgrind): New.
12168 2017-07-06 Olivier Hainque <hainque@adacore.com>
12170 * gcc.c (process_command): When deciding if undefined variables
12171 should be ignored when processing specs, accept "gcc -v" as well.
12173 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12175 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12176 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12178 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12180 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12181 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12182 * config/arm/arm-tables.opt: Regenerate.
12183 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12185 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12187 2017-07-06 Carl Love <cel@us.ibm.com>
12189 * ChangeLog: Clean up from mid air collision
12191 2017-07-06 Carl Love <cel@us.ibm.com>
12193 * config/rs6000/rs6000-c.c: Add support for built-in functions
12194 vector signed int vec_subc (vector signed int, vector signed int);
12195 vector signed __int128 vec_subc (vector signed __int128,
12196 vector signed __int128);
12197 vector unsigned __int128 vec_subc (vector unsigned __int128,
12198 vector unsigned __int128);
12199 vector signed int vec_sube (vector signed int, vector signed int,
12200 vector signed int);
12201 vector unsigned int vec_sube (vector unsigned int,
12202 vector unsigned int,
12203 vector unsigned int);
12204 vector signed __int128 vec_sube (vector signed __int128,
12205 vector signed __int128,
12206 vector signed__int128);
12207 vector unsigned __int128 vec_sube (vector unsigned __int128,
12208 vector unsigned __int128,
12209 vector unsigned __int128);
12210 vector signed int vec_subec (vector signed int, vector signed int,
12211 vector signed int);
12212 vector unsigned int vec_subec (vector unsigned int,
12213 vector unsigned int,
12214 vector unsigned int);
12215 vector signed __int128 vec_subec (vector signed __int128,
12216 vector signed __int128,
12217 vector signed__int128);
12218 vector unsigned __int128 vec_subec (vector unsigned __int128,
12219 vector unsigned __int128,
12220 vector unsigned __int128);
12221 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12222 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12223 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12224 BU_ALTIVEC_OVERLOAD_X definitions.
12225 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12226 * doc/extend.texi: Update the built-in documentation file for the new
12227 built-in functions.
12229 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12232 * diagnostic-show-locus.c (layout::layout): Use start and finish
12233 spelling location for the start and finish of each range.
12234 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12235 Add unused aspect param.
12236 * input.c (expand_location_1): Add "aspect" param, and use it
12237 to access the correct part of the location.
12238 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12240 (expand_location_to_spelling_point): Likewise.
12241 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12242 param, and pass it to expand_location_1.
12244 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12246 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12247 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12248 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12249 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12250 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12251 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12252 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12253 _mm_maskz_getmant_ss): New intrinsics.
12254 (__builtin_ia32_getexpss128_mask): Changed to ...
12255 __builtin_ia32_getexpss128_round ... this.
12256 (__builtin_ia32_getexpsd128_mask): Changed to ...
12257 __builtin_ia32_getexpsd128_round ... this.
12258 * config/i386/i386-builtin-types.def
12259 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12260 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12261 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12262 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12263 __builtin_ia32_getmantss_mask_round): New builtins.
12264 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12265 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12266 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12267 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12268 * config/i386/sse.md
12269 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12270 avx512f_sgetexp<mode><mask_scalar_name>
12271 <round_saeonly_scalar_name> ... this.
12272 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12273 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12274 vgetexp<ssescalarmodesuffix>
12275 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12276 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12277 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12278 avx512f_vgetmant<mode><mask_scalar_name>
12279 <round_saeonly_scalar_name> ... this.
12280 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12281 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12282 vgetmant<ssescalarmodesuffix>
12283 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12284 %0<mask_scalar_operand4>, %1, %2
12285 <round_saeonly_scalar_mask_op4>, %3} ... this.
12286 * config/i386/subst.md (mask_scalar_operand4,
12287 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12288 round_saeonly_scalar_nimm_predicate): New subst attributes.
12290 2017-07-06 Julia Koval <julia.koval@intel.com>
12292 * config/i386/i386.c (ix86_erase_embedded_rounding):
12293 Remove code for old rounding pattern.
12295 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12297 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12299 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12301 * doc/sourcebuild.texi (Test Directives, Variants of
12302 dg-require-support): Add documentation for dg-require-stack-check.
12304 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12306 * config/i386/subst.md (mask_scalar, round_scalar,
12307 round_saeonly_scalar): New meta-templates.
12308 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12309 round_scalar_mask_operand3, round_scalar_mask_op3,
12310 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12311 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12312 round_saeonly_scalar_constraint,
12313 round_saeonly_scalar_prefix): New subst attribute.
12314 * config/i386/sse.md
12315 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12316 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12317 <round_scalar_name> ... this.
12318 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12319 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12320 <round_scalar_name> ... this.
12321 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12322 <sse>_vm<code><mode>3<mask_scalar_name>
12323 <round_saeonly_scalar_name> ... this.
12324 (v<plusminus_mnemonic><ssescalarmodesuffix>
12325 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12326 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12327 v<plusminus_mnemonic><ssescalarmodesuffix>
12328 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12329 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12330 (v<multdiv_mnemonic><ssescalarmodesuffix>
12331 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12332 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12333 v<multdiv_mnemonic><ssescalarmodesuffix>
12334 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12335 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12336 (v<maxmin_float><ssescalarmodesuffix>
12337 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12338 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12339 v<maxmin_float><ssescalarmodesuffix>
12340 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12341 %0<mask_scalar_operand3>, %1, %<iptr>2
12342 <round_saeonly_scalar_mask_op3>} ... this.
12344 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12346 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12347 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12349 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12350 Alan Hayward <alan.hayward@arm.com>
12351 David Sherwood <david.sherwood@arm.com>
12353 * combine.c (simplify_if_then_else): Remove "enum" before
12355 * compare-elim.c (can_eliminate_compare): Likewise.
12356 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12358 (aarch64_lookup_simd_builtin_type): Likewise.
12359 (aarch64_simd_builtin_type): Likewise.
12360 (aarch64_init_simd_builtin_types): Likewise.
12361 (aarch64_simd_expand_args): Likewise.
12362 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12364 (aarch64_reverse_mask): Likewise.
12365 (aarch64_simd_emit_reg_reg_move): Likewise.
12366 (aarch64_gen_adjusted_ldpstp): Likewise.
12367 (aarch64_ccmp_mode_to_code): Likewise.
12368 (aarch64_operands_ok_for_ldpstp): Likewise.
12369 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12370 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12372 (aarch64_min_divisions_for_recip_mul): Likewise.
12373 (aarch64_reassociation_width): Likewise.
12374 (aarch64_get_condition_code_1): Likewise.
12375 (aarch64_simd_emit_reg_reg_move): Likewise.
12376 (aarch64_simd_attr_length_rglist): Likewise.
12377 (aarch64_reverse_mask): Likewise.
12378 (aarch64_operands_ok_for_ldpstp): Likewise.
12379 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12380 (aarch64_gen_adjusted_ldpstp): Likewise.
12381 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12383 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12384 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12385 (arm_lookup_simd_builtin_type): Likewise.
12386 (arm_simd_builtin_type): Likewise.
12387 (arm_init_simd_builtin_types): Likewise.
12388 (arm_expand_builtin_args): Likewise.
12389 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12390 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12391 (ft32_setup_incoming_varargs): Likewise.
12392 (ft32_function_arg): Likewise.
12393 (ft32_function_arg_advance): Likewise.
12394 (ft32_pass_by_reference): Likewise.
12395 (ft32_arg_partial_bytes): Likewise.
12396 (ft32_valid_pointer_mode): Likewise.
12397 (ft32_addr_space_pointer_mode): Likewise.
12398 (ft32_addr_space_legitimate_address_p): Likewise.
12399 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12401 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12402 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12403 (iamcu_alignment): Likewise.
12404 (canonicalize_vector_int_perm): Likewise.
12405 (ix86_noce_conversion_profitable_p): Likewise.
12406 (ix86_mpx_bound_mode): Likewise.
12407 (ix86_operands_ok_for_move_multiple): Likewise.
12408 * config/microblaze/microblaze-protos.h
12409 (microblaze_expand_conditional_branch_reg): Likewise.
12410 * config/microblaze/microblaze.c
12411 (microblaze_expand_conditional_branch_reg): Likewise.
12412 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12414 (rs6000_reassociation_width): Likewise.
12415 (rs6000_invalid_binary_op): Likewise.
12416 (fusion_p9_p): Likewise.
12417 (emit_fusion_p9_load): Likewise.
12418 (emit_fusion_p9_store): Likewise.
12419 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12421 (riscv_hard_regno_mode_ok_p): Likewise.
12422 (riscv_address_insns): Likewise.
12423 (riscv_split_symbol): Likewise.
12424 (riscv_legitimize_move): Likewise.
12425 (riscv_function_value): Likewise.
12426 (riscv_hard_regno_nregs): Likewise.
12427 (riscv_expand_builtin): Likewise.
12428 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12429 (riscv_build_integer): Likewise.
12430 (riscv_split_integer): Likewise.
12431 (riscv_legitimate_constant_p): Likewise.
12432 (riscv_cannot_force_const_mem): Likewise.
12433 (riscv_regno_mode_ok_for_base_p): Likewise.
12434 (riscv_valid_base_register_p): Likewise.
12435 (riscv_valid_offset_p): Likewise.
12436 (riscv_valid_lo_sum_p): Likewise.
12437 (riscv_classify_address): Likewise.
12438 (riscv_legitimate_address_p): Likewise.
12439 (riscv_address_insns): Likewise.
12440 (riscv_load_store_insns): Likewise.
12441 (riscv_force_binary): Likewise.
12442 (riscv_split_symbol): Likewise.
12443 (riscv_force_address): Likewise.
12444 (riscv_legitimize_address): Likewise.
12445 (riscv_move_integer): Likewise.
12446 (riscv_legitimize_const_move): Likewise.
12447 (riscv_legitimize_move): Likewise.
12448 (riscv_address_cost): Likewise.
12449 (riscv_subword): Likewise.
12450 (riscv_output_move): Likewise.
12451 (riscv_canonicalize_int_order_test): Likewise.
12452 (riscv_emit_int_order_test): Likewise.
12453 (riscv_function_arg_boundary): Likewise.
12454 (riscv_pass_mode_in_fpr_p): Likewise.
12455 (riscv_pass_fpr_single): Likewise.
12456 (riscv_pass_fpr_pair): Likewise.
12457 (riscv_get_arg_info): Likewise.
12458 (riscv_function_arg): Likewise.
12459 (riscv_function_arg_advance): Likewise.
12460 (riscv_arg_partial_bytes): Likewise.
12461 (riscv_function_value): Likewise.
12462 (riscv_pass_by_reference): Likewise.
12463 (riscv_setup_incoming_varargs): Likewise.
12464 (riscv_print_operand): Likewise.
12465 (riscv_elf_select_rtx_section): Likewise.
12466 (riscv_save_restore_reg): Likewise.
12467 (riscv_for_each_saved_reg): Likewise.
12468 (riscv_register_move_cost): Likewise.
12469 (riscv_hard_regno_mode_ok_p): Likewise.
12470 (riscv_hard_regno_nregs): Likewise.
12471 (riscv_class_max_nregs): Likewise.
12472 (riscv_memory_move_cost): Likewise.
12473 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12474 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12475 (rl78_addr_space_address_mode): Likewise.
12476 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12478 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12479 (rs6000_reassociation_width): Likewise.
12480 (rs6000_invalid_binary_op): Likewise.
12481 (fusion_p9_p): Likewise.
12482 (emit_fusion_p9_load): Likewise.
12483 (emit_fusion_p9_store): Likewise.
12484 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12485 (ok_for_simple_move_operands): Likewise.
12486 (ok_for_simple_move_strict_operands): Likewise.
12487 (ok_for_simple_arith_logic_operands): Likewise.
12488 (visium_legitimize_reload_address): Likewise.
12489 (visium_select_cc_mode): Likewise.
12490 (output_cbranch): Likewise.
12491 (visium_split_double_move): Likewise.
12492 (visium_expand_copysign): Likewise.
12493 (visium_expand_int_cstore): Likewise.
12494 (visium_expand_fp_cstore): Likewise.
12495 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12496 (visium_function_arg): Likewise.
12497 (visium_function_arg_advance): Likewise.
12498 (visium_libcall_value): Likewise.
12499 (visium_setup_incoming_varargs): Likewise.
12500 (visium_legitimate_constant_p): Likewise.
12501 (visium_legitimate_address_p): Likewise.
12502 (visium_legitimize_address): Likewise.
12503 (visium_secondary_reload): Likewise.
12504 (visium_register_move_cost): Likewise.
12505 (visium_memory_move_cost): Likewise.
12506 (prepare_move_operands): Likewise.
12507 (ok_for_simple_move_operands): Likewise.
12508 (ok_for_simple_move_strict_operands): Likewise.
12509 (ok_for_simple_arith_logic_operands): Likewise.
12510 (visium_function_value_1): Likewise.
12511 (rtx_ok_for_offset_p): Likewise.
12512 (visium_legitimize_reload_address): Likewise.
12513 (visium_split_double_move): Likewise.
12514 (visium_expand_copysign): Likewise.
12515 (visium_expand_int_cstore): Likewise.
12516 (visium_expand_fp_cstore): Likewise.
12517 (visium_split_cstore): Likewise.
12518 (visium_select_cc_mode): Likewise.
12519 (visium_split_cbranch): Likewise.
12520 (output_cbranch): Likewise.
12521 (visium_print_operand_address): Likewise.
12522 * expmed.c (flip_storage_order): Likewise.
12523 * expmed.h (emit_cstore): Likewise.
12524 (flip_storage_order): Likewise.
12525 * genrecog.c (validate_pattern): Likewise.
12526 * hsa-gen.c (gen_hsa_addr): Likewise.
12527 * internal-fn.c (expand_arith_overflow): Likewise.
12528 * ira-color.c (allocno_copy_cost_saving): Likewise.
12529 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12530 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12531 (process_invariant_for_inheritance): Likewise.
12532 * lra-eliminations.c (move_plus_up): Likewise.
12533 * omp-low.c (lower_oacc_reductions): Likewise.
12534 * simplify-rtx.c (simplify_subreg): Likewise.
12535 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12536 (TARGET_CHKP_BOUND_MODE): Likewise..
12537 * targhooks.c (default_chkp_bound_mode): Likewise.
12538 (default_setup_incoming_vararg_bounds): Likewise.
12539 * targhooks.h (default_chkp_bound_mode): Likewise.
12540 (default_setup_incoming_vararg_bounds): Likewise.
12541 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12542 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12543 (have_whole_vector_shift): Likewise.
12544 * tree-vect-stmts.c (vectorizable_load): Likewise.
12545 * doc/tm.texi: Regenerate.
12547 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12549 Graceful degrade if Binutils PR21472 is not available.
12552 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12553 .rodata in flash test fails.
12554 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12555 * confgure: Regenerate.
12556 * config.in: Regenerate.
12557 * config/avr/avr.c (avr_asm_named_section)
12558 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12559 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12560 (avr_asm_init_sections): Same.
12562 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12564 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12565 (fma<VH:mode>4_intrinsic): Likewise.
12566 (*fmsub<VCVTF:mode>4): Likewise.
12567 (*fmsub<VH:mode>4_intrinsic): Likewise.
12569 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12572 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12573 Don't depend on "optimize > 0".
12574 (out_movhi_r_mr, out_movqi_mr_r): Same.
12575 (out_movhi_mr_r, out_movqi_r_mr): Same.
12576 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12577 io_address_operand on "optimize > 0".
12579 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12581 * tree-loop-distribution.c: Add general explanantion on the pass.
12582 (generate_loops_for_partition): Mark distributed loop.
12583 (pg_add_dependence_edges): New parameter. Handle alias data
12584 dependence specially and record it in the parameter if asked.
12585 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12586 (init_partition_graph_vertices, add_partition_graph_edge): New.
12587 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12588 (free_partition_graph_vdata, build_partition_graph): New.
12589 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12590 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12591 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12592 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12593 (version_for_distribution_p, finalize_partitions): New.
12594 (distribute_loop): Handle alias data dependence specially. Factor
12595 out loop fusion code as functions and call these functions.
12597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12599 * tree-loop-distribution.c (classify_partition): New parameter and
12600 better handle reduction statement.
12601 (rdg_build_partitions): Revise comment.
12602 (distribute_loop): Compute statements in all partitions and pass it
12603 to classify_partition.
12605 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12607 * tree-loop-distribution.c (enum partition_type): New.
12608 (struct partition): New field type.
12609 (partition_merge_into): Add parameter. Update partition type.
12610 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12611 (build_rdg_partition_for_vertex): Compute partition type.
12612 (rdg_build_partitions): Dump partition type.
12613 (distribute_loop): Update calls to partition_merge_into.
12615 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12617 * tree-loop-distribution.c (struct ddr_hasher): New.
12618 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12620 (classify_partition): Call get_data_dependence.
12621 (pg_add_dependence_edges): Ditto.
12622 (distribute_loop): Release data dependence hash table.
12624 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12626 * tree-loop-distribution.c (ref_base_address): Delete.
12627 (similar_memory_accesses): Rename ...
12628 (share_memory_accesses): ... to this. Check if partitions access
12629 the same memory reference.
12630 (distribute_loop): Call share_memory_accesses.
12632 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12634 * tree-loop-distribution.c (struct partition): New field recording
12635 its data reference.
12636 (partition_alloc, partition_free): Init and release data refs.
12637 (partition_merge_into): Merge data refs.
12638 (build_rdg_partition_for_vertex): Collect data refs for partition.
12639 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12641 (distribute_loop): Remve data refs from vertice data of partition
12644 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12646 * tree-loop-distribution.c (params.h): Include header file.
12647 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12648 (datarefs_vec): New global var.
12649 (create_rdg_vertices): Use datarefs_vec directly.
12650 (free_rdg): Don't free data references.
12651 (build_rdg): Update use. Don't free data references.
12652 (distribute_loop): Compute global variable for data references.
12653 Bail out if there are too many data references.
12655 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12657 * tree-loop-distribution.c (loop_nest): New global var.
12658 (build_rdg): Use loop directly, rather than loop nest.
12659 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12661 (distribute_loop): Compute global variable loop nest. Update use.
12663 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12665 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12666 (partition_merge_into): New parameter. Dump reason for fusion.
12667 (distribute_loop): Update use of partition_merge_into.
12669 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12671 * tree-loop-distribution.c (bb_top_order_index): New.
12672 (bb_top_order_index_size, bb_top_order_cmp): New.
12673 (stmts_from_loop): Use topological order.
12674 (pass_loop_distribution::execute): Compute and release topological
12675 order for basic blocks.
12677 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12679 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12684 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12685 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12686 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12687 * internal-fn.def (LOOP_DIST_ALIAS): New.
12688 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12689 (fold_loop_internal_call): ... this.
12690 (vect_loop_dist_alias_call): New function.
12691 (set_uid_loop_bbs): Call fold_loop_internal_call.
12692 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12695 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12698 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12699 Require dead FLAGS_REG at the beginning of a peephole.
12701 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12704 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12705 arguments in the call to __builtin_ia32_sbb_u32.
12706 (_subborrow_u64): Swap _X and _Y arguments in the call to
12707 __builtin_ia32_sbb_u64.
12709 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12712 * tree-vrp.c (compare_assert_loc): Turn into a function template
12713 with stable template parameter. Only test if a->e is NULL,
12714 !a->e == !b->e has been verified already. Use e == NULL or
12715 e != NULL instead of e or ! e tests. If stable is true, don't use
12716 iterative_hash_expr, on the other side allow a or b or both NULL
12717 and sort the NULLs last.
12718 (process_assert_insertions): Sort using compare_assert_loc<false>
12719 instead of compare_assert_loc, later sort using
12720 compare_assert_loc<true> before calling process_assert_insertions_for
12721 in a loop. Use break instead of continue once seen NULL pointer.
12723 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12725 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12726 Cortex-R7 and Cortex-R8 processors.
12728 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12730 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12731 uninitialized while src is not.
12733 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12735 * common/config/arm/arm-common.c: Adjust include path for
12737 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12738 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12739 (arm-cpu-data.h): Likewise.
12740 (arm-cpu-cdata.h): Likewise.
12741 * config/arm/arm-cpu.h: Delete.
12742 * config/arm/arm-cpu-cdata.h: Delete.
12743 * config/arm/arm-cpu-data.h: Delete.
12745 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12747 * config/arm/arm-cpus.in (cortex-a55): New.
12748 (cortex-a75): Likewise.
12749 (cortex-a75.cortex-a55): Likewise.
12750 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12752 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12753 * config/arm/arm-cpu-cdata.h: Regenerate.
12754 * config/arm/arm-cpu-data.h: Regenerate.
12755 * config/arm/arm-cpu.h: Regenerate.
12756 * config/arm/arm-tables.opt: Regenerate.
12757 * config/arm/arm-tune.md: Regenerate.
12759 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12761 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12763 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12765 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12768 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12770 PR tree-optimization/81292
12771 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12772 full_string_p, also call adjust_related_strinfos if the adjustment
12773 is simple, otherwise invalidate related strinfos.
12775 2017-07-04 Martin Liska <mliska@suse.cz>
12778 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12779 newly created variable as DECL_IGNORED_P.
12781 2017-07-04 Martin Liska <mliska@suse.cz>
12784 * ipa-inline.c (inline_small_functions):
12785 Use xstrdup_for_dump.
12787 2017-07-04 Tom de Vries <tom@codesourcery.com>
12789 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12791 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12794 * config/darwin.c (darwin_function_switched_text_sections):
12797 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12799 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12801 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12803 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12805 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12807 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12808 min_profitable_iters, and th as inclusive lower bounds.
12809 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12810 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12811 for min_profitable_iters and min_profitable_estimate.
12812 (vect_transform_loop): Treat th as an inclusive lower bound.
12813 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12815 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12818 * config/darwin.c (darwin_function_switched_text_sections):
12819 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12820 in two pieces, and suppress the use of buf.
12822 2017-07-03 Nathan Sidwell <nathan@acm.org>
12824 * hash-table.h (hash_table_mod1): Fix indentation.
12826 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12828 PR middle-end/81290
12829 * predict.c (force_edge_cold): Be more careful about propagation
12831 * profile-count.h (profile_probability::guessed,
12832 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12834 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12836 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12838 * doc/invoke.texi (rcpc architecture extension): Document it.
12840 2017-07-03 Richard Biener <rguenther@suse.de>
12842 PR tree-optimization/60510
12843 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
12844 the scalar reduction PHI and use it.
12845 (vectorizable_reduction): Properly guard the single_defuse_cycle
12846 path for non-SLP reduction chains where we cannot use it.
12847 Rework reduc_def/index and vector type deduction. Rework
12848 vector operand gathering during reduction op code-gen.
12849 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
12850 chains dissolve the chain and leave it to non-SLP reduction
12853 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12855 * tree-data-ref.h (dr_alignment): Declare.
12856 * tree-data-ref.c (dr_alignment): New function.
12857 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
12858 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
12860 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
12862 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12864 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
12865 and base_misalignment fields.
12866 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
12867 * tree-data-ref.c: Include builtins.h.
12868 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
12869 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
12870 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
12871 * tree-vect-data-refs.c: Include tree-cfg.h.
12872 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
12873 fields instead of calculating an alignment here.
12874 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
12875 innermost_loop_behavior fields.
12877 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12879 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
12881 (DR_STEP_ALIGNMENT): New macro.
12882 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
12883 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
12884 (create_data_ref): Print it.
12885 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
12886 to tell whether the step preserves vector (mis)alignment.
12887 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12888 Move the check for an integer step and generalise to all INTEGER_CST.
12889 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
12890 Print the outer step alignment.
12892 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12894 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
12895 with offset_alignment.
12896 (DR_ALIGNED_TO): Delete.
12897 (DR_OFFSET_ALIGNMENT): New macro.
12898 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
12899 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
12900 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
12901 (create_data_ref): Likewise.
12902 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
12903 (vect_analyze_data_refs): Likewise.
12904 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
12905 creating dummy innermost behavior.
12907 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12909 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
12910 with a "innermost_loop_behavior *" and refeence tree.
12911 * tree-data-ref.c (dr_analyze_innermost): Likewise.
12912 (create_data_ref): Update call accordingly.
12913 * tree-predcom.c (find_looparound_phi): Likewise.
12915 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12917 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
12918 fields with dr_wrt_vec_loop.
12919 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
12920 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
12921 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
12922 (vect_dr_behavior): New function.
12923 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12924 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
12925 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
12926 track whether the step preserves the misalignment.
12927 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
12928 Use vect_dr_behavior.
12929 (vect_setup_realignment): Update call accordingly.
12930 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
12931 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
12932 call to vect_create_addr_base_for_vector_ref.
12933 (vect_create_cond_for_align_checks): Likewise.
12934 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
12935 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
12936 (vect_recog_mask_conversion_pattern): Likewise.
12937 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
12938 (new_stmt_vec_info): Remove redundant zeroing.
12940 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12942 * common/config/arm/arm-common.c (arm_be8_option): New function.
12943 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
12944 (ISA_ARMv6): Add isa_bit_be8.
12945 * config/arm/arm.h (arm_be8_option): Add prototype.
12946 (BE8_SPEC_FUNCTION): New define.
12947 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
12948 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
12949 (mlittle-endian): Similarly.
12950 (mbe8, mbe32): New options.
12951 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
12952 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
12954 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12956 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
12958 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12960 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
12961 (cleanup_tree_cfg_bb): Use it.
12962 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
12964 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
12966 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12969 * loop-doloop.c (add_test): Update profile.
12971 2017-07-03 Martin Liska <mliska@suse.cz>
12974 * sanopt.c (rewrite_usage_of_param): New function.
12975 (sanitize_rewrite_addressable_params): Likewise.
12976 (pass_sanopt::execute): Call rewrite_usage_of_param.
12978 2017-07-03 Richard Biener <rguenther@suse.de>
12980 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
12981 back to using VIEW_CONVERT_EXPR.
12983 2017-07-03 Martin Liska <mliska@suse.cz>
12986 * doc/extend.texi: Document when a resolver function is
12987 generated for target_clones.
12989 2017-07-03 Martin Liska <mliska@suse.cz>
12991 * asan.c (asan_emit_stack_protection): Unpoison just red zones
12992 and shadow memory of auto variables which are subject of
12993 use-after-scope sanitization.
12994 (asan_expand_mark_ifn): Add do set only when is_poison.
12996 2016-07-03 Richard Biener <rguenther@suse.de>
12998 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13000 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13001 (vectorizable_reduction): Perform reduction PHI creation when
13002 visiting a reduction PHI and adjust and simplify code generation
13003 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13004 (vect_transform_loop): Visit reduction PHIs.
13005 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13006 defs into the SLP tree.
13007 (vect_build_slp_tree): Reduction defs terminate the recursion.
13008 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13010 (vect_get_vec_defs_for_stmt_copy): Export.
13011 (vect_get_vec_defs): Likewise.
13012 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13014 (vect_get_vec_defs_for_stmt_copy): Declare.
13015 (vect_get_vec_defs): Likewise.
13017 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13019 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13020 parameter with a "loop" parameter and use it instead of the
13021 loop containing DR_STMT. Don't check simple_iv when doing
13022 BB analysis. Describe the two analysis modes in the comment.
13024 2017-07-03 Tom de Vries <tom@codesourcery.com>
13026 PR tree-optimization/69468
13027 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13028 (find_same_succ_bb): Handle ignore_edge_flags.
13030 2017-07-03 Tom de Vries <tom@codesourcery.com>
13032 PR tree-optimization/81192
13033 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13035 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13037 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13039 2017-07-03 Tom de Vries <tom@codesourcery.com>
13041 PR tree-optimization/81192
13042 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13043 BB_SAME_SUCC (bb) == NULL.
13045 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13047 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13050 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13052 * dumpfile.c: Include profile-count.h
13053 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13055 (insert_cond_bb): Update profile.
13056 * tree-cfg.h (insert_cond_bb): Update prototype.
13057 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13058 * tree-dump.c: Do not include tree-cfg.
13060 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13062 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13064 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13066 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13069 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13071 * tree-complex.c (expand_complex_div_wide): update profile.
13073 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13074 Alan Hayward <alan.hayward@arm.com>
13075 David Sherwood <david.sherwood@arm.com>
13077 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13078 (CORETYPES_H): New define.
13079 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13080 (insn-modes-inline.h, s-modes-inline-h): New rules.
13081 (generated_files): Add insn-modes-inline.h.
13082 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13083 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13084 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13085 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13086 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13087 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13088 (build/gencodes.o, build/genconditions.o): Likewise.
13089 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13090 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13091 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13092 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13093 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13094 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13095 * coretypes.h: Include everything up to real.h for generators.
13096 Include insn-modes.h first. Include wide-int-print.h after
13097 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13098 * machmode.h: Don't include insn-modes.h here.
13099 * function-tests.c: Remove includes of signop.h, machmode.h,
13100 double-int.h and wide-int.h.
13102 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13104 * optc-save-gen.awk: Likewise.
13105 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13106 * godump.c: Remove include of wide-int-print.h.
13107 * pretty-print.h: Likewise.
13108 * wide-int-print.cc: Likewise.
13109 * wide-int.cc: Likewise.
13110 * hash-map-tests.c: Remove include of signop.h.
13111 * hash-set-tests.c: Likewise.
13112 * rtl-tests.c: Likewise.
13113 * mkconfig.sh: Remove include of machmode.h.
13114 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13116 (emit_insn_modes_inline_h): ...this new function. Emit the code
13117 into an insn-modes-inline.h header file, adding appropriate
13118 include guards and end comments.
13119 (emit_insn_modes_c_header): Remove include of machmode.h.
13120 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13122 (main): Handle -i flag and call emit_insn_modes_inline_h when
13125 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13127 * tree-ssa-strlen.c (strinfo): Rename the length field to
13128 nonzero_chars. Add a full_string_p field.
13129 (compare_nonzero_chars, zero_length_string_p): New functions.
13130 (get_addr_stridx): Add an offset_out parameter.
13131 Use compare_nonzero_chars.
13132 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13133 (new_strinfo): Update after above changes to strinfo.
13134 (set_endptr_and_length): Set full_string_p.
13135 (get_string_length): Update after above changes to strinfo.
13136 (unshare_strinfo): Update call to new_strinfo.
13137 (maybe_invalidate): Likewise.
13138 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13139 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13140 as a uhwi instead of an shwi. Update after above changes to
13141 strinfo and new_strinfo.
13142 (zero_length_string): Assert that chainsi contains full strings.
13143 Use zero_length_string_p. Update call to new_strinfo.
13144 (adjust_related_strinfos): Update after above changes to strinfo.
13145 Copy full_string_p from origsi.
13146 (adjust_last_stmt): Use zero_length_string_p.
13147 (handle_builtin_strlen): Update after above changes to strinfo and
13148 new_strinfo. Install the lhs as the string length if the previous
13149 entry didn't describe a full string.
13150 (handle_builtin_strchr): Update after above changes to strinfo
13152 (handle_builtin_strcpy): Likewise.
13153 (handle_builtin_strcat): Likewise.
13154 (handle_builtin_malloc): Likewise.
13155 (handle_pointer_plus): Likewise.
13156 (handle_builtin_memcpy): Likewise. Track nonzero characters
13157 that aren't necessarily followed by a nul terminator.
13158 (handle_char_store): Likewise.
13160 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13162 PR tree-optimization/80769
13163 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13164 for malloc and calloc. Document the new invariant that all related
13165 strinfos have delayed lengths or none do.
13166 (verify_related_strinfos): Move earlier in file.
13167 (set_endptr_and_length): New function, split out from...
13168 (get_string_length): ...here. Also set the lengths of related
13170 (zero_length_string): Assert that chainsi has known (rather than
13172 (adjust_related_strinfos): Likewise.
13174 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13176 PR tree-optimization/81136
13177 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13178 assert that two references with the same misalignment have the same
13179 compile-time misalignment if those compile-time misalignments
13182 2017-07-01 Andi Kleen <ak@linux.intel.com>
13184 * print-tree.c (print_node): Print all attributes.
13186 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13188 * cfg.c (scale_bbs_frequencies): New function.
13189 * cfg.h (scale_bbs_frequencies): Declare it.
13190 * cfgloopanal.c (single_likely_exit): Cleanup.
13191 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13193 (scale_loop_profile): Likewise.
13194 (loop_version): Likewise.
13195 (create_empty_loop_on_edge): Update.
13196 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13197 scale_loop_frequencies, scale_loop_profile, loopify,
13198 loop_version): Update prototypes.
13199 * modulo-sched.c (sms_schedule): Update.
13200 * predict.c (unlikely_executed_edge_p): Also check probability.
13201 (probably_never_executed_edge_p): Fix typo.
13202 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13203 * tree-parloops.c (gen_parallel_loop): Update.
13204 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13205 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13206 * tree-ssa-loop-split.c (split_loop): Update.
13207 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13208 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13209 (vect_loop_versioning): Update.
13210 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13212 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13214 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13216 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13218 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13219 to keep profile consistent.
13221 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13223 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13224 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13225 * profile-count.h (max_safe_multiplier): Make unsigned.
13226 (profile_count::guessed_zero): New.
13228 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13230 * bb-reorder.c (fix_up_crossing_landing_pad,
13231 fix_crossing_conditional_branches): Use make_single_succ_edge
13232 to keep profile consistent.
13234 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13236 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13239 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13242 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13243 the right scopes, make sure cond_jump isn't preserved between multiple
13244 iterations. Search for fallthru edge whenever there are 3+ edges and
13245 use find_fallthru_edge for it.
13247 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13249 Patch by Alexander Monakov <amonakov@ispras.ru>
13250 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13251 probabilities consistently.
13253 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13255 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13256 emit_cmp_and_jump_insns.
13258 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13261 * tree-inline.c (expand_call_inline): Combine profile statuses.
13263 2017-06-30 Andrew Pinski <apinski@cavium.com>
13265 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13266 fold_stmt returned true.
13268 2017-06-30 Nathan Sidwell <nathan@acm.org>
13270 * ggc.h (empty_string): Delete.
13271 * cfgexpand.c (expand_asm_stmt): Use plain "".
13272 * optabs.c (expand_asm_memory_barrier): Likewise.
13273 * stringpool.c (empty_string): Delete.
13274 (digit_vector, digit_string): Delete.
13275 (ggc_alloc_string): Use plain "", don't optimize single digit
13276 strings. Use ggc_alloc_atomic.
13278 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13280 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13281 comparison set and one other set, use the cost of the non-comparison
13284 2017-06-30 Nathan Sidwell <nathan@acm.org>
13286 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13289 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13291 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13292 loops. Remove now unneeded calls to gimple_switch_set_label() that
13293 just set removed labels to NULL_TREE.
13295 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13297 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13298 (set_range_info): ...here. Only call set_range_info_raw if domain
13300 (set_nonzero_bits): Call set_range_info_raw.
13301 * tree-ssanames.h (set_range_info_raw): New.
13303 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13306 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13307 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13308 of nonimmediate_operand and <store_mask_constraint> instead of m
13309 for the input operand. For V8FI iterator, always split if input
13310 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13311 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13312 <store_mask_predicate> instead of register_operand and
13313 <store_mask_constraint> instead of v for the input operand. Make
13314 sure both operands aren't MEMs for if not <mask_applied>.
13316 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13318 * lto-wrapper.c (copy_file) Close both file descriptors before
13321 2017-06-30 Martin Liska <mliska@suse.cz>
13324 * multiple_target.c (create_dispatcher_calls): Make ifunc
13325 also for function that don't have calls or are not referenced.
13327 2017-06-30 Richard Biener <rguenther@suse.de>
13329 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13330 analyze the first scalar stmt. Move vector type computation
13331 for the BB case here from ...
13332 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13333 live operation processing in the SLP case properly.
13335 2017-06-30 Richard Biener <rguenther@suse.de>
13337 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13339 2017-06-30 Martin Liska <mliska@suse.cz>
13342 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13343 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13345 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13347 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13350 2017-06-30 Richard Biener <rguenther@suse.de>
13352 PR tree-optimization/81249
13353 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13354 condition reduction result to original scalar type.
13356 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13358 * profile-count.h (enum profile_quality): Fix typos and whitespace
13361 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13363 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13364 type for branch probabilities.
13366 2017-06-29 Julian Brown <julian@codesourcery.com>
13367 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13369 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13370 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13371 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13372 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13374 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13376 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13377 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13378 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13379 CC usage from generic code to here.
13380 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13381 CC usage into the target macros.
13383 2017-06-29 Maya Rashish <coypu@sdf.org>
13385 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13388 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13390 * arm/arm-builtins.c: Include profile-count.h
13391 * except.c (sjlj_emit_function_enter): Use
13392 profile_probability::unlikely.
13394 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13396 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13397 and tocrel_offset be pointer args rather than implicitly using
13399 (legitimate_constant_pool_address_p, rs6000_emit_move,
13400 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13401 tocrel_offset and use in toc_relative_expr_p call.
13402 (print_operand, print_operand_address): Use static tocrel_base_oac
13403 and tocrel_offset_oac.
13404 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13407 2017-06-29 Maya Rashish <coypu@sdf.org>
13409 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13411 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13413 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13414 objects, take into account only the alignment of 'op0' and 'mode1' if
13417 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13419 * ccmp.c (ccmp_tree_comparison_p): New function.
13420 (ccmp_candidate_p): Update to use above function.
13421 (get_compare_parts): New function.
13422 (expand_ccmp_next): Update to use new functions.
13423 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13425 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13426 take mode as argument.
13427 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13428 * expr.c (expand_expr_real_1): Pass mode as argument.
13430 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13432 * combine.c (combine_instructions): Print insns to dump_file, together
13435 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13437 * asan.c (asan_emit_stack_protection): Update.
13438 (create_cond_insert_point): Update.
13439 * auto-profile.c (afdo_propagate_circuit): Update.
13440 * basic-block.h (struct edge_def): Turn probability to
13441 profile_probability.
13442 (EDGE_FREQUENCY): Update.
13443 * bb-reorder.c (find_traces_1_round): Update.
13444 (better_edge_p): Update.
13445 (sanitize_hot_paths): Update.
13446 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13447 (make_single_succ_edge): Update.
13448 (check_bb_profile): Update.
13449 (dump_edge_info): Update.
13450 (update_bb_profile_for_threading): Update.
13451 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13452 probabilitycount to 0.
13453 * cfgbuild.c (compute_outgoing_frequencies): Update.
13454 * cfgcleanup.c (try_forward_edges): Update.
13455 (outgoing_edges_match): Update.
13456 (try_crossjump_to_edge): Update.
13457 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13458 (expand_gimple_tailcall): Update.
13459 (construct_init_block): Use make_single_succ_edge.
13460 (construct_exit_block): Use make_single_succ_edge.
13461 * cfghooks.c (verify_flow_info): Update.
13462 (redirect_edge_succ_nodup): Update.
13463 (split_edge): Update.
13464 (account_profile_record): Update.
13465 * cfgloopanal.c (single_likely_exit): Update.
13466 * cfgloopmanip.c (scale_loop_profile): Update.
13467 (set_zero_probability): Remove.
13468 (duplicate_loop_to_header_edge): Update.
13469 * cfgloopmanip.h (loop_version): Update prototype.
13470 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13471 (force_nonfallthru_and_redirect): Update.
13472 (update_br_prob_note): Update.
13473 (rtl_verify_edges): Update.
13474 (purge_dead_edges): Update.
13475 (rtl_lv_add_condition_to_bb): Update.
13476 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13477 * cgraphunit.c (init_lowered_empty_function): Update.
13478 (cgraph_node::expand_thunk): Update.
13479 * cilk-common.c: Include profile-count.h
13480 * dojump.c (inv): Remove.
13481 (jumpifnot): Update.
13482 (jumpifnot_1): Update.
13483 (do_jump_1): Update.
13485 (do_jump_by_parts_greater_rtx): Update.
13486 (do_compare_rtx_and_jump): Update.
13487 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13488 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13489 * dwarf2cfi.c: Include profile-count.h
13490 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13491 (sjlj_emit_dispatch_table): Likewise.
13492 * explow.c: Include profile-count.h
13493 * expmed.c (emit_store_flag_force): Update.
13494 (do_cmp_and_jump): Update.
13495 * expr.c (compare_by_pieces_d::generate): Update.
13496 (compare_by_pieces_d::finish_mode): Update.
13497 (emit_block_move_via_loop): Update.
13498 (store_expr_with_bounds): Update.
13499 (store_constructor): Update.
13500 (expand_expr_real_2): Update.
13501 (expand_expr_real_1): Update.
13502 * expr.h (try_casesi, try_tablejump): Update prototypes.
13503 * gimple-pretty-print.c (dump_probability): Update.
13504 (dump_profile): New.
13505 (dump_gimple_label): Update.
13506 (dump_gimple_bb_header): Update.
13507 * graph.c (draw_cfg_node_succ_edges): Update.
13508 * hsa-gen.c (convert_switch_statements): Update.
13509 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13510 (find_if_case_1): Update.
13511 (find_if_case_2): Update.
13512 * internal-fn.c (expand_arith_overflow_result_store): Update.
13513 (expand_addsub_overflow): Update.
13514 (expand_neg_overflow): Update.
13515 (expand_mul_overflow): Update.
13516 (expand_vector_ubsan_overflow): Update.
13517 * ipa-cp.c (good_cloning_opportunity_p): Update.
13518 * ipa-split.c (split_function): Use make_single_succ_edge.
13519 * ipa-utils.c (ipa_merge_profiles): Update.
13520 * loop-doloop.c (add_test): Update.
13521 (doloop_modify): Update.
13522 * loop-unroll.c (compare_and_jump_seq): Update.
13523 (unroll_loop_runtime_iterations): Update.
13524 * lra-constraints.c (lra_inheritance): Update.
13525 * lto-streamer-in.c (input_cfg): Update.
13526 * lto-streamer-out.c (output_cfg): Update.
13527 * mcf.c (adjust_cfg_counts): Update.
13528 * modulo-sched.c (sms_schedule): Update.
13529 * omp-expand.c (expand_omp_for_init_counts): Update.
13530 (extract_omp_for_update_vars): Update.
13531 (expand_omp_ordered_sink): Update.
13532 (expand_omp_for_ordered_loops): Update.
13533 (expand_omp_for_generic): Update.
13534 (expand_omp_for_static_nochunk): Update.
13535 (expand_omp_for_static_chunk): Update.
13536 (expand_cilk_for): Update.
13537 (expand_omp_simd): Update.
13538 (expand_omp_taskloop_for_outer): Update.
13539 (expand_omp_taskloop_for_inner): Update.
13540 * omp-simd-clone.c (simd_clone_adjust): Update.
13541 * optabs.c (expand_doubleword_shift): Update.
13542 (expand_abs): Update.
13543 (emit_cmp_and_jump_insn_1): Update.
13544 (expand_compare_and_swap_loop): Update.
13545 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13546 * predict.c (predictable_edge_p): Update.
13547 (edge_probability_reliable_p): Update.
13548 (set_even_probabilities): Update.
13549 (combine_predictions_for_insn): Update.
13550 (combine_predictions_for_bb): Update.
13551 (propagate_freq): Update.
13552 (estimate_bb_frequencies): Update.
13553 (force_edge_cold): Update.
13554 * profile-count.c (profile_count::dump): Add missing space into dump.
13555 (profile_count::debug): Add newline.
13556 (profile_count::differs_from_p): Explicitly convert to unsigned.
13557 (profile_count::stream_in): Update.
13558 (profile_probability::dump): New member function.
13559 (profile_probability::debug): New member function.
13560 (profile_probability::differs_from_p): New member function.
13561 (profile_probability::differs_lot_from_p): New member function.
13562 (profile_probability::stream_in): New member function.
13563 (profile_probability::stream_out): New member function.
13564 * profile-count.h (profile_count_quality): Rename to ...
13565 (profile_quality): ... this one.
13566 (profile_probability): New.
13567 (profile_count): Update.
13568 * profile.c (compute_branch_probabilities): Update.
13569 * recog.c (peep2_attempt): Update.
13570 * sched-ebb.c (schedule_ebbs): Update.
13571 * sched-rgn.c (find_single_block_region): Update.
13572 (compute_dom_prob_ps): Update.
13573 (schedule_region): Update.
13574 * sel-sched-ir.c (compute_succs_info): Update.
13575 * stmt.c (struct case_node): Update.
13576 (do_jump_if_equal): Update.
13577 (get_outgoing_edge_probs): Update.
13578 (conditional_probability): Update.
13579 (emit_case_dispatch_table): Update.
13580 (expand_case): Update.
13581 (expand_sjlj_dispatch_table): Update.
13582 (emit_case_nodes): Update.
13583 * targhooks.c: Update.
13584 * tracer.c (better_p): Update.
13585 (find_best_successor): Update.
13586 * trans-mem.c (expand_transaction): Update.
13587 * tree-call-cdce.c: Update.
13588 * tree-cfg.c (gimple_split_edge): Upate.
13589 (move_sese_region_to_fn): Upate.
13590 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13591 * tree-eh.c (lower_resx): Upate.
13592 (cleanup_empty_eh_move_lp): Upate.
13593 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13594 * tree-inline.c (copy_edges_for_bb): Update.
13595 (copy_cfg_body): Update.
13596 * tree-parloops.c (gen_parallel_loop): Update.
13597 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13598 (gimple_gen_time_profiler): Update.
13599 * tree-ssa-dce.c (remove_dead_stmt): Update.
13600 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13601 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13602 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13603 (unloop_loops): Update.
13604 (try_peel_loop): Update.
13605 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13606 * tree-ssa-loop-split.c (connect_loops): Update.
13607 (split_loop): Update.
13608 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13609 (hoist_guard): Update.
13610 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13611 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13612 (value_replacement): Update.
13613 * tree-ssa-reassoc.c (branch_fixup): Update.
13614 * tree-ssa-tail-merge.c (replace_block_by): Update.
13615 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13616 (create_edge_and_update_destination_phis): Update.
13617 (compute_path_counts): Update.
13618 (recompute_probabilities): Update.
13619 (update_joiner_offpath_counts): Update.
13620 (freqs_to_counts_path): Update.
13621 (duplicate_thread_path): Update.
13622 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13623 (struct switch_conv_info): Update.
13624 (gen_inbound_check): Update.
13625 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13626 (vect_do_peeling): Update.
13627 (vect_loop_versioning): Update.
13628 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13629 (optimize_mask_stores): Update.
13630 * ubsan.c (ubsan_expand_null_ifn): Update.
13631 * value-prof.c (gimple_divmod_fixed_value): Update.
13632 (gimple_divmod_fixed_value_transform): Update.
13633 (gimple_mod_pow2): Update.
13634 (gimple_mod_pow2_value_transform): Update.
13635 (gimple_mod_subtract): Update.
13636 (gimple_mod_subtract_transform): Update.
13637 (gimple_ic): Update.
13638 (gimple_stringop_fixed_value): Update.
13639 (gimple_stringops_transform): Update.
13640 * value-prof.h: Update.
13642 2017-06-29 Carl Love <cel@us.ibm.com>
13644 * config/rs6000/rs6000-c.c: Add support for built-in functions
13645 vector signed int vec_signed (vector float);
13646 vector signed long long vec_signed (vector double);
13647 vector signed int vec_signed2 (vector double, vector double);
13648 vector signed int vec_signede (vector double);
13649 vector signed int vec_signedo (vector double);
13650 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13651 instruction generator.
13652 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13653 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13654 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13656 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13657 vunsignede_v2df): Add define_expands.
13658 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13659 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13660 VEC_UNSIGNEDO): Add definitions.
13661 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13662 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13663 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13664 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13665 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13666 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13667 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13668 vec_unsignede, vec_unsignedo): Add builtin defines.
13669 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13671 * doc/extend.texi: Update the built-in documentation file for the
13672 new built-in functions.
13674 2017-06-29 Richard Biener <rguenther@suse.de>
13676 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13677 reduction chains to LOOP_VINFO_REDUCTIONS.
13678 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13679 SLP reductions after processing reduction chains.
13681 2017-06-29 Nathan Sidwell <nathan@acm.org>
13683 * builtins.c (fold_builtin_FUNCTION): Use
13684 lang_hooks.decl_printable_name.
13686 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13688 PR middle-end/81194
13689 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13690 with only one label.
13691 * stmt.c (expand_case): Assert NCASES is greater than one.
13693 2017-06-29 Richard Biener <rguenther@suse.de>
13695 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13697 (group_case_labels): Likewise.
13698 (find_taken_edge): Push sanity checking on val to workers...
13699 (find_taken_edge_cond_expr): ... here
13700 (find_taken_edge_switch_expr): ... and here, handle cases
13701 with just a default label.
13702 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13703 (group_case_labels): Likewise.
13704 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13705 group_case_labels does anything cleanup the CFG again.
13707 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13709 PR tree-optimization/81196
13710 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13711 exit condition comparing two IVs.
13713 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13715 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13716 profile to the dummy entry at the end of the list of architectures.
13717 * config/arm/arm-cpu-cdata.h: Regenerated.
13719 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13720 Michael Collison <michael.collison@arm.com>
13723 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13725 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13726 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13727 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13728 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13729 of shift when the shift amount is masked with constant equal to
13730 the size of the mode.
13731 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13734 2017-06-29 Martin Liska <mliska@suse.cz>
13736 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13738 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13741 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13744 (edge_growth_cache_entry::edge_growth_cache_entry): New
13746 (reset_edge_growth_cache): Update to use constructor.
13748 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13750 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13751 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13752 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13754 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13756 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13757 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13759 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13761 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13762 (*-linux-uclibc*): Add t-uclibc tmake_file.
13763 * config/t-musl: New.
13764 * config/t-uclibc: New.
13766 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13768 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13770 (gen_comm_data): Emit architectural setting of arch_prof.
13771 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13773 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13774 (armv8-m.base, armv8-m.main): Likewise.
13775 * arm-protos.h (arm_build_target): Add profile field.
13776 (arch_option): Likewise.
13777 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13779 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13780 arm_active_target.profile.
13782 2017-06-28 Richard Biener <rguenther@suse.de>
13784 PR middle-end/81227
13785 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13786 TYPE_OVERFLOW_WRAPS.
13787 * match.pd (negate_expr_p): Likewise.
13788 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13789 fold_build2, not fold_binary.
13791 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13793 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13794 Convert memory address to Pmode.
13795 (aarch64_print_operand): Assert MEM operands are always Pmode.
13797 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13800 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13801 Remove redundant if.
13802 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13803 * config/arm/aarch-common-protos.h
13804 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13805 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13807 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13810 * multiple_target.c (create_dispatcher_calls): Set the default
13811 clone to be static, not public.
13813 2017-06-28 Richard Biener <rguenther@suse.de>
13815 * tree-vect-loop.c (vectorizable_reduction): Move special
13816 cond reduction IV var creation ...
13817 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13818 parameter. Use STMT_VINFO_VECTYPE.
13819 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13822 2017-06-28 Martin Liska <mliska@suse.cz>
13825 * ipa-visibility.c (non_local_p): Handle visibility.
13827 2017-06-28 Martin Liska <mliska@suse.cz>
13830 * common.opt: Add IntegerRange to various options.
13831 * opt-functions.awk (integer_range_info): New function.
13832 * optc-gen.awk: Add integer_range_info to cl_options struct.
13833 * opts-common.c (decode_cmdline_option): Handle
13834 CL_ERR_INT_RANGE_ARG.
13835 (cmdline_handle_error): Likewise.
13836 * opts.c (print_filtered_help): Show valid interval in
13837 when --help is provided.
13838 * opts.h (struct cl_option): Add range_min and range_max fields.
13839 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
13841 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
13843 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
13844 (x * C EQ/NE y * C): New transformation.
13846 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
13848 * genmultilib (combination_space): Accept '+' in option names.
13850 2017-06-28 Martin Liska <mliska@suse.cz>
13853 * asan.c (instrument_derefs): Bail out inner references
13854 that are hard register variables.
13856 2017-06-28 Jakub Jelinek <jakub@redhat.com>
13859 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
13860 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
13862 2017-06-28 Richard Biener <rguenther@suse.de>
13864 * tree-vectorizer.h (vect_get_vec_defs): Remove.
13865 (vect_get_slp_defs): Adjust.
13866 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
13868 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
13870 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
13871 get_initial_defs_for_reduction instead of vect_get_vec_defs.
13872 (vectorizable_reduction): Adjust.
13873 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
13875 (vect_get_slp_defs): Likewise.
13876 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
13877 (vectorizable_bswap): Adjust.
13878 (vectorizable_call): Likewise.
13879 (vectorizable_conversion): Likewise.
13880 (vectorizable_assignment): Likewise.
13881 (vectorizable_shift): Likewise.
13882 (vectorizable_operation): Likewise.
13883 (vectorizable_store): Likewise.
13884 (vectorizable_condition): Likewise.
13885 (vectorizable_comparison): Likewise.
13887 2017-06-28 Michael Collison <michael.collison@arm.com>
13890 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
13892 (arm_gen_movmemqi): Removed unused variable 'i'.
13893 Convert 'for' loop into 'while' loop.
13894 (arm_expand_prologue): Remove last unnecessary set of insn.
13895 (thumb_pop): Remove unused variable 'pushed_words'.
13896 (thumb_exit): Remove last unnecessary set of regs_to_pop.
13898 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13900 * config/s390/predicates.md: Use s390_rel_address_ok_p.
13901 * config/s390/s390-protos.h: Add prototype of
13902 s390_rel_address_ok_p.
13903 * config/s390/s390.c (s390_got_symbol): New function.
13904 (s390_rel_address_ok_p): New function.
13905 (legitimize_pic_address): Use s390_rel_address_ok_p.
13906 (s390_load_got): Use s390_got_symbol.
13907 (s390_option_override): Issue error if
13908 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
13909 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
13911 * config/s390/s390.opt: New option mpic-data-is-text-relative.
13913 2017-06-27 Andrew Pinski <apinski@cavium.com>
13915 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
13916 (X * copysign (1.0, X)): New pattern.
13917 (X * copysign (1.0, -X)): New pattern.
13918 (copysign (-1.0, CST)): New pattern.
13920 2017-06-27 Joseph Myers <joseph@codesourcery.com>
13922 * genmultilib (combination_space): Remove variable.
13923 Validate reuse rules against regular expression for any sequence
13924 of multilib options in any order.
13926 2017-06-27 Michael Collison <michael.collison@arm.com>
13928 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
13929 call aarch64_split_simd_combine.
13930 * (aarch64_combine_internal<mode>): Delete pattern.
13931 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
13932 Allow register and subreg operands.
13934 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13936 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
13937 specific need, just fallback on defaults.
13938 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
13940 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13941 Olivier Hainque <hainque@adacore.com>
13943 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
13945 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
13946 targets. Pick a default if no particular attempt applied.
13947 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
13950 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13952 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
13953 (x86_64-wrs-vxworks7): Likewise.
13955 2017-06-27 Marek Polacek <polacek@redhat.com>
13958 * ubsan.c (instrument_null): Check get_base_address's result for null.
13960 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13962 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
13964 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
13966 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
13967 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
13968 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
13969 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
13970 New function types.
13971 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
13972 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
13973 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
13974 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
13975 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
13976 BUILT_IN_FEUPDATEENV): New builtins.
13977 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
13978 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
13979 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
13980 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
13982 (builtin_structptr_types): Adjust size.
13983 * tree.c (builtin_structptr_types): Add four entries.
13985 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
13986 Olivier Hainque <hainque@adacore.com>
13988 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
13989 (TLS_SYM): New local macro, forcing reference to __tls__ on
13990 link command lines for VxWorks 7 RTPs, triggering initialization
13992 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
13993 OS features TLS support, true for RTPs on VxWorks 7.
13994 * config/vxworks.c (vxworks_override_options): Setup emutls
13997 2017-06-27 Jakub Jelinek <jakub@redhat.com>
13999 * predict.c (test_prediction_value_range): Use -1U instead of -1
14000 to avoid narrowing conversion warning.
14001 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14002 to avoid narrowing conversion warning.
14003 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14005 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14007 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14009 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14010 64bit configurations.
14011 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14012 (SIZE_TYPE): Likewise.
14013 * config/vxworks.c (vxworks_emutls_var_fields): Use
14014 long_unsigned_type_node instead of unsigned_type_node as the offset
14015 field type, which is "pointer" mode in emutls.c.
14017 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14020 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14022 PR middle-end/81207
14023 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14024 gimple_vuse copying separately from gimple_vdef copying.
14026 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14028 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14031 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14032 Olivier Hainque <hainque@adacore.com>
14034 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14035 for all vxworks7 targets.
14036 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14037 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14038 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14039 variations for VX6/VX7 and 32/64bits later on in ...
14040 (VXWORKS_LIB_SPEC): Leverage new macros.
14041 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14042 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14044 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14046 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14047 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14049 2017-06-26 Carl Love <cel@us.ibm.com>
14051 * config/rs6000/rs6000-c.c: Add support for built-in functions
14052 vector bool char vec_reve (vector bool char);
14053 vector signed char vec_reve (vector signed char);
14054 vector unsigned char vec_reve (vector unsigned char);
14055 vector bool int vec_reve (vector bool int);
14056 vector signed int vec_reve (vector signed int);
14057 vector unsigned int vec_reve (vector unsigned int);
14058 vector bool long long vec_reve (vector bool long long);
14059 vector signed long long vec_reve (vector signed long long);
14060 vector unsigned long long vec_reve (vector unsigned long long);
14061 vector bool short vec_reve (vector bool short);
14062 vector signed short vec_reve (vector signed short);
14063 vector double vec_reve (vector double);
14064 vector float vec_reve (vector float);
14065 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14066 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14067 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14068 (altivec_vreve): New pattern.
14069 * config/rs6000/altivec.h (vec_reve): New define.
14070 * doc/extend.texi (vec_rev): Update the built-in documentation file
14071 for the new built-in functions.
14073 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14075 PR tree-optimization/71815
14076 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14078 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14080 (slsr_process_phi): Likewise.
14081 (replace_uncond_cands_and_profitable_phis): Don't replace a
14082 multiply candidate with a stride of 1 (copy or cast).
14083 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14085 (lowest_cost_path): Likewise.
14086 (total_savings): Likewise.
14088 2017-06-26 Richard Biener <rguenther@suse.de>
14091 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14092 Use def_builtin_pure for all gather builtins.
14094 2017-06-26 Richard Biener <rguenther@suse.de>
14096 PR tree-optimization/81203
14097 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14098 non-dominating BBs.
14100 2017-06-26 Marek Polacek <polacek@redhat.com>
14103 * doc/invoke.texi: Document -Wmultistatement-macros.
14105 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14107 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14108 arm_neon_ok_no_float_abi effective target.
14110 2017-06-26 Richard Biener <rguenther@suse.de>
14112 PR tree-optimization/80928
14113 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14114 (copy_bbs): Set BB_DUPLICATED flag early.
14115 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14117 (execute_on_shrinking_pred): Likewise.
14118 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14119 BB_DUPLICATED blocks.
14120 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14121 iterate over all PHIs considering removal of *gsi.
14123 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14125 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14128 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14130 * config/rs6000/rs6000-string.c: (expand_block_clear,
14131 do_load_for_compare, select_block_compare_mode,
14132 compute_current_alignment, expand_block_compare,
14133 expand_strncmp_align_check, expand_strn_compare,
14134 expand_block_move, rs6000_output_load_multiple)
14135 Move functions related to string/block move/compare
14136 to a separate file.
14137 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14138 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14139 for this function which is now used in two files.
14140 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14141 * config.gcc: Add rs6000-string.o to extra_objs for
14142 targets powerpc*-*-* and rs6000*-*-*.
14144 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14147 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14148 32-bit, since indexed is not valid for DImode.
14149 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14150 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14151 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14152 (define_peephole2 for Altivec d-form store): Likewise.
14155 * multiple_target.c (create_dispatcher_calls): Only create the
14156 dispatcher call if the function is the default clone of a
14157 versioned function.
14159 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14161 PR middle-end/80902
14162 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14163 a call, force the call to not be a tail call.
14165 2017-06-23 Jeff Law <law@redhat.com>
14167 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14170 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14172 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14173 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14174 handling for early expansion of vector shifts (sl,sr,sra,rl).
14175 (builtin_function_type): Add vector shift right instructions
14176 to the unsigned argument list.
14178 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14180 rtl-optimizatoin/79286
14181 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14182 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14183 trap. PIC register plus a const unspec without offset can never trap.
14185 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14187 * tree.h (builtin_structptr_type): New type.
14188 (builtin_structptr_types): Declare new array.
14189 * tree.c (builtin_structptr_types): New array.
14190 (free_lang_data, build_common_tree_nodes): Use it.
14192 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14195 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14198 2017-06-22 Matt Turner <mattst88@gmail.com>
14200 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14201 Lake models to skylake case. Assume skylake for unknown
14202 models with clflushopt.
14204 2017-06-22 Jeff Law <law@redhat.com>
14206 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14207 frame sizes that do not satisfy aarch64_uimm12_shift.
14209 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14211 * profile-count.h (apply_probability,
14212 apply_scale, probability_in): Fix checks for zero.
14214 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14216 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14217 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14219 2016-06-22 Richard Biener <rguenther@suse.de>
14221 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14222 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14223 REDUC_MAX_EXPR support.
14224 (vectorizable_reduction): Likewise.
14225 (vect_create_epilog_for_reduction): Likewise.
14227 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14229 * match.pd (A / (1 << B) -> A >> B): New.
14230 * generic-match-head.c: Include optabs-tree.h.
14231 * gimple-match-head.c: Likewise.
14232 * optabs-tree.h (target_supports_op_p): New.
14233 * optabs-tree.c (target_supports_op_p): New.
14235 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14237 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14238 $gcc_cv_ld --help output.
14239 (gcc_cv_ld_demangle): Likewise.
14240 (gcc_cv_ld_eh_frame_hdr): Likewise.
14241 (gcc_cv_ld_pie): Likewise.
14242 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14243 (gcc_cv_ld_buildid): Likewise.
14244 (gcc_cv_ld_sysroot): Likewise.
14245 (ld_bndplt_support): Likewise.
14246 (ld_pushpopstate_support): Likewise.
14247 * configure: Regenerate.
14248 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14250 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14253 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14254 operands indexes to avoid gap between operands and match_dups.
14256 2017-06-21 Andrew Pinski <apinski@cavium.com>
14258 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14259 Increment Arith_shift and Arith_shift_reg by 1.
14260 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14262 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14263 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14264 (aarch64_strip_extend): Add new argument and test for it.
14265 (aarch64_cheap_mult_shift_p): New function.
14266 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14267 add a cost if it is true.
14268 Update calls to aarch64_strip_extend.
14269 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14271 2017-06-21 Andrew Pinski <apinski@cavium.com>
14273 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14275 (thunderxt88): Likewise.
14276 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14277 (thunderx_prefetch_tune): New variable.
14278 (thunderx2t99_prefetch_tune): Update for the correct values.
14279 (thunderxt88_tunings): New variable.
14280 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14281 generic_prefetch_tune.
14282 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14284 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14286 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14287 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14288 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14289 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14290 (aarch64_atomic_cas<mode>, GPI): Likewise.
14292 2017-06-21 Martin Liska <mliska@suse.cz>
14294 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14295 statements on cold and hot labels.
14296 * predict.c (tree_estimate_probability_bb): Remove the
14297 prediction from this place.
14299 2017-06-21 Martin Liska <mliska@suse.cz>
14301 PR tree-optimization/79489
14302 * gimplify.c (maybe_add_early_return_predict_stmt): New
14304 (gimplify_return_expr): Call the function.
14305 * predict.c (tree_estimate_probability_bb): Remove handling
14307 * predict.def: Update comment about early return predictor.
14308 * gimple-predict.h (is_gimple_predict): New function.
14309 * predict.def: Change default value of early return to 66.
14310 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14312 * passes.def: Put pass_strip_predict_hints to the beginning of
14315 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14317 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14318 FUNCTION_DECL declarations.
14319 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14321 (dwaf2out_decl): Likewise.
14322 * godump.c (go_early_global_decl): Skip call to the real debug hook
14323 for FUNCTION_DECL declarations.
14324 * passes.c (rest_of_decl_compilation): Skip call to the
14325 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14326 -fdump-go-spec is passed.
14328 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14330 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14331 Reorder for compactness.
14332 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14333 (def_builtin_pure, def_builtin_pure2): New functions.
14334 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14336 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14338 * match.pd (nop_convert): New predicate.
14339 ((A +- CST1) +- CST2): Allow some NOP conversions.
14341 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14344 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14345 with ctors/dtors if GOVD_SHARED is set.
14347 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14349 * config/aarch64/aarch64.md (movti_aarch64):
14350 Emit mov rather than orr.
14351 (movtf_aarch64): Likewise.
14352 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14353 Emit mov rather than orr.
14355 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14357 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14358 Swap alternatives, make integer dup more expensive.
14360 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14362 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14363 Return true for non-tls symbols.
14365 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14367 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14368 (cortex-a75): Likewise.
14369 (cortex-a75.cortex-a55): Likewise.
14370 * config/aarch64/aarch64-tune.md: Regenerate.
14371 * doc/invoke.texi (-mtune): Document new values for -mtune.
14373 2017-06-21 Tom de Vries <tom@codesourcery.com>
14375 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14376 stack_size feature.
14377 (Effective-Target Keywords, Other attributes): Suggest using
14378 dg-add-options stack_size feature to get stack limit in stack_size
14379 effective target documentation.
14381 2017-06-21 Julian Brown <julian@codesourcery.com>
14382 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14384 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14385 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14386 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14388 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14389 attribute type list for neon_multiply.
14390 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14391 attribute type list for neon_multiply.
14392 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14393 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14394 attribute type list for neon_multiply.
14395 * config/arm/types.md (crypto_pmull): Add.
14396 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14397 attribute type list.
14399 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14401 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14404 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14406 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14407 to make sure not to dereference a NULL cost_classes_ptr pointer.
14409 2017-06-20 Carl Love <cel@us.ibm.com>
14411 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14412 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14413 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14414 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14415 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14416 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14417 VMULOSW): New enum "unspec" values.
14418 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14419 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14420 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14421 altivec_vmulosw): New patterns.
14422 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14423 VMULOSW): Add definitions.
14425 2017-06-20 Julia Koval <julia.koval@intel.com>
14427 * config/i386/i386.c: Fix rounding expand for new pattern.
14428 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14430 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14432 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14433 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14435 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14437 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14440 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14442 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14443 architecture, then by implementer ID.
14444 * config/aarch64/aarch64-tune.md: Regenerate.
14446 2017-06-20 Richard Biener <rguenther@suse.de>
14448 PR middle-end/81097
14449 * fold-const.c (split_tree): Fold to type before negating.
14451 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14453 * diagnostic-show-locus.c
14454 (selftest::test_fixit_deletion_affecting_newline): New function.
14455 (selftest::diagnostic_show_locus_c_tests): Call it.
14457 2017-06-20 Andreas Schwab <schwab@suse.de>
14460 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14463 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14465 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14466 __ARM_FEATURE_COPROC according to support.
14468 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14470 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14471 Rewritten to avoid overflow for > 32-bit pointers.
14474 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14475 by removing enum keyword.
14476 (ubsan_type_descriptor): Likewise. Formatting fix.
14479 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14480 splitter): Require TARGET_SSE2 in the condition.
14482 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14485 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14486 for doing vector set of SFmode on ISA 3.0.
14487 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14488 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14490 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14491 SFmode value into a V4SF variable that was extracted from another
14492 V4SF variable without converting the element to double precision
14493 and back to single precision vector format.
14494 (vsx_insert_extract_v4sf_p9_2): Likewise.
14496 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14498 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14499 in UWHI to avoid undefined overflow.
14502 * ubsan.h (enum ubsan_encode_value_phase): New.
14503 (ubsan_encode_value): Change second argument to
14504 enum ubsan_encode_value_phase with default value of
14505 UBSAN_ENCODE_VALUE_GENERIC.
14506 * ubsan.c (ubsan_encode_value): Change second argument to
14507 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14508 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14509 create_tmp_var_raw instead of create_tmp_var and use a
14511 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14512 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14513 ubsan_encode_value callers.
14516 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14517 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14518 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14520 2017-06-19 Richard Biener <rguenther@suse.de>
14522 PR middle-end/81118
14523 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14524 estimates if we changed anything.
14526 2017-06-19 Richard Biener <rguenther@suse.de>
14528 PR tree-optimization/80887
14529 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14530 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14531 simplified lookups, then reset mprts_hook.
14532 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14534 (try_to_simplify): Likewise.
14536 2017-06-19 Martin Liska <mliska@suse.cz>
14539 * gimplify.c (gimplify_switch_expr):
14540 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14542 2017-06-19 Martin Liska <mliska@suse.cz>
14544 * doc/install.texi: Document that PGO runs in 4 stages.
14546 2017-06-19 Martin Liska <mliska@suse.cz>
14549 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14550 to dispatcher function name.
14551 * multiple_target.c (replace_function_decl): New function.
14552 (create_dispatcher_calls): Redirect both edges and references.
14554 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14556 * profile-count.c (profile_count::dump): Dump quality.
14557 (profile_count::differs_from_p): Update for unsigned val.
14558 * profile-count.h (profile_count_quality): New enum.
14559 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14561 2017-06-19 Richard Biener <rguenther@suse.de>
14563 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14564 struct function as arg.
14565 (estimate_numbers_of_iterations): Export overload with loop arg.
14566 (free_numbers_of_iterations_estimates_loop): Use an overload of
14567 free_numbers_of_iterations_estimates instead.
14568 * tree-cfg.c (remove_bb): Adjust.
14569 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14570 * tree-parloops.c (gen_parallel_loop): Likewise.
14571 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14573 (tree_unroll_loops_completely): Likewise.
14574 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14575 Use an overload instead and export.
14576 (estimated_loop_iterations): Adjust.
14577 (max_loop_iterations): Likewise.
14578 (likely_max_loop_iterations): Likewise.
14579 (estimate_numbers_of_iterations): Take struct function as arg
14581 (loop_exits_before_overflow): Adjust.
14582 (free_numbers_of_iterations_estimates_loop): Use an overload.
14583 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14584 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14586 2017-06-19 Richard Biener <rguenther@suse.de>
14589 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14590 RANGE_EXPR conservatively.
14592 2017-06-16 Carl Love <cel@us.ibm.com>
14594 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14595 definitions for vec_float, vec_float2, vec_floato,
14596 vec_floate built-ins.
14597 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14598 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14600 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14601 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14602 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14603 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14604 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14605 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14606 vec_floato): Add builtin defines.
14607 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14608 Update the built-in documentation file for the new built-in
14611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14613 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14614 (mthumb): Mark as the negative of -marm.
14616 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14618 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14620 (ARM Options, -mtune): Document that this accepts the same
14621 extension options as -mcpu.
14622 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14626 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14627 permitted extensions.
14629 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14631 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14632 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14633 (armv8-m.main): Add option +nodsp.
14634 * config/arm/arm-cpu-cdata.h: Regenerated.
14636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14638 * config/arm/t-fuchsia: New file.
14639 * config.gcc (arm*-*-fuchsia*): Use it.
14641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14643 * config/arm/t-symbian: Rewrite for new option infrastructure.
14645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14647 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14648 (MULTILIB_REQUIRED): Likewise.
14650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14652 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14653 (MULTILIB_RESUE): Likewise.
14654 (MULTILIB_MATCHES): Likewise.
14655 (MULTLIB_REQUIRED): Likewise.
14657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14659 * config/arm/t-rtems: Rewrite for new option framework.
14661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14663 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14664 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14665 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14666 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14667 * config/arm/t-multilib: ... here.
14668 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14669 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14670 armv7-a and armv8*-a when A-profile libraries have not been built.
14671 * config/arm/t-rmprofile: Rewrite.
14673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14675 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14676 with a backslash. Remove the backslash after substituting unescaped
14678 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14682 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14683 the driver to pass through the default setting of -mfloat-abi.
14684 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14686 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14687 (all_feat_combs): New rule.
14688 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14690 * config/arm/t-aprofile: Rewrite.
14692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14694 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14695 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14696 fpu is not specified by the user/command-line.
14697 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14698 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14699 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14700 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14701 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14702 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14704 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14706 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14707 * config/arm/t-arm-elf: Rewritten.
14709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14711 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14712 have some floating-point instructions.
14713 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14714 (TARGET_MAYBE_HARD_FLOAT): New macro.
14715 * config/arm/arm-builtins.c (arm_init_builtins): Use
14716 TARGET_MAYBE_HARD_FLOAT.
14717 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14721 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14722 (configargs.h): Include it.
14723 (arm_print_hint_for_fpu_option): New function.
14724 (arm_parse_fpu_option): New function.
14725 (candidate_extension): New class.
14726 (arm_canon_for_multilib): New function.
14727 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14728 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14729 (ARCH_CANONICAL_SPECS): New macro.
14730 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14732 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14734 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14735 are set after handling multilib fragments. Set target_cpu_default2
14738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14740 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14742 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14743 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14745 (check_arch): Likewise.
14746 * config/arm/arm.c (arm_configure_build_target): Handle
14747 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14748 options in the default.
14750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14752 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14753 when an option is an alias of another.
14754 * config/arm/parsecpu.awk (optalias): New parser token.
14755 (gen_comm_data): Mark non-alias options as such. Emit entries
14756 for extension aliases.
14757 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14758 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14759 (armv6kz, armv6zk, armv6t2): Likewise.
14760 (armv7): Make vfpv3-d16 an alias.
14761 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14763 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14764 Sort in canonical order.
14765 (armv8-a): Sort in canonical order.
14766 (armv8.1-a, armv8.2-a): Likewise.
14767 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14769 (cortex-a9): Sort in canonical order.
14770 * config/arm/arm.c (selftests.h): Include it.
14771 (arm_test_cpu_arch_data): New function.
14772 (arm_run_self_tests): New function.
14773 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14774 (targetm): Move declaration to the end of the file.
14775 * arm-cpu-cdata.h: Regenerated.
14777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14779 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14780 call to target_mode_check describing the type of option passed.
14781 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14782 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14783 arm_parse_cpu_option_name to match parameters against list of
14785 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14786 arm_arch_core_flags data structure.
14787 * config/arm/arm-cpu_cdata.h: Regenerated.
14789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14791 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14793 (arm_print_hint_for_cpu_option): Likewise.
14794 (arm_print_hint_for_arch_option): Likewise.
14795 (arm_parse_cpu_option_name): Likewise.
14796 (arm_parse_arch_option_name): Likewise.
14797 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14798 of entries in the all_fpus list.
14799 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14800 (arm_parse_cpu_option_name): Declare.
14801 (arm_parse_arch_option_name): Declare.
14802 (arm_parse_option_features): Declare.
14803 (arm_intialize_isa): Declare.
14804 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14806 (gen_comm_data): ... here. Make definitions non-static.
14807 * config/arm/arm-cpu-data.h: Regenerated.
14808 * config/arm/arm-cpu-cdata.h: Regenerated.
14810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14812 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14813 (cpu_arch_extension): New structure.
14814 (cpu_arch_option, arch_option, cpu_option): New structures.
14815 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14816 architecture types.
14817 (gen_data): Generate new format data tables.
14818 * config/arm/arm.c (cpu_tune): New structure.
14819 (cpu_option, processors): Delete.
14820 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14821 (arm_print_hint_for_cpu_option): ... this and ...
14822 (arm_print_hint_for_arch_option): ... this.
14823 (arm_parse_arch_cpu_name): Delete. Replace with ...
14824 (arm_parse_cpu_option_name): ... this and ...
14825 (arm_parse_arch_option_name): ... this.
14826 (arm_unrecognized_feature): Change type of target parameter to
14828 (arm_parse_arch_cpu_features): Delete. Replace with ...
14829 (arm_parse_option_features): ... this.
14830 (arm_configure_build_target): Rework to use new configuration data
14832 (arm_print_tune_info): Rework for new configuration data tables.
14833 * config/arm/arm-cpu-data.h: Regenerated.
14834 * config/arm/arm-cpu.h: Regenerated.
14836 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14838 * Makefile.in (OBJS): Move sbitmap.o from here ...
14839 (OBJS-libcommon): ... to here.
14841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14843 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
14844 (ISA_ALL_CRYPTO): New macro.
14845 (ISA_ALL_SIMD): New macro
14846 (ISA_ALL_FP): New macro.
14847 * config/arm/arm.c (fpu_bitlist): Update initializer.
14848 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
14850 (arm9e): Add fpu. Add option for nofp
14851 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
14852 (arm926ej-s, arm1026ej-s): Likewise.
14853 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
14854 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
14855 neon-fp16, neon-vfpv4, nofp and nosimd.
14856 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
14857 (cortex-a8): Add fpu. Add option for nofp.
14858 (cortex-a9): Add fpu. Add options for nosimd and nofp.
14859 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
14860 (cortex-r4f): Add fpu.
14861 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
14862 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
14864 (cortex-r8): Likewise.
14865 (cortex-m4): Add fpu. Add option for nofp.
14866 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
14867 (cortex-a17.cortex-a7): Likewise.
14868 (cortex-a32): Add fpu. Add options for crypto and nofp.
14869 (cortex-a35, cortex-a53): Likewise.
14870 (cortex-a57): Add fpu. Add option for crypto.
14871 (cortex-a72, cortex-a73): Likewise.
14872 (exynos-m1): Likewise.
14873 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
14874 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
14875 (cortex-m33): Add fpu. Add option for nofp.
14876 * config/arm/arm-cpu-cdata.h: Regenerated
14877 * config/arm/arm-cpu-data.h: Regenerated.
14879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14881 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
14882 (armv5te, armv5tej): Likewise.
14883 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
14884 (armv7): Add options fp and vfpv3-d16.
14885 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
14886 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
14888 (armv7ve): Likewise.
14889 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
14890 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
14891 (armv8-a): Add nocrypto option.
14892 (armv8.1-a, armv8.2-a): Likewise.
14893 (armv8-m.main): add options fp, fp.dp and nofp.
14895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14897 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
14899 (armv8-a+crc): Delete.
14900 (armv8.1-a): Add options simd, crypto and nofp.
14901 (armv8.2-a): Add options fp16, simd, crypto and nofp.
14902 (armv8.2-a+fp16): Delete.
14903 (armv8-m.main): Add option dsp.
14904 (armv8-m.main+dsp): Delete.
14905 (cortex-a8): Add fpu. Add nofp option.
14906 (cortex-a9): Add fpu. Add nofp and nosimd options.
14907 * config/arm/parsecpu.awk (gen_data): Generate option tables and
14908 link to main cpu and architecture data structures.
14909 (gen_comm_data): Only put isa attributes from the main architecture
14911 (option): New statement for architecture and CPU entries.
14912 * arm.c (struct cpu_option): New structure.
14913 (struct processors): Add entry for options.
14914 (arm_unrecognized_feature): New function.
14915 (arm_parse_arch_cpu_name): Ignore any characters after the first
14917 (arm_parse_arch_cpu_feature): New function.
14918 (arm_configure_build_target): Separate out any CPU and architecture
14919 features and parse separately. Don't error out if -mfpu=auto is
14920 used with only an architecture string.
14921 (arm_print_asm_arch_directives): New function.
14922 (arm_file_start): Call it.
14923 * config/arm/arm-cpu-cdata.h: Regenerated.
14924 * config/arm/arm-cpu-data.h: Likewise.
14925 * config/arm/arm-tables.opt: Likewise.
14927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14929 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
14930 assembler when it is not -mfpu=auto.
14932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14934 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
14935 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
14936 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
14937 (ASM_CPU_SPEC): Rewrite.
14938 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
14939 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
14940 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
14941 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
14942 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
14943 copied string is NUL-terminated. Also strip any characters prefixed
14945 (arm_rewrite_selected_arch): New function.
14946 (arm_rewrite_march): New function.
14948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14950 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
14951 (x_arm_cpu_string, x_arm_tune_string): Likewise.
14952 (march, mcpu, mtune): Convert to string-based options.
14953 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
14954 (arm_parse_arch_cpu_name): New function.
14955 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
14956 identify selected architecture or CPU.
14957 (arm_option_save): New function.
14958 (TARGET_OPTION_SAVE): Redefine.
14959 (arm_option_restore): Restore string options.
14960 (arm_option_print): Print string options.
14962 2017-06-16 Martin Sebor <msebor@redhat.com>
14964 PR tree-optimization/80933
14965 PR tree-optimization/80934
14966 * builtins.c (fold_builtin_3): Do not handle bcmp here.
14967 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
14968 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
14969 (gimple_fold_builtin): Call them.
14971 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14973 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
14974 as unlikely; update profile.
14976 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14978 * predict.c (force_edge_cold): Handle declaring edges impossible
14981 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14983 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
14985 (try_unroll_loop_completely): Fix reporting.
14987 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
14989 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
14991 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
14994 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
14995 if given a non-constant argument for an intrinsic which requires a
14998 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15000 * profile.c (compare_freqs): New function.
15001 (branch_prob): Sort edge list.
15002 (find_spanning_tree): Assume that the list is priority sorted.
15004 2017-06-16 Richard Biener <rguenther@suse.de>
15006 PR tree-optimization/81090
15007 * passes.def (pass_record_bounds): Remove.
15008 * tree-pass.h (make_pass_record_bounds): Likewise.
15009 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15010 make_pass_record_bounds): Likewise.
15011 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15012 not free niter estimates at the beginning but at the end.
15013 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15015 2017-06-16 Richard Biener <rguenther@suse.de>
15017 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15018 initializer to workaround ICE in host GCC 4.8.
15020 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15022 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15024 (clone_inlined_nodes): Update.
15026 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15028 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15029 prefetch settings, and enable prefetching by default at -O3.
15031 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15033 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15034 Set flag_prefetch_loop_arrays according to tuning data.
15036 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15038 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15039 New tune structure.
15040 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15041 [Unrelated to main purpose of the patch] Place the pointer field last
15042 to enable type checking errors when tune structure are wrongly merged.
15043 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15044 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15045 (thunderx2t99_prefetch_tune): New tune constants.
15046 (tune_params *_tunings): Update all tunings (no functional change).
15047 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15048 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15049 from tunings structures.
15051 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15054 * ubsan.c (instrument_null): Add T argument, use it instead
15055 of computing it based on IS_LHS.
15056 (instrument_object_size): Likewise.
15057 (pass_ubsan::execute): Adjust instrument_null and
15058 instrument_object_size callers to pass gimple_get_lhs or
15059 gimple_assign_rhs1 result to it. Use instrument_null instead of
15060 calling get_base_address and instrument_mem_ref. Handle
15061 aggregate call arguments for object-size sanitization.
15063 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15065 PR tree-optimization/81089
15066 * tree-vrp.c (is_masked_range_test): Validate operands of
15069 2017-06-15 Martin Sebor <msebor@redhat.com>
15072 * dumpfile.c (dump_register): Avoid calling memset to initialize
15073 a class with a default ctor.
15074 * gcc.c (struct compiler): Remove const qualification.
15075 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15076 * hash-table.h: Ditto.
15077 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15079 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15080 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15082 * params.h (struct param_info): Make struct members non-const.
15083 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15084 with default initialization.
15085 * vec.h (vec_copy_construct, vec_default_construct): New helper
15087 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15088 with vec_copy_construct.
15089 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15090 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15091 * doc/invoke.texi (-Wclass-memaccess): Document.
15093 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15095 * emit-rtl.h (is_leaf): Update comment about local
15096 register allocator.
15098 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15101 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15102 for a variable to have a section before checking if the section has a
15104 Set section to.persistent if persistent attribute is set.
15105 Warn if .persistent attribute is used on an automatic variable.
15107 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15109 PR rtl-optimization/80474
15110 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15112 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15114 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15117 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15119 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15120 (casesi): Emit bounds checking as RTL.
15121 (casesi_internal_mips16_<mode>): Remove bounds checking.
15123 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15125 * config/xtensa/xtensa.c (xtensa_option_override): Append
15126 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15127 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15128 xtensa_doloop_hooks): Define unconditionally.
15129 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15131 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15132 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15133 for it in xtensa_option_override.
15134 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15135 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15137 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15139 * doc/cppopts.texi: Document '-' special value to -MF.
15141 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15143 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15144 (cortex_a53_fconst): Likewise.
15145 (cortex_a53_fpmul): Likewise.
15146 (cortex_a53_f_load_64): Likewise.
15147 (cortex_a53_f_load_many): Likewise.
15148 (cortex_a53_advsimd_alu): Likewise.
15149 (cortex_a53_advsimd_alu_q): Likewise.
15150 (cortex_a53_advsimd_mul): Likewise.
15151 (cortex_a53_advsimd_mul_q): Likewise.
15152 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15153 Add missing fmul, r2f_cvt and fconst cases.
15155 2017-06-14 Richard Biener <rguenther@suse.de>
15157 PR middle-end/81088
15158 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15160 (fold_binary_loc): When associating do not treat pre-existing
15161 TREE_OVERFLOW on literal constants as a reason to allow
15162 TREE_OVERFLOW on associated literal constants.
15164 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15166 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15167 (MASK_FEATURES): New macro.
15168 * config/sparc/sparc.c (sparc_option_override): Remove the special
15169 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15171 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15173 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15174 a division of 0 if non-call exceptions are enabled.
15176 2017-06-14 Andrew Pinski <apinski@cavium.com>
15177 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15180 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15181 Improve vector initialization code gen for only variable case.
15183 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15185 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15187 2017-06-14 Richard Biener <rguenther@suse.de>
15189 PR tree-optimization/81083
15190 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15193 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15195 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15196 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15197 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15198 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15199 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15200 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15202 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15204 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15205 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15207 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15209 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15211 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15213 * config/rs6000/t-rtems: Don't handle SPE.
15215 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15217 * config/rs6000/t-linux: Don't handle SPE.
15219 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15221 * config/rs6000/eabispe.h: Delete file.
15223 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15225 * config/rs6000/t-spe: Delete file.
15227 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15229 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15230 (rs6000_legitimate_offset_address_p): Return false for anything in
15231 V2SImode or V2SFmode.
15233 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15235 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15236 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15237 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15239 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15240 (rs6000_legitimate_offset_address_p): Ditto.
15241 (rs6000_emit_move): Ditto.
15242 (rs6000_init_builtins): Remove V4HI_type_node.
15244 2017-06-13 Martin Liska <mliska@suse.cz>
15247 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15248 (gate_asan): Likewise.
15249 * asan.h (asan_no_sanitize_address_p): Remove the function.
15250 (sanitize_flags_p): New function.
15251 * builtins.def: Fix coding style.
15252 * common.opt: Use renamed enum value.
15253 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15254 * doc/extend.texi: Document no_sanitize attribute.
15255 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15256 to SANITIZE_UNDEFINED_NONDEFAULT.
15257 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15258 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15259 Use sanitize_flags_p.
15260 * gimplify.c (gimplify_function_tree): Likewise.
15261 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15262 * opts.c (parse_no_sanitize_attribute): New function.
15263 (common_handle_option): Use renamed enum value.
15264 * opts.h (parse_no_sanitize_attribute): Declare.
15265 * tree.c (sanitize_flags_p): New function.
15266 * tree.h: Declared here.
15267 * tsan.c: Use sanitize_flags_p.
15268 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15269 (instrument_mem_ref): Likewise.
15270 (instrument_bool_enum_load): Likewise.
15271 (do_ubsan_in_current_function): Remove the function.
15272 (pass_ubsan::execute): Use sanitize_flags_p.
15273 * ubsan.h: Remove do_ubsan_in_current_function
15274 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15275 (dump_function_to_file): Use it here.
15277 2017-06-13 Martin Jambor <mjambor@suse.cz>
15279 PR tree-optimization/80803
15280 PR tree-optimization/81063
15281 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15282 (propagate_subaccesses_across_link): Enqueue subtree whenever
15283 necessary instead of relying on the caller.
15285 2017-06-13 Martin Jambor <mjambor@suse.cz>
15287 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15288 that have a first_link.
15289 (sort_and_splice_var_accesses): Do not check first_link before
15291 (subtree_mark_written_and_enqueue): Likewise.
15292 (propagate_all_subaccesses): Likewise and do not stop at first
15293 parent with a first_link.
15295 2017-06-13 Martin Jambor <mjambor@suse.cz>
15297 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15300 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15302 * match.pd: New pattern.
15304 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15306 * tree-vrp.c (is_masked_range_test): New function.
15307 (register_edge_assert_for): Determine ranges for
15310 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15312 PR tree-optimization/67328
15313 * fold-const.c (maskable_range_p): New function.
15314 (build_range_check): Generate bittests if possible.
15316 2017-06-13 Martin Liska <mliska@suse.cz>
15318 * gimple-pretty-print.c (dump_probability): Add new argument.
15319 (dump_edge_probability): Dump both probability and count.
15320 (dump_gimple_label): Likewise.
15321 (dump_gimple_bb_header): Likewise.
15323 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15326 * config/avr/avr-devices.c: Fix indentation.
15327 * config/avr/gen-avr-mmcu-specs.c: Dito.
15329 2017-06-13 Richard Biener <rguenther@suse.de>
15331 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15332 instead get vector type from stmt_info.
15333 (vectorizable_reduction): Adjust. Remove dead code.
15335 2017-06-13 Richard Biener <rguenther@suse.de>
15337 PR middle-end/81065
15338 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15339 case of C * (x * C2 + C3).
15340 (fold_addr_of_array_ref_difference): Properly fold index difference.
15342 2017-06-12 David S. Miller <davem@davemloft.net>
15345 * config/sparc/sparc.md (return expander): Emit frame blockage if
15346 function uses alloca.
15348 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15350 * combine.c (make_field_assignment): Check len rather than the mode
15351 precision when calling force_to_mode.
15353 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15355 Support multilibs and devices that see flash in RAM address range.
15358 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15359 (avr_mcu_t) <flash_pm_offset>: New field.
15360 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15361 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15362 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15363 (AVR_TINY_PM_OFFSET): Remove macro.
15364 * config/avr/avr.opt (-mshort-calls): New option.
15365 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15366 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15367 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15368 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15369 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15370 instead of avr_arch->have_jmp_call.
15371 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15372 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15373 avr_arch->flash_pm_offset to define.
15374 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15375 new field flash_pm_offset. Add entry for avrxmega3.
15376 (avr_texinfo): Add entry for avrxmega3.
15377 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15378 attiny212, attiny214,
15379 attiny412, attiny414, attiny416, attiny417,
15380 attiny814, attiny816, attiny817,
15381 attiny1614, attiny1616, attiny1617,
15382 attiny3214, attiny3216, attiny3217.
15383 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15384 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15385 (avr_print_operand_address) [AVR_TINY]: Same.
15386 (avr_asm_init_sections) <readonly_data_section>: Only patch
15387 callback if avr_arch->flash_pm_offset = 0.
15388 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15389 for rodata if avr_arch->flash_pm_offset != 0.
15390 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15391 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15392 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15393 (m_options): Append opt_rcall.
15394 (m_dirnames): Append dir_rcall.
15395 * config/avr/t-multilib: Regenerate.
15397 * configure.ac [target=avr]: Check whether avrxmega3 default
15398 linker description file works as needed.
15399 * configure: Regenerate.
15400 * doc/avr-mmcu.texi: Regenerate.
15401 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15402 <__AVR_ARCH__>: Document avrxmega3 and 103.
15403 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15404 <__AVR_SHORT_CALLS__>: Document it.
15405 <__AVR_PM_BASE_ADDRESS__>: Document it.
15406 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15407 (AVR Variable Attributes) <progmem>: Document this is
15408 not needed for avrxmega3.
15409 (AVR Named Address Spaces) <__flash>: Dito.
15411 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15413 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15415 2017-06-12 Doug Rupp <rupp@adacore.com>
15417 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15418 Append vxworks-stdint.h to the tm_file list.
15419 * config/vxworks-stdint.h: New file.
15421 2017-06-12 Martin Liska <mliska@suse.cz>
15423 PR tree-optimization/81041
15424 * tree-profile.c (gimple_gen_ic_func_profiler):
15425 Create an extra BB in profile-generate
15426 (gimple_gen_time_profiler): Likewise.
15428 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15430 PR tree-optimization/81003
15431 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15432 (update_range_test): Use it instead of force_gimple_operand_gsi.
15434 2017-06-12 Richard Biener <rguenther@suse.de>
15436 PR tree-optimization/81053
15437 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15438 with backedge value not defined in loop. Simplify def stmt
15441 2017-06-11 Tom de Vries <tom@codesourcery.com>
15444 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15446 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15447 nvptx_cannot_force_const_mem.
15449 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15451 * opts.c (finish_options): Move test for flag_split_stack after
15452 it has been initialized.
15454 2017-06-11 Jason Merrill <jason@redhat.com>
15456 * tree.h (id_equal): New.
15457 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15458 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15459 instead of strcmp of IDENTIFIER_POINTER.
15461 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15463 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15464 (mark_all_inlined_calls_cdtor): Fix formating.
15465 (inline_transform): Rescale profile before inlining.
15467 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15469 * cgraph.h (cgraph_edge::clone): Update prototype.
15470 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15471 (cgraph_node::create_clone): Update.
15472 (cgraph_node::create_version_clone): Update.
15473 * tree-inline.c (copy_bb): Update.
15474 (expand_call_inline): Update.
15476 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15478 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15479 factored out from ...
15480 (rs6000_emit_prologue): ... here.
15482 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15484 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15485 factored out from ...
15486 (rs6000_emit_prologue): ... here.
15488 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15490 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15492 (handle_missing_profiles): Fix computation of tp_first_run.
15493 (counts_to_freqs): Do not touch freqs when count is 0.
15495 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15497 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15500 2017-06-10 Tom de Vries <tom@codesourcery.com>
15502 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15503 attributes): Document signal effective target.
15505 2017-06-10 Tom de Vries <tom@codesourcery.com>
15507 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15508 Document effective target stack_size.
15510 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15512 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15513 to the edit_context if they can be auto-applied.
15515 2017-06-9 Ian Lance Taylor <iant@golang.org>
15517 * opts.c (finish_options): If -fsplit-stack, disable implicit
15518 -forder-blocks-and-partition.
15519 * doc/invoke.texi (Optimize Options): Document that when using
15520 -fsplit-stack -forder-blocks-and-partition is not implicitly
15523 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15525 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15526 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15527 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15528 * builtins.def (abort, trap, unreachable): Declare cold.
15529 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15530 * tree-core.h (ECF_COLD): New.
15531 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15532 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15534 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15536 * predict.c (unlikely_executed_stmt_p): Cleanup.
15538 2017-06-09 Richard Biener <rguenther@suse.de>
15540 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15541 model if the ref is always written to.
15543 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15545 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15547 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15549 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15552 2017-06-09 Tom de Vries <tom@codesourcery.com>
15555 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15556 "target cannot support label values" when encountering LABEL_REF.
15558 2017-06-09 Martin Liska <mliska@suse.cz>
15560 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15561 (gimple_gen_ic_func_profiler): Emit direct comparison
15562 of __gcov_indirect_call_callee with NULL.
15563 (gimple_gen_time_profiler): Change probability from
15564 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15566 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15568 * profile.c (edge_gcov_counts): Turn to pointer.
15569 (compute_branch_probabilities, compute_branch_probabilities): Update.
15570 (branch_prob): Do not clear edge_gcov_count.
15571 * profile.h (edge_gcov_counts): Turn to pointer.
15572 (edge_gcov_count): Update.
15574 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15576 * gimple.h (gimple_check_failed): Mark cold.
15578 2017-06-09 Richard Biener <rguenther@suse.de>
15580 PR tree-optimization/66623
15581 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15582 refactor check_reduction into two parts, properly computing
15583 whether we have to check reduction validity for outer loop
15586 2017-06-09 Richard Biener <rguenther@suse.de>
15588 PR tree-optimization/79483
15589 * graphite-scop-detection.c (order): New global.
15590 (get_order): Compute bb to order mapping that satisfies code
15591 generation constraints.
15592 (cmp_pbbs): New helper.
15593 (build_scops): Start domwalk at entry block, sort generated
15596 2017-06-09 Richard Biener <rguenther@suse.de>
15598 PR middle-end/81007
15599 * ipa-polymorphic-call.c
15600 (ipa_polymorphic_call_context::restrict_to_inner_class):
15601 Skip FIELD_DECLs with error_mark_node type.
15602 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15605 2017-06-09 Martin Liska <mliska@suse.cz>
15607 * predict.c (struct branch_predictor): New struct.
15608 (test_prediction_value_range): New test.
15609 (predict_c_tests): New function.
15610 * selftest-run-tests.c (selftest::run_tests): Run the function.
15611 * selftest.h: Declare new tests.
15613 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15616 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15617 gen_add3_insn did not fail.
15618 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15619 r0, construct that number in a temporary reg and add that reg to r0.
15620 If asked to put the result in r0 as well, fail.
15622 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15624 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15625 for early expansion of vec_eqv.
15627 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15629 PR middle-end/81005
15630 * ubsan.c (instrument_null): Avoid pointless code temporary.
15631 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15633 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15637 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15640 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15643 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15645 * predict.c (unlikely_executed_edge_p): Move ahead.
15646 (probably_never_executed_edge_p): Use it.
15648 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15650 PR middle-end/79988
15651 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15652 gimple_call_builtin_p call.
15654 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15656 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15657 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15658 rtl_check_failed_type2, rtl_check_failed_code1,
15659 rtl_check_failed_code2, rtl_check_failed_code_mode,
15660 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15661 rtvec_check_failed_bounds, rtl_check_failed_flag,
15662 _fatal_insn_not_found, _fatal_insn): Likewise.
15663 * tree.h (tree_contains_struct_check_failed,
15664 tree_check_failed, tree_not_check_failed,
15665 tree_class_check_failed, tree_range_check_failed,
15666 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15667 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15668 tree_operand_check_failed, omp_clause_check_failed,
15669 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15672 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15674 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15675 flag_branch_probabilities.
15676 * ipa-inline.c (edge_badness): Likewise.
15677 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15678 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15679 * predict.c (maybe_hot_frequency_p): Likewise.
15680 (probably_never_executed): Likewise.
15681 * sched-ebb.c (schedule_ebbs): Likewise.
15682 * sched-rgn.c (find_single_block_region): Likewise.
15683 * tracer.c (tail_duplicate): Likewise.
15685 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15687 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15688 longer requires x_flag_profile_use.
15690 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15692 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15693 instead of flag_reorder_blocks_and_partition.
15694 * dbxout.c (dbxout_function_end): Likewise.
15695 * dwarf2out.c (gen_subprogram_die): Likewise.
15696 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15697 * hw-doloop.c (reorg_loops): Likewise.
15698 * varasm.c (assemble_start_function,
15699 assemble_end_function): Likewise.
15700 (decide_function_section): Do not check for
15701 flag_reorder_blocks_and_partition.
15703 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15705 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15707 (chkp_get_hard_register_fake_addr_expr): Ditto.
15708 (chkp_build_addr_expr): Add check for hard reg case.
15709 (chkp_parse_array_and_component_ref): Ditto.
15710 (chkp_find_bounds_1): Ditto.
15711 (chkp_process_stmt): Don't generate bounds store for
15714 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15716 * predict.c (maybe_hot_bb_p): Do not check profile status.
15717 (maybe_hot_edge_p): Likewise.
15718 (probably_never_executed): Check for zero counts even if profile
15720 (unlikely_executed_edge_p): New function.
15721 (unlikely_executed_stmt_p): New function.
15722 (unlikely_executed_bb_p): New function.
15723 (set_even_probabilities): Use unlikely predicates.
15724 (combine_predictions_for_bb): Likewise.
15725 (predict_paths_for_bb): Likewise.
15726 (predict_paths_leading_to_edge): Likewise.
15727 (determine_unlikely_bbs): New function.
15728 (estimate_bb_frequencies): Use it.
15729 (compute_function_frequency): Use zero counts even if profile is
15731 * profile-count.h: Fix typo.
15733 2017-08-08 Julia Koval <julia.koval@intel.com>
15735 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15736 _mm512_mask_cvtsepi16_storeu_epi8,
15737 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15738 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15739 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15740 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15741 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15742 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15743 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15744 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15745 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15746 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15747 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15748 __builtin_ia32_pmovuswb256mem_mask,
15749 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15750 __builtin_ia32_pmovwb512mem_mask): New builtins.
15752 2017-08-08 Julia Koval <julia.koval@intel.com>
15754 PR target/73350,80862
15755 * config/i386/subst.md (round): Fix round pattern.
15756 * config/i386/i386.c (ix86_erase_embedded_rounding):
15757 Fix erasing rounding for the fixed pattern.
15759 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15761 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15763 2017-06-08 Martin Liska <mliska@suse.cz>
15765 PR gcov-profile/80911
15766 * gcov.c (block_info::block_info): New constructor.
15768 2017-06-07 Carl Love <cel@us.ibm.com>
15770 * config/rs6000/rs6000-c: The return type of the following
15771 built-in functions was implemented as int not long long. Fix sign
15772 of return value for the unsigned version of vec_mulo and vec_mule.
15773 vector unsigned long long vec_bperm (vector unsigned long long,
15774 vector unsigned char)
15775 vector signed long long vec_mule (vector signed int,
15777 vector unsigned long long vec_mule (vector unsigned int,
15778 vector unsigned int)
15779 vector signed long long vec_mulo (vector signed int,
15781 vector unsigned long long vec_mulo (vector unsigned int,
15782 vector unsigned int)
15783 * doc/extend.texi: Fix the documentation for the built-in
15786 2017-06-07 Carl Love <cel@us.ibm.com>
15789 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15792 2017-06-07 Carl Love <cel@us.ibm.com>
15794 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15795 support, Generate doublehv for signed int/float for BE case only.
15797 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15799 * doc/invoke.texi (mcx16): Rewrite.
15801 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15803 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15804 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15805 *mov<mode>_softfloat, and an anonymous splitter): Use
15806 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15810 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15812 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15813 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15814 (rs6000_debug_reg_global): Adjust.
15815 (rs6000_init_hard_regno_mode_ok): Adjust.
15816 (rs6000_dbx_register_number): Adjust.
15817 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15818 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15819 Remove SPE_ACC and SPEFSCR.
15820 (REG_ALLOC_ORDER): Ditto.
15821 (FRAME_POINTER_REGNUM): Change to 111.
15822 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15823 (REG_CLASS_NAMES): Ditto.
15824 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15825 (REGISTER_NAMES): Ditto.
15826 (ADDITIONAL_REG_NAMES): Ditto.
15827 (rs6000_reg_names): Ditto.
15828 * config/rs6000/rs6000.md: Renumber some register number
15831 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15833 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15835 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15836 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15838 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15839 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15840 Delete the SPE high registers.
15841 (REG_ALLOC_ORDER): Ditto.
15842 (enum reg_class): Remove SPE_HIGH_REGS.
15843 (REG_CLASS_NAMES): Ditto.
15844 (REG_CLASS_CONTENTS): Delete the SPE high registers.
15845 (REGISTER_NAMES): Ditto.
15846 (rs6000_reg_names): Ditto.
15847 * doc/tm.texi.in: Remove SPE as example.
15848 * doc/tm.texi: Regenerate.
15850 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15852 * config/rs6000/8540.md (ppc8540_brinc): Delete.
15853 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
15854 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
15855 * config/rs6000/rs6000.md (type): Remove "brinc".
15857 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15859 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
15860 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
15861 * config/rs6000/linuxspe.h: Delete file.
15862 * config/rs6000/rs6000.md: Don't include spe.md.
15863 * config/rs6000/spe.h: Delete file.
15864 * config/rs6000/spe.md: Delete file.
15865 * config/rs6000/t-rs6000: Remove spe.md.
15867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15869 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
15870 (reg_or_none500mem_operand): Delete.
15871 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
15872 instead of reg_or_none500mem_operand.
15874 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15876 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
15877 handling of SPE flags.
15878 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
15880 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15882 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15884 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
15885 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
15886 paired_divv2sf3): Similar.
15887 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
15888 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
15889 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
15891 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
15892 Rename the paired_* instruction patterns.
15893 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
15895 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
15896 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
15897 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
15898 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
15899 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
15900 PAIRED_VECTOR_MODE.
15901 (struct machine_function): Delete field spe_insn_chain_scanned_p.
15902 (spe_func_has_64bit_regs_p): Delete.
15903 (spe_expand_predicate_builtin): Delete.
15904 (spe_expand_evsel_builtin): Delete.
15905 (TARGET_DWARF_REGISTER_SPAN): Do not define.
15906 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
15907 (invalid_e500_subreg): Delete.
15908 (rs6000_legitimize_address): Always force_reg op2 as well, for
15909 paired single memory accesses.
15910 (rs6000_member_type_forces_blk): Delete.
15911 (rs6000_spe_function_arg): Delete.
15912 (rs6000_expand_unop_builtin): Delete SPE handling.
15913 (rs6000_expand_binop_builtin): Ditto.
15914 (spe_expand_stv_builtin): Delete.
15915 (bdesc_2arg_spe): Delete.
15916 (spe_expand_builtin): Delete.
15917 (spe_expand_predicate_builtin): Delete.
15918 (spe_expand_evsel_builtin): Delete.
15919 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
15920 (spe_init_builtins): Delete.
15921 (spe_func_has_64bit_regs_p): Delete.
15922 (savres_routine_name): Delete "info" parameter. Adjust callers.
15923 (rs6000_emit_stack_reset): Ditto.
15924 (rs6000_dwarf_register_span): Delete.
15925 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
15926 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
15927 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
15929 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
15931 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
15932 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
15933 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
15934 mulv2sf3, divv2sf3): Delete expanders.
15936 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15938 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
15940 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15942 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
15943 * config/rs6000/rs6000.c: Ditto.
15945 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15947 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
15948 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
15949 comparison_operator.
15951 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15953 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
15954 * config/rs6000/rs6000.opt: Ditto.
15955 * config/rs6000/t-rtems: Ditto.
15957 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15959 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
15960 TARGET_E500_SINGLE by 0, simplify.
15961 * config/rs6000/rs6000.c: Ditto.
15962 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
15963 (spe_build_register_parallel): Delete.
15964 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
15965 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
15966 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
15967 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
15968 (E500_CONVERT): Delete.
15969 * config/rs6000/spe.md: Remove many patterns and all define_constants.
15971 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15973 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
15974 * config/rs6000/dfp.md: Ditto.
15975 (negdd2, *negdd2_fpr): Merge.
15976 (absdd2, *absdd2_fpr): Merge.
15977 (negtd2, *negtd2_fpr): Merge.
15978 (abstd2, *abstd2_fpr): Merge.
15979 * config/rs6000/e500.h: Delete file.
15980 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
15981 TARGET_FPRS by 1 and simplify.
15982 * config/rs6000/rs6000-c.c: Ditto.
15983 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
15984 TARGET_DF_SPE by 0.
15985 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
15987 * config/rs6000/rs6000.md: Ditto.
15988 (floatdidf2, *floatdidf2_fpr): Merge.
15989 (move_from_CR_gt_bit): Delete.
15990 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
15991 (E500_CR_IOR_COMPARE): Delete.
15992 (All patterns that require !TARGET_FPRS): Delete.
15993 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
15995 2017-06-07 Bin Cheng <bin.cheng@arm.com>
15997 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
15999 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16001 * graphds.c (add_edge): Intitialize edge's attached data.
16002 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16003 pointer parameter. Call pointed function on each edge during
16004 graph traversing. Skip traversing the edge when the function
16006 (graphds_dfs, graphds_scc): Ditto.
16007 (for_each_edge): New parameter. Pass the new parameter to callback
16009 * graphds.h (skip_edge_callback): New function pointer type.
16010 (graphds_dfs, graphds_scc): New function pointer parameter.
16011 (graphds_edge_callback, for_each_edge): New parameter.
16013 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16015 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16016 out code checking if runtime alias check is possible to below ...
16017 Call the new function.
16018 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16019 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16021 2017-06-07 Marek Polacek <polacek@redhat.com>
16024 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16025 TYPE_OVERFLOW_WRAPS check.
16027 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16029 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16030 if versioning is required.
16031 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16032 peeling with the check for versioning.
16034 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16036 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16037 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16038 Set true to new parameter if new ssa variable is defined.
16039 (vect_gen_vector_loop_niters): Refactor. Set range information
16040 for the new vector loop bound variable.
16041 (vect_do_peeling): Ditto.
16043 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16045 * tree-affine.c (ssa.h): Include header file.
16046 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16047 has wrapping overflow behavior.
16049 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16051 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16053 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16055 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16056 (tree_to_aff_combination): ... here.
16058 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16060 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16061 reg_pressure model function.
16062 (ivopts_global_cost_for_size): Delete.
16063 (determine_set_costs, iv_ca_recount_cost): Call new model function
16064 ivopts_estimate_reg_pressure.
16066 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16068 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16069 expensive than udiv. Remove floating point cases from mod.
16071 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16073 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16074 Increase idiv cost.
16076 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16078 * config/aarch64/aarch64.md
16079 (copysignsf3): Fix mask generation.
16081 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16083 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16085 (class dump_manager): Add register_dumps method.
16086 * dumpfile.c: Include langhooks.h.
16087 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16088 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16089 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16090 (dump_manager::dump_register): Start with 512 entries instead of 32.
16091 (dump_manager::register_dumps): New method.
16092 * toplev.c (general_init): Instead of invoking register_dumps
16093 langhook, invoke register_dumps method on the dump manager.
16094 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16097 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16099 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16100 Say that # requires an associated define_split to exist, and that
16101 the define_split must be suitable for use after register allocation.
16103 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16105 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16106 (compute_outgoing_frequencies): Also initialize zero counts.
16107 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16108 around loops; preserve more of profile when nothing changes.
16110 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16112 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16114 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16115 * config/arm/arm-cpu-cdata.h: Regenerate.
16116 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16117 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16118 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16119 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16120 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16122 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16124 * config/arm/t-rmprofile: Likewise.
16125 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16127 2017-06-06 David S. Miller <davem@davemloft.net>
16130 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16131 blockage if function uses alloca.
16133 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16135 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16136 New "uid" fields to hold pretty-print IDs of group and ref.
16137 Memory references are now identified as <group_id>:<ref_id>
16138 instead of using [random] addresses.
16139 (dump_mem_details): Simplify, no functional change.
16140 (dump_mem_ref): Simplify and make output more concise.
16141 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16142 (find_or_create_group): Initialize group uid.
16143 (record_ref): Initialize ref uid. Improve debug output.
16144 (prune_group_by_reuse, should_issue_prefetch_p,)
16145 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16146 (mark_nontemporal_store, determine_loop_nest_reuse):
16147 Improve debug output.
16149 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16151 * dbgcnt.def (prefetch): New debug counter.
16152 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16153 (schedule_prefetches): Stop issueing prefetches if debug counter
16156 2017-06-06 Tom de Vries <tom@codesourcery.com>
16158 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16159 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16161 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16163 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16164 Use aarch64_reg_or_zero predicate for operand 4.
16165 (aarch64_compare_and_swap<mode> define_insn_and_split):
16166 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16167 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16169 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16171 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16172 (arm_compute_save_core_reg_mask): This.
16173 (thumb1_compute_save_reg_mask): Rename into ...
16174 (thumb1_compute_save_core_reg_mask): This.
16175 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16176 (arm_compute_frame_layout): Likewise.
16178 2017-06-06 Richard Biener <rguenther@suse.de>
16180 PR tree-optimization/80974
16181 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16182 keep or clear leaders SSA info.
16184 2017-06-06 Tom de Vries <tom@codesourcery.com>
16186 * config/nvptx/nvptx.c (split_mode_p): New function.
16187 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16189 2017-06-06 Tom de Vries <tom@codesourcery.com>
16191 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16193 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16196 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16199 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16201 * shrink-wrap.c (handle_simple_exit): Update profile.
16202 (try_shrink_wrapping): Upate profile.
16204 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16206 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16207 (tree_guess_outgoing_edge_probabilities): New.
16208 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16209 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16211 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16213 * ipa-split.c (split_function): Initialize return bb profile.
16215 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16217 * profile.c (compute_branch_probabilities): Also initialize
16218 EXIT_BLOCK profile.
16220 2017-06-06 Richard Biener <rguenther@suse.de>
16222 PR tree-optimization/80928
16223 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16224 (vect_analyze_loop_operations): Properly guard analysis for
16226 (vect_transform_loop): Likewise.
16227 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16228 (vect_model_induction_cost): Do not cost for pure SLP.
16229 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16230 of induction in inner loop vectorization.
16231 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16232 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16233 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16235 (vect_analyze_slp_cost_1): Cost induction.
16236 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16237 (vect_get_slp_vect_defs): Likewise.
16238 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16239 (vect_transform_stmt): Handle SLP reductions.
16240 * tree-vectorizer.h (vectorizable_induction): Adjust.
16242 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16244 * config/rs6000/rs6000.c (make_resolver_func): Update
16245 init_lowered_empty_function call.
16247 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16249 * doc/invoke.texi: Document the -fprofile-abs-path option.
16250 * common.opt (fprofile-abs-path): New option.
16251 * gcov-io.h (gcov_write_filename): Declare.
16252 * gcov-io.c (gcov_write_filename): New function.
16253 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16254 * profile.c (output_location): Likewise.
16256 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16258 * shring-wrap.c: Revert accidental commit.
16260 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16262 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16264 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16266 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16268 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16269 profile in callgraph edge.
16270 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16271 (apply_scale): Likewise.
16272 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16273 Also scale profile when inlining function with zero profile.
16274 (initialize_cfun): Update exit block profile even when it is zero.
16275 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16276 when profile is read.
16278 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16280 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16281 (CLONE_*): New constants to define the processors we can generate
16282 code for with the target_clone attribute.
16283 (rs6000_clone_map): New array to identify which clone processors
16284 the current program is running on.
16285 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16286 target_clone attribute.
16287 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16288 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16289 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16290 (cpu_expand_builtin): Add support for target_clone attribute.
16291 (rs6000_valid_attribute_p): Allow "default" attribute.
16292 (get_decl_name): New debug function to simplify printing the
16293 current function name in debugging statements.
16294 (rs6000_clone_priority): New functions to support the target_clone
16295 attribute, and be able to generate code to switch between ISA 2.05
16296 through ISA 3.0 (power6 through power9).
16297 (rs6000_compare_version_priority): Likewise.
16298 (rs6000_get_function_versions_dispatcher): Likewise.
16299 (make_resolver_func): Likewise.
16300 (add_condition_to_bb): Likewise.
16301 (dispatch_function_versions): Likewise.
16302 (rs6000_generate_version_dispatcher_body): Likewise.
16303 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16304 (fusion_gpr_load_p): Fix a spacing issue.
16305 * doc/extend.texi (Common Function Attributes): Document that the
16306 PowerPC supports the target_clone attribute.
16308 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16310 * config/arm/arm.h: explain F symbol found in description of ARM
16311 register allocation in its legend.
16313 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16315 * config/mips/frame-header-opt.c: Include profile-count.h.
16316 * config/riscv/riscv.c: Include profile-count.h
16318 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16320 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16322 (sm_set_flag_if_changed): Add bbs field.
16323 (execute_sm_if_changed_flag_set): Pass BBS.
16324 (execute_sm): Update.
16326 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16328 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16331 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16333 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16334 (peephole2): New peephole2 to emit the above.
16335 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16337 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16339 * config/aarch64/aarch64.c (define_peephole2 above
16340 *sub_<shift>_<mode>): New peephole.
16342 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16344 * config/i386/i386.c (make_resolver_func): Update.
16345 * Makefile.in: Add profile-count.h and profile-count.o
16346 * auto-profile.c (afdo_indirect_call): Update to new API.
16347 (afdo_set_bb_count): Update.
16348 (afdo_propagate_edge): Update.
16349 (afdo_propagate_circuit): Update.
16350 (afdo_calculate_branch_prob): Update.
16351 (afdo_annotate_cfg): Update.
16352 * basic-block.h: Include profile-count.h
16353 (struct edge_def): Turn count to profile_count.
16354 (struct basic_block_def): Likewie.
16355 (REG_BR_PROB_BASE): Move to profile-count.h
16356 (RDIV): Move to profile-count.h
16357 * bb-reorder.c (max_entry_count): Turn to profile_count.
16358 (find_traces): Update.
16359 (rotate_loop):Update.
16360 (connect_traces):Update.
16361 (sanitize_hot_paths):Update.
16362 * bt-load.c (migrate_btr_defs): Update.
16363 * cfg.c (RDIV): Remove.
16364 (init_flow): Use alloc_block.
16365 (alloc_block): Uninitialize count.
16366 (unchecked_make_edge): Uninitialize count.
16367 (check_bb_profile): Update.
16368 (dump_edge_info): Update.
16369 (dump_bb_info): Update.
16370 (update_bb_profile_for_threading): Update.
16371 (scale_bbs_frequencies_int): Update.
16372 (scale_bbs_frequencies_gcov_type): Update.
16373 (scale_bbs_frequencies_profile_count): New.
16374 * cfg.h (update_bb_profile_for_threading): Update.
16375 (scale_bbs_frequencies_profile_count): Declare.
16376 * cfgbuild.c (compute_outgoing_frequencies): Update.
16377 (find_many_sub_basic_blocks): Update.
16378 * cfgcleanup.c (try_forward_edges): Update.
16379 (try_crossjump_to_edge): Update.
16380 * cfgexpand.c (expand_gimple_tailcall): Update.
16381 (construct_exit_block): Update.
16382 * cfghooks.c (verify_flow_info): Update.
16383 (dump_bb_for_graph): Update.
16384 (split_edge): Update.
16385 (make_forwarder_block): Update.
16386 (duplicate_block): Update.
16387 (account_profile_record): Update.
16388 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16389 (get_estimated_loop_iterations): Update.
16390 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16391 (single_likely_exit): Update.
16392 * cfgloopmanip.c (scale_loop_profile): Update.
16394 (set_zero_probability): Update.
16395 (lv_adjust_loop_entry_edge): Update.
16396 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16397 (purge_dead_edges): Update.
16398 (rtl_account_profile_record): Update.
16399 * cgraph.c (cgraph_node::create): Uninitialize count.
16400 (symbol_table::create_edge): Uninitialize count.
16401 (cgraph_update_edges_for_call_stmt_node): Update.
16402 (cgraph_edge::dump_edge_flags): Update.
16403 (cgraph_node::dump): Update.
16404 (cgraph_edge::maybe_hot_p): Update.
16405 * cgraph.h: Include profile-count.h
16406 (create_clone), create_edge, create_indirect_edge): Update.
16407 (cgraph_node): Turn count to profile_count.
16408 (cgraph_edge0: Likewise.
16409 (make_speculative, clone): Update.
16410 (create_edge): Update.
16411 (init_lowered_empty_function): Update.
16412 * cgraphclones.c (cgraph_edge::clone): Update.
16413 (duplicate_thunk_for_node): Update.
16414 (cgraph_node::create_clone): Update.
16415 * cgraphunit.c (cgraph_node::analyze): Update.
16416 (cgraph_node::expand_thunk): Update.
16417 * final.c (dump_basic_block_info): Update.
16418 * gimple-streamer-in.c (input_bb): Update.
16419 * gimple-streamer-out.c (output_bb): Update.
16420 * graphite.c (print_global_statistics): Update.
16421 (print_graphite_scop_statistics): Update.
16422 * hsa-brig.c: Include basic-block.h.
16423 * hsa-dump.c: Include basic-block.h.
16424 * hsa-gen.c (T sum_slice): Update.
16425 (convert_switch_statements):Update.
16426 * hsa-regalloc.c: Include basic-block.h.
16427 * ipa-chkp.c (chkp_produce_thunks): Update.
16428 * ipa-cp.c (struct caller_statistics): Update.
16429 (init_caller_stats): Update.
16430 (gather_caller_stats): Update.
16431 (ipcp_cloning_candidate_p): Update.
16432 (good_cloning_opportunity_p): Update.
16433 (get_info_about_necessary_edges): Update.
16434 (dump_profile_updates): Update.
16435 (update_profiling_info): Update.
16436 (update_specialized_profile): Update.
16437 (perhaps_add_new_callers): Update.
16438 (decide_about_value): Update.
16439 (ipa_cp_c_finalize): Update.
16440 * ipa-devirt.c (struct odr_type_warn_count): Update.
16441 (struct decl_warn_count): Update.
16442 (struct final_warning_record): Update.
16443 (possible_polymorphic_call_targets): Update.
16444 (ipa_devirt): Update.
16445 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16446 * ipa-icf.c (sem_function::merge): Update.
16447 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16448 * ipa-inline.c (compute_uninlined_call_time): Update.
16449 (compute_inlined_call_time): Update.
16450 (want_inline_small_function_p): Update.
16451 (want_inline_self_recursive_call_p): Update.
16452 (edge_badness): Update.
16453 (lookup_recursive_calls): Update.
16454 (recursive_inlining): Update.
16455 (inline_small_functions): Update.
16456 (dump_overall_stats): Update.
16457 (dump_inline_stats): Update.
16458 * ipa-profile.c (ipa_profile_generate_summary): Update.
16459 (ipa_propagate_frequency): Update.
16460 (ipa_profile): Update.
16461 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16462 * ipa-utils.c (ipa_merge_profiles): Update.
16463 * loop-doloop.c (doloop_modify): Update.
16464 * loop-unroll.c (report_unroll): Update.
16465 (unroll_loop_runtime_iterations): Update.
16466 * lto-cgraph.c (lto_output_edge): Update.
16467 (lto_output_node): Update.
16468 (input_node): Update.
16469 (input_edge): Update.
16470 (merge_profile_summaries): Update.
16471 * lto-streamer-in.c (input_cfg): Update.
16472 * lto-streamer-out.c (output_cfg): Update.
16473 * mcf.c (create_fixup_graph): Update.
16474 (adjust_cfg_counts): Update.
16475 (sum_edge_counts): Update.
16476 * modulo-sched.c (sms_schedule): Update.
16477 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16478 * predict.c (maybe_hot_count_p): Update.
16479 (probably_never_executed): Update.
16480 (dump_prediction): Update.
16481 (combine_predictions_for_bb): Update.
16482 (propagate_freq): Update.
16483 (handle_missing_profiles): Update.
16484 (counts_to_freqs): Update.
16485 (rebuild_frequencies): Update.
16486 (force_edge_cold): Update.
16487 * predict.h: Include profile-count.h
16488 (maybe_hot_count_p, counts_to_freqs): UPdate.
16489 * print-rtl-function.c: Do not include cfg.h
16490 * print-rtl.c: Include basic-block.h
16491 * profile-count.c: New file.
16492 * profile-count.h: New file.
16493 * profile.c (is_edge_inconsistent): Update.
16494 (correct_negative_edge_counts): Update.
16495 (is_inconsistent): Update.
16496 (set_bb_counts): Update.
16497 (read_profile_edge_counts): Update.
16498 (compute_frequency_overlap): Update.
16499 (compute_branch_probabilities): Update; Initialize and deinitialize
16501 (branch_prob): Update.
16502 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16503 (edge_gcov_count): New.
16504 (bb_gcov_count): New.
16505 * shrink-wrap.c (try_shrink_wrapping): Update.
16506 * tracer.c (better_p): Update.
16507 * trans-mem.c (expand_transaction): Update.
16508 (ipa_tm_insert_irr_call): Update.
16509 (ipa_tm_insert_gettmclone_call): Update.
16510 * tree-call-cdce.c: Update.
16511 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16512 (gimple_duplicate_sese_tail): Update.
16513 (gimple_account_profile_record): Update.
16514 (execute_fixup_cfg): Update.
16515 * tree-inline.c (copy_bb): Update.
16516 (copy_edges_for_bb): Update.
16517 (initialize_cfun): Update.
16518 (freqs_to_counts): Update.
16519 (copy_cfg_body): Update.
16520 (expand_call_inline): Update.
16521 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16522 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16523 (try_unroll_loop_completely): Update.
16524 (try_peel_loop): Update.
16525 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16526 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16527 * tree-ssa-loop-split.c (connect_loops): Update.
16528 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16529 * tree-ssa-reassoc.c (branch_fixup): Update.
16530 * tree-ssa-tail-merge.c (replace_block_by): Update.
16531 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16532 (compute_path_counts): Update.
16533 (update_profile): Update.
16534 (recompute_probabilities): Update.
16535 (update_joiner_offpath_counts): Update.
16536 (estimated_freqs_path): Update.
16537 (freqs_to_counts_path): Update.
16538 (clear_counts_path): Update.
16539 (ssa_fix_duplicate_block_edges): Update.
16540 (duplicate_thread_path): Update.
16541 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16542 (struct switch_conv_info): Update.
16543 * tree-tailcall.c (decrease_profile): Update.
16544 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16545 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16546 * value-prof.c (check_counter): Update.
16547 (gimple_divmod_fixed_value): Update.
16548 (gimple_mod_pow2): Update.
16549 (gimple_mod_subtract): Update.
16550 (gimple_ic_transform): Update.
16551 (gimple_stringop_fixed_value): Update.
16552 * value-prof.h (gimple_ic): Update.
16554 2017-06-02 Carl Love <cel@us.ibm.com>
16556 * config/rs6000/rs6000-c: Add support for built-in functions
16557 vector double vec_doublee (vector signed int);
16558 vector double vec_doublee (vector unsigned int);
16559 vector double vec_doublee (vector float);
16560 vector double vec_doubleh (vector signed int);
16561 vector double vec_doubleh (vector unsigned int);
16562 vector double vec_doubleh (vector float);
16563 vector double vec_doublel (vector signed int);
16564 vector double vec_doublel (vector unsigned int);
16565 vector double vec_doublel (vector float);
16566 vector double vec_doubleo (vector signed int);
16567 vector double vec_doubleo (vector unsigned int);
16568 vector double vec_doubleo (vector float);.
16569 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16570 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16572 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16573 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16574 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16576 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16577 vec_doublel, vec_doubleh.
16578 * doc/extend.texi: Update the built-in documentation file for the
16579 new built-in functions.
16581 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16584 * ipa-inline-analysis.c (free_growth_caches): Set
16585 edge_removal_hook_holder to NULL after removing it.
16587 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16589 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16590 comparision with zero.
16592 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16593 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16594 for early expansion of vec_min and vec_max builtins.
16595 (builtin_function_type): Add min/max unsigned variants to those
16596 identified as having unsigned arguments.
16598 2017-06-02 Olivier Hainque <hainque@adacore.com>
16600 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16602 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16604 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16605 Use VALL_F16 iterator rather than VALL.
16607 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16609 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16610 Emit CBNZ inside loop when doing a strong exchange and comparing
16611 against zero. Generate the CC flags after the loop.
16613 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16615 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16616 (dl_section_ref): New.
16617 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16618 On AIX, append an expression to subtract the size of the
16619 section length to dl_section_ref.
16621 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16623 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16624 for early expansion of vector absolute builtins.
16626 2017-06-02 Richard Biener <rguenther@suse.de>
16628 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16629 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16631 2017-06-02 Richard Biener <rguenther@suse.de>
16633 PR tree-optimization/80948
16634 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16635 stmt order as well.
16637 2017-06-02 Richard Biener <rguenther@suse.de>
16639 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16641 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16642 for inductions as relevant.
16644 2017-06-02 Richard Biener <rguenther@suse.de>
16646 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16647 (vectorizable_induction): ... this. Remove dead code.
16649 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16651 * builtins. (expand_builtin_alloca): Remove second parameter and
16652 infer its value from the first parameter instead.
16653 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16655 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16657 PR rtl-optimization/80903
16658 * loop-doloop.c (add_test): Unshare sequence.
16660 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16662 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16664 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16666 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16668 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16669 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16670 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16671 (xlogue_layout::get_stub_rtx): Make static.
16672 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16673 (xlogue_layout::compute_stub_managed_regs): Rename to...
16674 (xlogue_layout::count_stub_managed_regs): ...this.
16675 (xlogue_layout::is_stub_managed_reg): New function.
16676 (xlogue_layout::m_stub_names): Rename to...
16677 (xlogue_layout::s_stub_names): ...this, make static.
16678 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16679 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16680 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16681 xlogue_layout::s_stub_names): Instantiate statics.
16682 (stub_managed_regs): Remove.
16683 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16684 (disable_call_ms2sysv_xlogues): Rename to...
16685 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16686 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16688 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16689 change after reload_completed.
16690 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16692 (ix86_expand_prologue): Likewise.
16693 (ix86_expand_epilogue): Likewise.
16694 (ix86_expand_split_stack_prologue): Likewise.
16695 (ix86_compute_frame_layout): Remove frame parameter ...
16696 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16697 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16699 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16700 (ix86_frame): Move from here ...
16701 * config/i386/i386.h (ix86_frame): ... to here.
16702 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16703 complete ix86_frame data structure instead. Remove some_ld_name.
16705 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16707 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16708 symbols that hold a DECL_VALUE_EXPR.
16710 2017-06-01 Martin Jambor <mjambor@suse.cz>
16712 PR tree-optimization/80898
16713 * tree-sra.c (process_subtree_disqualification): Removed.
16714 (disqualify_candidate): Do not acll
16715 process_subtree_disqualification.
16716 (subtree_mark_written_and_enqueue): New function.
16717 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16718 RHS has been disqualified and re-queue LHS if necessary. Apart
16719 from that, ignore disqualified RHS.
16721 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16723 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16724 address fetch for z10 or later.
16726 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16728 * config/arc/arc.md (tst_movb): Add guard when splitting.
16730 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16732 * config/arc/arc.c (arc_can_eliminate): Test against
16733 arc_frame_pointer_needed.
16735 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16737 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16738 to prevent store reordering.
16739 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16740 (type): Add block type.
16741 (stack_tie): Define special instruction to be used in
16744 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16746 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16747 constraint. It is not valid for the pattern.
16748 (noncommutative_binary_comparison): Likewise.
16750 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16752 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16755 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16757 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16758 be used by the reg-alloc.
16760 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16762 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16763 reg-alloc when having mul64 or mul32x16 instructions.
16764 (mulsidi3): Likewise.
16765 (umulsidi3): Likewise.
16766 (mulsi32x16): New pattern.
16767 (mulsi64): Likewise.
16768 (mulsidi64): Likewise.
16769 (umulsidi64): Likewise.
16770 (MUL32x16_REG): Define.
16771 (mul64_600): Use MUL32x16_REG.
16772 (mac64_600): Likewise.
16773 (umul64_600): Likewise.
16774 (umac64_600): Likewise.
16776 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16778 * config/arc/arc.md (mulsi3_700): Make it commutative.
16780 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16782 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16784 (*sign_extendsidi2_insn): Likewise for movstosw.
16786 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16788 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16789 the type of the input discriminant value. Convert the
16790 discriminant value of signedness vary.
16792 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16794 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16795 Add to -Wall section.
16797 2017-06-01 Richard Biener <rguenther@suse.de>
16799 PR middle-end/66313
16800 * fold-const.c (fold_plusminus_mult_expr): If the factored
16801 factor may be zero use a wrapping type for the inner operation.
16802 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16803 and handle moved defs.
16804 (process_assignment): Properly guard the unary op case. Return a
16805 tri-state indicating that moving the stmt before the call may allow
16806 to continue. Pass through to_move.
16807 (find_tail_calls): Handle moving unrelated defs before
16810 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16813 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16814 splitter result in the canonical way.
16816 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16818 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16819 also for 32bit target. Update insn attributes.
16820 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16822 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16824 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16825 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16826 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16827 (_mm_maskz_min_ss): New intrinsics.
16829 2017-05-31 Martin Liska <mliska@suse.cz>
16831 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16832 Change comment style to one we normally use.
16833 (vectorizable_reduction): Likewise.
16834 (vectorizable_induction): Likewise.
16835 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16836 (vectorizable_call): Likewise.
16837 (vectorizable_simd_clone_call): Likewise.
16838 (vectorizable_conversion): Likewise.
16839 (vectorizable_assignment): Likewise.
16840 (vectorizable_shift): Likewise.
16841 (vectorizable_operation): Likewise.
16842 (vectorizable_store): Likewise.
16843 (vectorizable_load): Likewise.
16844 * tree-vectorizer.h: Likewise.
16846 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
16848 * passes.c (emergency_dump_function): New.
16849 * tree-pass.h (emergency_dump_function): Declare.
16850 * plugin.c (plugins_internal_error_function): Remove.
16851 * plugin.h (plugins_internal_error_function): Remove declaration.
16852 * toplev.c (internal_error_function): New static function. Use it...
16853 (general_init): ...here.
16855 2017-05-31 Graham Markall <graham.markall@embecosm.com>
16857 * config/arc/arc.c (arc_print_operand): Handle constant operands.
16858 (arc_rtx_costs): Add costs for new patterns.
16859 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
16860 * config/arc/predicates.md: Add _1_2_3_operand predicate.
16862 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16864 * tree-ssa-strlen.c (get_next_strinfo): New function.
16865 (get_stridx_plus_constant): Use it.
16866 (zero_length_string): Likewise.
16867 (adjust_related_strinfos): Likewise.
16868 (adjust_last_stmt): Likewise.
16870 2017-05-31 Richard Biener <rguenther@suse.de>
16873 * config/i386/i386.c (ix86_expand_builtin): Remove assert
16874 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
16876 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
16878 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
16879 loop_vinfo argument and use of dependence distance vectors.
16880 Check instead whether the two references differ only in their
16881 initial value and assume that they have the same alignment if the
16882 difference is a multiple of the vector alignment.
16883 (vect_analyze_data_refs_alignment): Update call accordingly.
16885 2017-05-31 Martin Liska <mliska@suse.cz>
16888 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
16890 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16892 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
16893 (create_intersect_range_checks): Move from ...
16894 * tree-data-ref.c (create_intersect_range_checks_index)
16895 (create_intersect_range_checks): ... to here.
16896 (create_runtime_alias_checks): New function factored from ...
16897 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
16898 here. Call above function.
16899 * tree-data-ref.h (create_runtime_alias_checks): New function.
16901 2017-05-31 Bin Cheng <bin.cheng@arm.com>
16903 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
16904 segment length for dr_b and compute it in wide_int.
16906 2017-05-31 Richard Biener <rguenther@suse.de>
16908 PR tree-optimization/80906
16909 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
16910 and pass through iv_map.
16911 (copy_bb_and_scalar_dependences): Adjust.
16912 (translate_pending_phi_nodes): Likewise.
16913 (copy_loop_close_phi_args): Handle code-generating IVs instead
16916 2017-05-30 David Malcolm <dmalcolm@redhat.com>
16918 * diagnostic-color.c (color_dict): Add "type-diff".
16919 (parse_gcc_colors): Update comment.
16920 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
16921 -fdiagnostics-show-template-tree and -fno-elide-type.
16922 (GCC_COLORS): Add type-diff to example.
16924 (-fdiagnostics-show-template-tree): New.
16925 (-fno-elide-type): New.
16926 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
16927 the pp_format_decoder callback. Call any m_format_postprocessor's
16929 (pretty_printer::pretty_printer): Initialize
16930 m_format_postprocessor.
16931 (pretty_printer::~pretty_printer): Delete any
16932 m_format_postprocessor.
16933 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
16934 (class format_postprocessor): New class.
16935 (struct pretty_printer::format_decoder): Document the new parameters.
16936 (struct pretty_printer::m_format_postprocessor): New field.
16937 * tree-diagnostic.c (default_tree_printer): Update for new
16938 bool and const char ** params.
16939 * tree-diagnostic.h (default_tree_printer): Likewise.
16941 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
16943 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
16944 (lwa_operand): Delete rs6000_gen_cell_microcode test.
16945 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16946 rs6000_gen_cell_microcode code.
16947 (rs6000_final_prescan_insn): Delete.
16948 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
16949 "warn-cell-microcode" entries.
16950 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
16951 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
16952 throughout. Change cc_reg_not_micro_cr0_operand to
16953 cc_reg_not_cr0_operand throughout.
16954 (*extendhi<mode>2_noload): Delete.
16955 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
16956 (mwarn-cell-microcode): Delete.
16957 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
16958 -mgen-cell-microcode and -mwarn-cell-microcode.
16960 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
16963 * config/i386/constraints.md (Yd): New constraint.
16965 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
16966 and (?Yd, r) alternatives. Update insn attributes.
16967 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
16968 and (?*Yd, r) alternatives. Update insn attributes.
16969 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
16971 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
16973 * gimplify.c (gimplify_modify_expr): Don't create a
16974 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
16977 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16979 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
16981 2017-05-30 Richard Biener <rguenther@suse.de>
16983 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
16984 and reduc_def fields.
16985 (STMT_VINFO_REDUC_TYPE): New define.
16986 (STMT_VINFO_REDUC_DEF): Likewise.
16987 (vect_force_simple_reduction): Adjust prototype.
16988 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
16989 (vect_is_simple_reduction): Remove check_reduction argument.
16990 (vect_force_simple_reduction): Adjust and set
16991 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16992 (vectorizable_reduction): Do not re-do reduction analysis
16993 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
16994 * tree-parloops.c (gather_scalar_reductions): Adjust.
16996 2017-05-30 Richard Biener <rguenther@suse.de>
16998 PR middle-end/80901
16999 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17002 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17004 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17005 Introduce unknown_misalignment parameter and remove vf.
17006 (vect_peeling_hash_get_lowest_cost):
17007 Pass unknown_misalignment parameter.
17008 (vect_enhance_data_refs_alignment):
17009 Fix unsupportable data ref treatment.
17011 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17013 * tree-vect-data-refs.c (vect_get_data_access_cost):
17014 Workaround for SLP handling.
17015 (vect_enhance_data_refs_alignment):
17016 Compute costs for doing no peeling at all, compare to the best
17017 peeling costs so far and avoid peeling if cheaper.
17019 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17021 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17022 Return peeling info and set costs to zero for unlimited cost
17024 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17025 with unknown misalignment. Compute and costs for unknown
17026 misalignment, compare them to the costs for known misalignment
17027 and choose the cheapest for peeling.
17029 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17031 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17032 (vect_get_peeling_costs_all_drs): Create function.
17033 (vect_peeling_hash_get_lowest_cost):
17034 Use vect_get_peeling_costs_all_drs.
17035 (vect_peeling_supportable): Create function.
17036 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17038 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17040 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17041 DR_HAS_NEGATIVE_STEP.
17042 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17043 (vect_enhance_data_refs_alignment): Use.
17044 (vect_duplicate_ssa_name_ptr_info): Use.
17045 * tree-vectorizer.h (dr_misalignment): Use.
17046 (known_alignment_for_access_p): Use.
17048 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17051 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17053 (has_section_name): New function.
17055 2017-05-30 Martin Liska <mliska@suse.cz>
17058 * auto-profile.c (get_function_decl_from_block): Fix
17061 2017-05-30 Richard Biener <rguenther@suse.de>
17063 PR middle-end/80876
17064 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17066 2017-05-30 Martin Liska <mliska@suse.cz>
17068 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17069 * dumpfile.h (struct dump_file_info): Remove ctors.
17071 2017-05-30 Martin Liska <mliska@suse.cz>
17073 * predict.def: Fix GNU coding style.
17075 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17077 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17078 Mark 'to' argument with ATTRIBUTE_UNUSED.
17080 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17082 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17083 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17084 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17087 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17089 * doc/install.texi (Options specification): Restore entry of
17090 --enable-sjlj-exceptions.
17092 2017-05-27 Michael Eager <eager@eagercon.com>
17095 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17097 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17099 * config/microblaze/microblaze.h
17100 (FIXED_REGISTERS): Update in macro.
17101 (CALL_USED_REGISTERS): Update in macro.
17103 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17105 * doc/install.texi: Add links to macOS binary distributions.
17107 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17111 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17113 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17115 2017-05-26 Martin Liska <mliska@suse.cz>
17117 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17119 2017-05-26 Martin Liska <mliska@suse.cz>
17121 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17122 always leading ';; '.
17123 (dump_bb_info): Likewise.
17124 (brief_dump_cfg): Likewise.
17125 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17126 * dumpfile.c: Remove usage of TDF_VERBOSE.
17127 * dumpfile.h (enum dump_kind): Likewise.
17128 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17129 * print-tree.c (debug_verbose): Remove.
17130 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17131 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17132 * tree-diagnostic.c (default_tree_printer): Replace
17133 TDF_DIAGNOSTIC with TDF_SLIM.
17135 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17137 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17138 in parameter loop, rather than loop_vinfo.
17139 (create_intersect_range_checks): Ditto.
17140 (vect_create_cond_for_alias_checks): Update call to above functions.
17142 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17144 PR tree-optimization/80815
17145 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17146 for merging runtime alias checks. Handle negative DR_STEPs.
17148 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17150 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17152 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17153 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17154 out code pruning runtime alias checks.
17155 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17156 factored out from above.
17157 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17159 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17161 (prune_runtime_alias_test_list): New decalaration.
17163 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17165 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17166 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17167 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17168 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17169 (operator==, comp_dr_with_seg_len_pair): Ditto.
17170 (vect_prune_runtime_alias_test_list): Ditto.
17172 2017-05-26 Martin Liska <mliska@suse.cz>
17175 * params.def: Bound partial-inlining-entry-probability param.
17177 2017-05-26 Marek Polacek <polacek@redhat.com>
17180 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17183 2017-05-26 Richard Biener <rguenther@suse.de>
17185 PR tree-optimization/80842
17186 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17189 2017-05-26 Richard Biener <rguenther@suse.de>
17191 PR tree-optimization/80844
17192 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17194 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17196 * doc/md.texi (Machine Constraints): Update x86 family
17197 machine constraints section to match 'config/i386/constraints.md'.
17199 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17201 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17203 2017-05-25 Nathan Sidwell <nathan@acm.org>
17205 * doc/invoke.texi (--enable-languages): Update documentation.
17207 2017-05-25 Martin Liska <mliska@suse.cz>
17209 * dumpfile.c: Add TDF_FOLDING.
17210 * dumpfile.h (enum dump_kind): Likewise.
17211 * genmatch.c (dt_simplify::gen_1): Use it.
17213 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17215 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17217 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17219 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17220 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17222 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17224 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17225 * match.pd (X == C): Rewrite it here.
17226 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17227 with_certain_nonzero_bits2): New predicates.
17228 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17230 2017-05-24 Nathan Sidwell <nathan@acm.org>
17232 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17235 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17238 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17240 * config/powerpcspe: New port. Files are copied from the rs6000
17241 port, with "rs6000" in filenames replaced by "powerpcspe".
17243 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17245 PR rtl-optimization/80754
17246 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17248 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17250 * config/sparc/sparc.md (length): Return the correct value for -mflat
17251 sibcalls to match output_sibcall.
17253 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17257 * config/rs6000/rs6000.c (struct machine_function): Add new field
17259 (rs6000_get_separate_components): Init that field, use it.
17260 (rs6000_components_for_bb): Use the field.
17262 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17264 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17266 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17268 PR middle-end/80823
17269 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17271 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17274 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17275 address against address_operand predicate.
17276 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17278 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17280 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17281 parameters passed indirectly.
17283 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17285 * config/i386/i386.md (*movdi_internal): Remove SSE4
17286 alternative 18 (?r, *v). Update insn attributes.
17287 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17288 Update insn attributes.
17289 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17290 Update insn attributes.
17291 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17292 alternative 1 (r, v). Remove isa attribute.
17293 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17294 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17295 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17297 2017-05-23 Tom de Vries <tom@codesourcery.com>
17299 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17302 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17304 * cgraphunit.c (symbol_table::process_new_functions): Update.
17305 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17306 (inline_generate_summary): Rename to ...
17307 (ipa_fn_summary_generate): ... this one.
17308 (inline_read_summary): Rename to ...
17309 (ipa_fn_summary_read): ... this one.
17310 (inline_write_summary): Rename to ...
17311 (ipa_fn_summary_write): ... this one.
17312 (inline_free_summary): Rename to ...
17313 (ipa_free_fn_summary): ... this one.
17314 (pass_data_local_fn_summary, pass_local_fn_summary,
17315 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17316 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17317 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17318 make_pass_ipa_fn_summary): New.
17319 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17320 inline_write_summary, inline_free_summary): Remove.
17321 (ipa_free_fn_summary) : New.
17322 * ipa-inline.c (ipa_inline): Update.
17323 (pass_ipa_inline): Do not generate summaries.
17324 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17326 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17327 and add pass_ipa_fn_summary.
17328 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17330 (make_pass_inline_parameters): Remove.
17332 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17334 * omp-low.c (struct omp_context): Remove "default_kind" member.
17337 * omp-offload.c (execute_oacc_device_lower): Remove the
17338 parallelism dimensions function attributes for unparallelized
17339 OpenACC kernels constructs.
17341 2017-05-23 Martin Liska <mliska@suse.cz>
17343 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17345 (cgraph_edge::make_speculative): Likewise.
17346 (cgraph_edge::resolve_speculation): Likewise.
17347 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17348 (cgraph_node::dump): Likewise.
17349 * cgraph.h: Likewise.
17350 * cgraphunit.c (analyze_functions): Likewise.
17351 (symbol_table::compile): Likewise.
17352 * ipa-cp.c (print_all_lattices): Likewise.
17353 (determine_versionability): Likewise.
17354 (initialize_node_lattices): Likewise.
17355 (ipcp_verify_propagated_values): Likewise.
17356 (estimate_local_effects): Likewise.
17357 (update_profiling_info): Likewise.
17358 (create_specialized_node): Likewise.
17359 (perhaps_add_new_callers): Likewise.
17360 (decide_about_value): Likewise.
17361 (decide_whether_version_node): Likewise.
17362 (identify_dead_nodes): Likewise.
17363 (ipcp_store_bits_results): Likewise.
17364 * ipa-devirt.c (dump_targets): Likewise.
17365 (ipa_devirt): Likewise.
17366 * ipa-icf.c (sem_item::dump): Likewise.
17367 (sem_function::equals): Likewise.
17368 (sem_variable::equals): Likewise.
17369 (sem_item_optimizer::read_section): Likewise.
17370 (sem_item_optimizer::execute): Likewise.
17371 (congruence_class::dump): Likewise.
17372 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17373 (dump_inline_summary): Likewise.
17374 (estimate_node_size_and_time): Likewise.
17375 (inline_analyze_function): Likewise.
17376 * ipa-inline-transform.c (inline_call): Likewise.
17377 * ipa-inline.c (report_inline_failed_reason): Likewise.
17378 (want_early_inline_function_p): Likewise.
17379 (edge_badness): Likewise.
17380 (update_edge_key): Likewise.
17381 (inline_small_functions): Likewise.
17382 * ipa-profile.c (ipa_profile): Likewise.
17383 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17384 (ipa_make_edge_direct_to_target): Likewise.
17385 (remove_described_reference): Likewise.
17386 (ipa_impossible_devirt_target): Likewise.
17387 (propagate_controlled_uses): Likewise.
17388 (ipa_print_node_params): Likewise.
17389 (ipcp_transform_function): Likewise.
17390 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17391 (propagate_pure_const): Likewise.
17392 * ipa-reference.c (generate_summary): Likewise.
17393 (read_write_all_from_decl): Likewise.
17394 (propagate): Likewise.
17395 (ipa_reference_read_optimization_summary): Likewise.
17396 * ipa-utils.c (ipa_merge_profiles): Likewise.
17397 * ipa.c (walk_polymorphic_call_targets): Likewise.
17398 (symbol_table::remove_unreachable_nodes): Likewise.
17399 (ipa_single_use): Likewise.
17400 * passes.c (execute_todo): Likewise.
17401 * predict.c (drop_profile): Likewise.
17402 * symtab.c (symtab_node::get_dump_name): New function.
17403 (symtab_node::dump_name): Likewise.
17404 (symtab_node::dump_asm_name): Likewise.
17405 (symtab_node::dump_references): Likewise.
17406 (symtab_node::dump_referring): Likewise.
17407 (symtab_node::dump_base): Likewise.
17408 (symtab_node::debug_symtab): Likewise.
17409 * tree-sra.c (convert_callers_for_node): Likewise.
17410 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17411 * value-prof.c (init_node_map): Likewise.
17413 2017-05-23 Martin Liska <mliska@suse.cz>
17415 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17416 and symtab_node::debug_symtab to symbol_table::debug.
17417 * cgraphunit.c (analyze_functions): Use the renamed function.
17418 (symbol_table::compile): Likewise.
17419 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17420 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17421 * passes.c (execute_todo): Likewise.
17422 * symtab.c (symbol_table::dump): New function.
17423 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17425 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17427 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17428 that nonconst implies exec.
17430 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17432 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17433 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17434 (inline_edge_summary_vec): Turn into ...
17435 (ipa_call_summaries): ... this one.
17436 (redirect_to_unreachable, edge_set_predicate,
17437 evaluate_properties_for_edge, inline_summary_alloc,
17438 reset_ipa_call_summary, reset_inline_summary,
17439 inline_summary_t::duplicate): Update.
17440 (inline_edge_duplication_hook): Turn to ...
17441 (ipa_call_summary_t::duplicate): ... this one.
17442 (inline_edge_removal_hook): Turn to ...
17443 (ipa_call_summary_t::remove): ... this one.
17444 (dump_inline_edge_summary): Turn to ...
17445 (dump_ipa_call_summary): ... this one.
17446 (estimate_function_body_sizes): Update.
17447 (inline_update_callee_summaries): Update.
17448 (remap_edge_change_prob): Update.
17449 (remap_edge_summaries): Update.
17450 (inline_merge_summary): Update.
17451 (do_estimate_edge_time): Update.
17452 (inline_generate_summary): Update.
17453 (inline_read_section): Update.
17454 (inline_read_summary): Update.
17455 (inline_free_summary): Update.
17456 * ipa-inline.c (can_inline_edge_p): Update.
17457 (compute_inlined_call_time): Update.
17458 (want_inline_small_function_p): Update.
17459 (edge_badness): Update.
17460 (early_inliner): Update.
17461 * ipa-inline.h (inline_edge_summary): Turn to ...
17462 (ipa_call_summary): ... this one.
17463 (ipa_call_summary_t): New class.
17464 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17465 (ipa_call_summaries): New.
17466 (inline_edge_summary): Remove.
17467 (estimate_edge_growth): Update.
17468 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17469 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17470 * ipa-split.c (execute_split_functions): Update.
17471 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17473 2017-05-23 Tom de Vries <tom@codesourcery.com>
17475 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17476 attributes): Document rdrand effective target.
17478 2017-05-23 Tom de Vries <tom@codesourcery.com>
17480 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17481 attributes): Sort alphabetically.
17483 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17485 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17487 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17490 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17491 V2DF/V2DI splat into two separate patterns, one that handles
17492 registers, and the other that only handles memory. Drop support
17493 for splatting from a GPR on ISA 2.07 and then splitting the
17494 splat into direct move and splat.
17495 (vsx_splat_<mode>_reg): Likewise.
17496 (vsx_splat_<mode>_mem): Likewise.
17498 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17500 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17502 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17504 PR middle-end/80809
17505 * omp-low.c (finish_taskreg_remap): New function.
17506 (finish_taskreg_scan): If unit size of ctx->record_type
17507 is non-constant, unshare the size expression and replace
17508 decls in it with possible outer var refs.
17510 PR middle-end/80809
17511 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17512 GOVD_SHARED rather than GOVD_PRIVATE with it.
17513 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17514 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17516 PR middle-end/80853
17517 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17518 as last argument to build_outer_var_ref for pointer bases of array
17519 section reductions.
17521 2017-05-19 Martin Sebor <msebor@redhat.com>
17523 * print-tree.c (print_node): Print DECL_READ_P flag.
17525 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17527 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17528 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17529 * cgraph.c: Likewise.
17530 * cgraphunit.c: Likewise.
17531 * gengtype.c: Likewise.
17532 * ipa-cp.c: Likewise.
17533 * ipa-devirt.c: Likewise.
17534 * ipa-icf.c: Likewise.
17535 * ipa-predicate.c: Likewise.
17536 * ipa-profile.c: Likewise.
17537 * ipa-prop.c: Likewise.
17538 * ipa-split.c: Likewise.
17540 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17541 edge_predicate_pool, dump_inline_hints,
17542 inline_summary::account_size_time, redirect_to_unreachable,
17543 edge_set_predicate, set_hint_predicate,
17544 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17545 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17546 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17547 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17548 ipa_call_summary_t::remove, initialize_growth_caches,
17549 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17550 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17551 mark_modified, unmodified_parm_1, unmodified_parm,
17552 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17553 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17554 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17555 will_be_nonconstant_predicate, record_modified_bb_info,
17556 get_minimal_bb, record_modified, param_change_prob,
17557 phi_result_unknown_predicate, predicate_for_phi_result,
17558 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17559 estimate_function_body_sizes, compute_inline_parameters,
17560 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17561 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17562 inline_update_callee_summaries, remap_edge_change_prob,
17563 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17564 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17565 inline_analyze_function, inline_summary_t::insert,
17566 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17567 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17568 inline_free_summary): Move to ipa-fnsummary.h
17569 (predicate_t): Remove.
17570 * ipa-fnsummary.c: New file.
17571 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17572 (enum inline_hints_vals, inline_hints, agg_position_info,
17573 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17574 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17575 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17576 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17577 inline_read_summary, inline_write_summary, inline_free_summary,
17578 inline_analyze_function, initialize_inline_failed,
17579 inline_merge_summary, inline_update_overall_summary,
17580 compute_inline_parameters): Move to ipa-fnsummary.h
17581 * ipa-fnsummary.h: New file.
17582 * ipa-inline-transform.h: Include ipa-inline.h.
17583 * ipa-inline.c: LIkewise.
17585 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17587 * ipa-inline.c (edge_badness): Use inlined_time instead of
17588 inline_summaries->get.
17590 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17592 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17594 2017-05-22 Nathan Sidwell <nathan@acm.org>
17596 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17597 (fdump-lang): Document 'raw' option.
17598 * dumpfile.h (TDI_tu): Delete.
17599 * dumpfile.c (dump_files): Remove translation-unit.
17600 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17602 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17604 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17605 command option from $(AWK) call.
17606 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17607 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17608 [FORMAT]: Remove handling of variable.
17609 * config/avr/t-multilib: Regenerate.
17611 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17613 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17615 (dump_inline_summary): Do not print self_time.
17616 (estimate_function_body_sizes): Do not set self_time.
17617 (compute_inline_parameters): Likewise.
17618 (inline_read_section, inline_write_summary): Do not stream self_time.
17619 * ipa-inline.h (inline_summary): Drop self_time.
17621 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17623 * ipa-inline-analysis.c (account_size_time): Rename to ...
17624 (inline_summary::account_size_time): ... this one.
17625 (reset_ipa_call_summary): Turn to ...
17626 (ipa_call_summary::reset): ... this one.
17627 (reset_inline_summary): Turn to ...
17628 (inline_summary::reset): ... this one.
17629 (inline_summary_t::remove): Update.
17630 (inline_summary_t::duplicate): Update.
17631 (ipa_call_summary_t::remove): Update.
17632 (dump_inline_summary): Update.
17633 (estimate_function_body_sizes): Update.
17634 (compute_inline_parameters): Update.
17635 (estimate_node_size_and_time): Update.
17636 (inline_merge_summary): Update.
17637 (inline_update_overall_summary): Update.
17638 (inline_read_section): Update.
17639 (inline_write_summary): Update.
17640 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17641 add account_size_time and reset member functions.
17642 (ipa_call_summary): Add reset function.
17643 * ipa-predicate.h (predicate::operator &): Constify.
17645 2017-05-22 Richard Biener <rguenther@suse.de>
17647 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17649 2017-05-19 Jason Merrill <jason@redhat.com>
17651 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17653 2017-05-19 Marek Polacek <polacek@redhat.com>
17656 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17657 TYPE_OVERFLOW_WRAPS checks.
17659 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17661 * tree-core.h (enum omp_clause_default_kind): Add
17662 "OMP_CLAUSE_DEFAULT_PRESENT".
17663 * tree-pretty-print.c (dump_omp_clause): Handle it.
17664 * gimplify.c (enum gimplify_omp_var_data): Add
17665 "GOVD_MAP_FORCE_PRESENT".
17666 (gimplify_adjust_omp_clauses_1): Map it to
17667 "GOMP_MAP_FORCE_PRESENT".
17668 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17670 * gimplify.c (oacc_default_clause): Clarify.
17672 2017-05-19 Nathan Sidwell <nathan@acm.org>
17674 LANG_HOOK_REGISTER_DUMPS
17675 * toplev.c (general_init): Call register dump lang hook.
17676 * doc/invoke.texi: Document -fdump-lang option family.
17677 * dumpfile.c (dump_files): Remove class dump here.
17678 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17679 * dumpfile.h (tree_dump_index): Remove TDI_class.
17680 * langhooks-def.h (lhd_register_dumps): Declare.
17681 (LANG_HOOKS_REGISTER_DUMPS): Define.
17682 (LANG_HOOKS_INITIALIZER): Add it.
17683 * langhooks.c (lhd_register_dumps): Define.
17684 * langhooks.h (struct lang_hooks): Add register_dumps.
17686 2017-05-19 Nathan Sidwell <nathan@acm.org>
17688 * context.h (context::set_passes): New.
17689 * context.c (context::context): Do not create pass manager.
17690 * toplev.c (general_init): Create pass manager here.
17692 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17694 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17695 use this splitter if two add or or instructions would also work for
17696 the constant we want to generate.
17698 2017-05-19 Richard Biener <rguenther@suse.de>
17701 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17702 predicate evaluation.
17704 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17706 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17708 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17711 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17713 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17714 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17715 (inline_edge_summary_vec): Turn into ...
17716 (ipa_call_summaries): ... this one.
17717 (redirect_to_unreachable, edge_set_predicate,
17718 evaluate_properties_for_edge, inline_summary_alloc,
17719 reset_ipa_call_summary, reset_inline_summary,
17720 inline_summary_t::duplicate): Update.
17721 (inline_edge_duplication_hook): Turn to ...
17722 (ipa_call_summary_t::duplicate): ... this one.
17723 (inline_edge_removal_hook): Turn to ...
17724 (ipa_call_summary_t::remove): ... this one.
17725 (dump_inline_edge_summary): Turn to ...
17726 (dump_ipa_call_summary): ... this one.
17727 (estimate_function_body_sizes): Update.
17728 (inline_update_callee_summaries): Update.
17729 (remap_edge_change_prob): Update.
17730 (remap_edge_summaries): Update.
17731 (inline_merge_summary): Update.
17732 (do_estimate_edge_time): Update.
17733 (inline_generate_summary): Update.
17734 (inline_read_section): Update.
17735 (inline_read_summary): Update.
17736 (inline_free_summary): Update.
17737 * ipa-inline.c (can_inline_edge_p): Update.
17738 (compute_inlined_call_time): Update.
17739 (want_inline_small_function_p): Update.
17740 (edge_badness): Update.
17741 (early_inliner): Update.
17742 * ipa-inline.h (inline_edge_summary): Turn to ...
17743 (ipa_call_summary): ... this one.
17744 (ipa_call_summary_t): New class.
17745 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17746 (ipa_call_summaries): New.
17747 (inline_edge_summary): Remove.
17748 (estimate_edge_growth): Update.
17749 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17750 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17751 * ipa-split.c (execute_split_functions): Update.
17752 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17754 2017-05-19 Richard Biener <rguenther@suse.de>
17756 PR middle-end/80764
17757 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17759 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17761 * config/rs6000/rs6000.c (struct machine_function): Add field
17762 fpr_is_wrapped_separately.
17763 (rs6000_get_separate_components): Use 64 components. Handle the
17764 new FPR components.
17765 (rs6000_components_for_bb): Handle the FPR components.
17766 (rs6000_emit_prologue_components): Handle the FPR components.
17767 (rs6000_emit_epilogue_components): Handle the FPR components.
17768 (rs6000_set_handled_components): Handle the FPR components.
17769 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17770 that are already separately shrink-wrapped.
17771 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17772 that are already separately shrink-wrapped.
17774 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17777 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17780 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17781 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17782 cases where the register allocator uses a move and an offsettable
17783 memory operation to/from a FPR register on ISA 2.06/2.07.
17784 (define_peephole2 for Altivec d-form store): Likewise.
17786 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17789 * config/i386/mmx.md (*mov<mode>_internal): Enable
17790 alternatives 11, 12, 13 and 14 also for 32bit targets.
17791 Remove alternatives 15, 16, 17 and 18.
17792 * config/i386/sse.md (vec_concatv2di): Change
17793 alternative (!x, *y) to (x, ?!*Yn).
17795 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17797 * dumpfile.h (enum dump_kind): Remove stray comma.
17799 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17801 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17802 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17803 predicate::num_conditions
17804 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17805 (CHANGED): turn into predicate::changed.
17806 (agg_position_info): Move to ipa-predicate.h
17807 (add_condition, predicate::add_clause, predicate::operator &=,
17808 predicate::or_with, predicate::evaluate, predicate::probability,
17809 dump_condition, dump_clause, predicate::dump,
17810 predicate::remap_after_duplication, predicate::remap_after_inlining,
17811 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17812 (evaluate_conditions_for_known_args): Update.
17813 (set_cond_stmt_execution_predicate): Update.
17814 * ipa-inline.h: Include ipa-predicate.h
17815 (condition, inline_param_summary, conditions, agg_position_info,
17816 predicate): Move to ipa-predicate.h
17817 * ipa-predicate.c: New file.
17818 * ipa-predicate.h: New file.
17820 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17822 * final.c (leaf_function_p): Check we are not in a sequence.
17824 2017-05-18 Martin Liska <mliska@suse.cz>
17826 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17827 * dumpfile.c (dump_register): Use new enum dump_kind.
17828 (get_dump_file_name): Likewise.
17829 (dump_enable_all): Likewise.
17830 (dump_switch_p_1): Likewise.
17831 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17832 * dumpfile.h (enum dump_kind): New enum type.
17833 (struct dump_file_info): Create constructor and
17834 format fields and comments.
17835 * passes.c (pass_manager::register_one_dump_file):
17837 * statistics.c (statistics_early_init): Likewise.
17838 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17839 TDF_TREE with TDF_SLIM.
17840 (gather_memory_references_ref): Likewise.
17842 2017-05-18 Martin Liska <mliska@suse.cz>
17844 * vec.h (struct vnull): Use it.
17846 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17848 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
17849 (true_predicate, false_predicate, true_predicate_p,
17850 false_predicate_p): Remove.
17851 (single_cond_predicate, not_inlined_predicate): Turn to member function
17853 (add_condition): Update.
17854 (add_clause): Turn to...
17855 (predicate::add_clause): ... this one; update; allow passing NULL
17857 (and_predicates): Turn to ...
17858 (predicate::operator &=): ... this one.
17859 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
17860 (or_predicates): Turn to ...
17861 (predicate::or_with): ... this one.
17862 (evaluate_predicate): Turn to ...
17863 (predicate::evaluate): ... this one.
17864 (predicate_probability): Turn to ...
17865 (predicate::probability): ... this one.
17866 (dump_condition): Update.
17867 (dump_predicate): Turn to ...
17868 (predicate::dump): ... this one.
17869 (account_size_time): Update.
17870 (edge_set_predicate): Update.
17871 (set_hint_predicate): UPdate.
17872 (evaluate_conditions_for_known_args): Update.
17873 (evaluate_properties_for_edge): Update.
17874 (remap_predicate_after_duplication): Turn to...
17875 (predicate::remap_after_duplication): ... this one.
17876 (remap_hint_predicate_after_duplication): Update.
17877 (inline_summary_t::duplicate): UPdate.
17878 (dump_inline_edge_summary): Update.
17879 (dump_inline_summary): Update.
17880 (set_cond_stmt_execution_predicate): Update.
17881 (set_switch_stmt_execution_predicate): Update.
17882 (compute_bb_predicates): Update.
17883 (will_be_nonconstant_expr_predicate): Update.
17884 (will_be_nonconstant_predicate): Update.
17885 (phi_result_unknown_predicate): Update.
17886 (predicate_for_phi_result): Update.
17887 (array_index_predicate): Update.
17888 (estimate_function_body_sizes): Update.
17889 (estimate_node_size_and_time): Update.
17890 (estimate_ipcp_clone_size_and_time): Update.
17891 (remap_predicate): Rename to ...
17892 (predicate::remap_after_inlining): ... this one.
17893 (remap_hint_predicate): Update.
17894 (inline_merge_summary): Update.
17895 (inline_update_overall_summary): Update.
17896 (estimate_size_after_inlining): Update.
17897 (read_predicate): Rename to ...
17898 (predicate::stream_in): ... this one.
17899 (read_inline_edge_summary): Update.
17900 (write_predicate): Rename to ...
17901 (predicate::stream_out): ... this one.
17902 (write_inline_edge_summary): Update.
17903 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
17904 (clause_t): Turn to uint32_t
17905 (predicate): Turn to class; implement constructor and operators
17907 (size_time_entry): Update.
17908 (inline_summary): Update.
17909 (inline_edge_summary): Update.
17911 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
17913 * fold-const.c (fold_binary_loc): Move transformation...
17914 * match.pd (C - X CMP X): ... here.
17916 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
17918 * config/sparc/sparc.c (sparc_option_override): Set function
17919 alignment for -mcpu=niagara7 to 64 to match the I$ line.
17920 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
17922 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
17924 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
17926 2017-05-18 Marek Polacek <polacek@redhat.com>
17929 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
17930 (pass_ubsan::execute): Call gimple_assign_single_p instead of
17931 gimple_assign_load_p.
17933 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
17935 PR middle-end/80692
17936 * real.c (do_compare): Give decimal_do_compare preference over
17937 comparing just the signs.
17939 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
17941 * doc/md.texi (Canonicalization of Instructions): Describe the
17942 canonical form of instructions that inherently set a condition
17945 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
17947 PR middle-end/80775
17948 * tree-cfg.c: Move deletion of unreachable case statements to after
17949 the merging of consecutive case labels.
17951 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
17953 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
17954 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
17955 restoring of callee-saved registers.
17957 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
17959 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
17960 * config/visium/visium.c (single_set_and_flags): Likewise.
17961 * config/visium/visium.md (Substitutions): Likewise.
17963 2017-05-17 Martin Liska <mliska@suse.cz>
17965 * cfg.c: Introduce dump_flags_t type and
17966 use it instead of int type.
17968 * cfghooks.c: Likewise.
17969 * cfghooks.h (struct cfg_hooks): Likewise.
17970 * cfgrtl.c: Likewise.
17971 * cfgrtl.h: Likewise.
17972 * cgraph.c (cgraph_node::get_body): Likewise.
17973 * coretypes.h: Likewise.
17974 * domwalk.c: Likewise.
17975 * domwalk.h: Likewise.
17976 * dumpfile.c (struct dump_option_value_info): Likewise.
17977 (dump_enable_all): Likewise.
17978 (dump_switch_p_1): Likewise.
17979 (opt_info_switch_p): Likewise.
17980 * dumpfile.h (enum tree_dump_index): Likewise.
17981 (struct dump_file_info): Likewise.
17982 * genemit.c: Likewise.
17983 * generic-match-head.c: Likewise.
17984 * gengtype.c (open_base_files): Likewise.
17985 * gimple-pretty-print.c: Likewise.
17986 * gimple-pretty-print.h: Likewise.
17987 * graph.c (print_graph_cfg): Likewise.
17988 * graphite-scop-detection.c (dot_all_sese): Likewise.
17989 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
17990 * loop-unroll.c (report_unroll): Likewise.
17991 * passes.c (pass_manager::register_one_dump_file): Likewise.
17992 * print-tree.c: Likewise.
17993 * statistics.c: Likewise.
17994 * tree-cfg.c: Likewise.
17995 * tree-cfg.h: Likewise.
17996 * tree-dfa.c: Likewise.
17997 * tree-dfa.h: Likewise.
17998 * tree-dump.c (dump_function): Likewise.
17999 * tree-dump.h (struct dump_info): Likewise.
18000 * tree-pretty-print.c: Likewise.
18001 * tree-pretty-print.h: Likewise.
18002 * tree-ssa-live.c: Likewise.
18003 * tree-ssa-live.h: Likewise.
18004 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18005 * tree-vect-loop.c: Likewise.
18006 * tree-vect-slp.c: Likewise.
18008 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18009 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18011 PR tree-optimization/80457
18012 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18013 of all arguments to a statement as scalar_to_vec operations.
18014 (vectorizable_call): Adjust call to vect_model_simple_cost for
18016 (vectorizable_conversion): Likewise.
18017 (vectorizable_assignment): Likewise.
18018 (vectorizable_shift): Likewise.
18019 (vectorizable_operation): Likewise.
18020 (vectorizable_comparison): Likewise.
18021 (vect_is_simple_cond): Record the def types for operands.
18022 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18023 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18024 for statement argument count.
18026 2017-05-16 Carl Love <cel@us.ibm.com>
18028 * config/rs6000/rs6000-c: Add support for built-in functions
18029 vector unsigned long long vec_bperm (vector unsigned long long,
18030 vector unsigned char)
18031 vector signed long long vec_mule (vector signed int,
18033 vector unsigned long long vec_mule (vector unsigned int,
18034 vector unsigned int)
18035 vector signed long long vec_mulo (vector signed int,
18037 vector unsigned long long vec_mulo (vector unsigned int,
18038 vector unsigned int)
18039 vector signed char vec_sldw (vector signed char,
18040 vector signed char,
18042 vector unsigned char vec_sldw (vector unsigned char,
18043 vector unsigned char,
18045 vector signed short vec_sldw (vector signed short,
18046 vector signed short,
18048 vector unsigned short vec_sldw (vector unsigned short,
18049 vector unsigned short,
18051 vector signed int vec_sldw (vector signed int,
18054 vector unsigned int vec_sldw (vector unsigned int,
18055 vector unsigned int,
18057 vector signed long long vec_sldw (vector signed long long,
18058 vector signed long long,
18060 vector unsigned long long vec_sldw (vector unsigned long long,
18061 vector unsigned long long,
18063 * config/rs6000/rs6000-c: Add support for built-in functions
18064 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18065 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18066 * doc/extend.texi: Update the built-in documentation for the
18067 new built-in functions.
18069 2017-05-16 Marek Polacek <polacek@redhat.com>
18073 * tree.c (save_expr): Don't fold the expression.
18075 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18077 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18078 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18079 and (?*y,m). Update insn attributes.
18081 2017-05-16 Martin Liska <mliska@suse.cz>
18083 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18084 flags argument of print_gimple_stmt, print_gimple_expr,
18085 print_generic_stmt and print_generic_expr.
18086 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18087 * coretypes.h: Likewise.
18088 * except.c (dump_eh_tree): Likewise.
18089 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18090 * gimple-pretty-print.h: Likewise.
18091 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18092 (backprop::push_to_worklist): Likewise.
18093 (backprop::pop_from_worklist): Likewise.
18094 (backprop::process_use): Likewise.
18095 (backprop::intersect_uses): Likewise.
18096 (note_replacement): Likewise.
18097 * gimple-ssa-store-merging.c
18098 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18099 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18100 (pass_store_merging::execute): Likewise.
18101 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18102 (ssa_base_cand_dump_callback): Likewise.
18103 (dump_incr_vec): Likewise.
18104 (replace_refs): Likewise.
18105 (replace_mult_candidate): Likewise.
18106 (create_add_on_incoming_edge): Likewise.
18107 (create_phi_basis): Likewise.
18108 (insert_initializers): Likewise.
18109 (all_phi_incrs_profitable): Likewise.
18110 (introduce_cast_before_cand): Likewise.
18111 (replace_one_candidate): Likewise.
18112 * gimplify.c (gimplify_expr): Likewise.
18113 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18114 (set_rename): Likewise.
18115 (rename_uses): Likewise.
18116 (copy_loop_phi_nodes): Likewise.
18117 (add_close_phis_to_merge_points): Likewise.
18118 (copy_loop_close_phi_args): Likewise.
18119 (copy_cond_phi_args): Likewise.
18120 (graphite_copy_stmts_from_block): Likewise.
18121 (translate_pending_phi_nodes): Likewise.
18122 * graphite-poly.c (print_pdr): Likewise.
18123 (dump_gbb_cases): Likewise.
18124 (dump_gbb_conditions): Likewise.
18125 (print_scop_params): Likewise.
18126 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18127 (build_cross_bb_scalars_use): Likewise.
18128 (gather_bbs::before_dom_children): Likewise.
18129 * hsa-dump.c (dump_hsa_immed): Likewise.
18130 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18131 (get_replacement_map): Likewise.
18132 * ipa-inline-analysis.c (dump_condition): Likewise.
18133 (estimate_function_body_sizes): Likewise.
18134 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18135 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18136 * ipa-prop.c (ipa_dump_param): Likewise.
18137 (ipa_print_node_jump_functions_for_edge): Likewise.
18138 (ipa_modify_call_arguments): Likewise.
18139 (ipa_modify_expr): Likewise.
18140 (ipa_dump_param_adjustments): Likewise.
18141 (ipa_dump_agg_replacement_values): Likewise.
18142 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18143 * ipa-pure-const.c (check_stmt): Likewise.
18144 (pass_nothrow::execute): Likewise.
18145 * ipa-split.c (execute_split_functions): Likewise.
18146 * omp-offload.c (dump_oacc_loop_part): Likewise.
18147 (dump_oacc_loop): Likewise.
18148 * trans-mem.c (tm_log_emit): Likewise.
18149 (tm_memopt_accumulate_memops): Likewise.
18150 (dump_tm_memopt_set): Likewise.
18151 (dump_tm_memopt_transform): Likewise.
18152 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18153 (print_loop): Likewise.
18154 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18155 (chkp_gather_checks_info): Likewise.
18156 (chkp_get_check_result): Likewise.
18157 (chkp_remove_check_if_pass): Likewise.
18158 (chkp_use_outer_bounds_if_possible): Likewise.
18159 (chkp_reduce_bounds_lifetime): Likewise.
18160 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18161 (chkp_mark_completed_bounds): Likewise.
18162 (chkp_register_incomplete_bounds): Likewise.
18163 (chkp_mark_invalid_bounds): Likewise.
18164 (chkp_maybe_copy_and_register_bounds): Likewise.
18165 (chkp_build_returned_bound): Likewise.
18166 (chkp_get_bound_for_parm): Likewise.
18167 (chkp_build_bndldx): Likewise.
18168 (chkp_get_bounds_by_definition): Likewise.
18169 (chkp_generate_extern_var_bounds): Likewise.
18170 (chkp_get_bounds_for_decl_addr): Likewise.
18171 * tree-chrec.c (chrec_apply): Likewise.
18172 * tree-data-ref.c (dump_data_reference): Likewise.
18173 (dump_subscript): Likewise.
18174 (dump_data_dependence_relation): Likewise.
18175 (analyze_overlapping_iterations): Likewise.
18176 * tree-inline.c (expand_call_inline): Likewise.
18177 (tree_function_versioning): Likewise.
18178 * tree-into-ssa.c (dump_defs_stack): Likewise.
18179 (dump_currdefs): Likewise.
18180 (dump_names_replaced_by): Likewise.
18181 (dump_update_ssa): Likewise.
18182 (update_ssa): Likewise.
18183 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18184 * tree-parloops.c (build_new_reduction): Likewise.
18185 (try_create_reduction_list): Likewise.
18186 (ref_conflicts_with_region): Likewise.
18187 (oacc_entry_exit_ok_1): Likewise.
18188 (oacc_entry_exit_single_gang): Likewise.
18189 * tree-pretty-print.h: Likewise.
18190 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18191 (get_scalar_evolution): Likewise.
18192 (add_to_evolution): Likewise.
18193 (get_loop_exit_condition): Likewise.
18194 (analyze_evolution_in_loop): Likewise.
18195 (analyze_initial_condition): Likewise.
18196 (analyze_scalar_evolution): Likewise.
18197 (instantiate_scev): Likewise.
18198 (number_of_latch_executions): Likewise.
18199 (gather_chrec_stats): Likewise.
18200 (final_value_replacement_loop): Likewise.
18201 (scev_const_prop): Likewise.
18202 * tree-sra.c (dump_access): Likewise.
18203 (disqualify_candidate): Likewise.
18204 (create_access): Likewise.
18205 (reject): Likewise.
18206 (maybe_add_sra_candidate): Likewise.
18207 (create_access_replacement): Likewise.
18208 (analyze_access_subtree): Likewise.
18209 (analyze_all_variable_accesses): Likewise.
18210 (sra_modify_assign): Likewise.
18211 (initialize_constant_pool_replacements): Likewise.
18212 (find_param_candidates): Likewise.
18213 (decide_one_param_reduction): Likewise.
18214 (replace_removed_params_ssa_names): Likewise.
18215 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18216 * tree-ssa-copy.c (dump_copy_of): Likewise.
18217 (copy_prop_visit_cond_stmt): Likewise.
18218 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18219 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18220 (record_equivalences_from_stmt): Likewise.
18221 * tree-ssa-dse.c (compute_trims): Likewise.
18222 (delete_dead_call): Likewise.
18223 (delete_dead_assignment): Likewise.
18224 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18225 (forward_propagate_into_cond): Likewise.
18226 (pass_forwprop::execute): Likewise.
18227 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18228 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18230 (move_computations_worker): Likewise.
18231 (execute_sm): Likewise.
18232 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18233 (remove_exits_and_undefined_stmts): Likewise.
18234 (remove_redundant_iv_tests): Likewise.
18235 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18236 (adjust_iv_update_pos): Likewise.
18237 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18238 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18239 (value_replacement): Likewise.
18240 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18241 * tree-ssa-pre.c (print_pre_expr): Likewise.
18242 (get_representative_for): Likewise.
18243 (create_expression_by_pieces): Likewise.
18244 (insert_into_preds_of_block): Likewise.
18245 (eliminate_insert): Likewise.
18246 (eliminate_dom_walker::before_dom_children): Likewise.
18247 (eliminate): Likewise.
18248 (remove_dead_inserted_code): Likewise.
18249 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18250 * tree-ssa-reassoc.c (get_rank): Likewise.
18251 (eliminate_duplicate_pair): Likewise.
18252 (eliminate_plus_minus_pair): Likewise.
18253 (eliminate_not_pairs): Likewise.
18254 (undistribute_ops_list): Likewise.
18255 (eliminate_redundant_comparison): Likewise.
18256 (update_range_test): Likewise.
18257 (optimize_range_tests_var_bound): Likewise.
18258 (optimize_vec_cond_expr): Likewise.
18259 (rewrite_expr_tree): Likewise.
18260 (rewrite_expr_tree_parallel): Likewise.
18261 (linearize_expr): Likewise.
18262 (break_up_subtract): Likewise.
18263 (linearize_expr_tree): Likewise.
18264 (attempt_builtin_powi): Likewise.
18265 (attempt_builtin_copysign): Likewise.
18266 (transform_stmt_to_copy): Likewise.
18267 (transform_stmt_to_multiply): Likewise.
18268 (dump_ops_vector): Likewise.
18269 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18270 (print_scc): Likewise.
18271 (set_ssa_val_to): Likewise.
18272 (visit_reference_op_store): Likewise.
18273 (visit_use): Likewise.
18274 (sccvn_dom_walker::before_dom_children): Likewise.
18275 (run_scc_vn): Likewise.
18276 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18278 (expr_hash_elt::print): Likewise.
18279 (const_and_copies::pop_to_marker): Likewise.
18280 (const_and_copies::record_const_or_copy_raw): Likewise.
18281 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18282 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18283 (dump_predicates): Likewise.
18284 (find_uninit_use): Likewise.
18285 (warn_uninitialized_phi): Likewise.
18286 (pass_late_warn_uninitialized::execute): Likewise.
18287 * tree-ssa.c (verify_vssa): Likewise.
18288 (verify_ssa): Likewise.
18289 (maybe_optimize_var): Likewise.
18290 * tree-vrp.c (dump_value_range): Likewise.
18291 (dump_all_value_ranges): Likewise.
18292 (dump_asserts_for): Likewise.
18293 (register_edge_assert_for_2): Likewise.
18294 (vrp_visit_cond_stmt): Likewise.
18295 (vrp_visit_switch_stmt): Likewise.
18296 (vrp_visit_stmt): Likewise.
18297 (vrp_visit_phi_node): Likewise.
18298 (simplify_cond_using_ranges_1): Likewise.
18299 (fold_predicate_in): Likewise.
18300 (evrp_dom_walker::before_dom_children): Likewise.
18301 (evrp_dom_walker::push_value_range): Likewise.
18302 (evrp_dom_walker::pop_value_range): Likewise.
18303 (execute_early_vrp): Likewise.
18305 2017-05-16 Richard Biener <rguenther@suse.de>
18307 * dwarf2out.c (loc_list_from_tree_1): Do not create
18308 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18310 2017-05-16 Richard Biener <rguenther@suse.de>
18312 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18314 (note_variable_value_in_expr): If we resolved the decl ref
18315 do not push to the stack.
18317 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18319 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18320 operations in fast-math mode.
18321 (vaddq_f16): Likewise.
18322 (vmul_f16): Likewise.
18323 (vmulq_f16): Likewise.
18324 (vsub_f16): Likewise.
18325 (vsubq_f16): Likewise.
18326 * config/arm/neon.md (add<mode>3): New.
18328 (fma:<VH:mode>3): New. Also remove outdated comment.
18331 2017-05-16 Martin Liska <mliska@suse.cz>
18335 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18336 (odr_types_equivalent_p): Likewise.
18338 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18340 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18342 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18345 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18346 non-interunit SSE move alternatives with '?'.
18347 (zero-extendsidi peephole2): New peephole to skip intermediate
18348 general register in SSE zero-extend sequence.
18350 2017-05-15 Jeff Law <law@redhat.com>
18352 * reorg.c (relax_delay_slots): Create a new variable to hold
18353 the temporary target rather than clobbering TARGET_LABEL.
18355 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18356 missing argument to extract_bit_field call.
18357 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18359 2017-05-15 Martin Liska <mliska@suse.cz>
18362 * gcc.c (process_command): Do not allow empty argument of -o option.
18364 2017-05-15 Renlin Li <renlin.li@arm.com>
18366 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18367 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18368 * config/aarch64/constraints.md (Usf): Add long call check.
18369 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18370 (call_value): Likewise.
18371 (sibcall): Likewise.
18372 (sibcall_value): Likewise.
18374 (call_value_insn): New.
18375 (sibcall_insn): Update rtx pattern.
18376 (sibcall_value_insn): Likewise.
18377 (call_internal): Remove.
18378 (call_value_internal): Likewise.
18379 (sibcall_internal): Likewise.
18380 (sibcall_value_internal): Likewise.
18381 (call_reg): Likewise.
18382 (call_symbol): Likewise.
18383 (call_value_reg): Likewise.
18384 (call_value_symbol): Likewise.
18386 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18389 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18391 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18393 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18394 compatible with CCGOCmode and with CCZmode.
18396 2017-05-14 Martin Sebor <msebor@redhat.com>
18398 PR middle-end/77671
18399 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18400 (gimple_fold_builtin_snprintf): Same.
18401 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18402 (gimple_fold_builtin_snprintf): Same.
18403 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18404 of character types.
18405 (is_call_safe): New function.
18406 (try_substitute_return_value): Call it.
18407 (try_simplify_call): New function.
18408 (pass_sprintf_length::handle_gimple_call): Call it.
18410 2017-05-14 Martin Sebor <msebor@redhat.com>
18412 PR middle-end/80669
18413 * builtins.c (expand_builtin_stpncpy): Simplify.
18415 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18417 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18418 * config/i386/i386.h
18419 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18420 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18421 (struct machine_function): Add new members call_ms2sysv,
18422 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18423 (struct machine_frame_state): New fields sp_realigned and
18424 sp_realigned_offset.
18425 * config/i386/i386.c
18426 (enum xlogue_stub): New enum.
18427 (enum xlogue_stub_sets): New enum.
18428 (class xlogue_layout): New class.
18429 (struct ix86_frame): New fields stack_realign_allocate_offset,
18430 stack_realign_offset and outlined_save_offset. Modify comments to
18431 detail stack layout when using out-of-line stubs.
18432 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18433 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18434 -mcall-ms2sysv-xlogues.
18435 (stub_managed_regs): New static variable.
18436 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18437 registers managed by out-of-line stub.
18438 (disable_call_ms2sysv_xlogues): New function.
18439 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18440 m->call_ms2sysv when appropriate and compute frame layout for
18442 (sp_valid_at, fp_valid_at): New inline functions.
18443 (choose_basereg): New function.
18444 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18446 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18447 Use align parameter of choose_baseaddr to generated aligned SSE movs
18449 (pro_epilogue_adjust_stack): Modify to track
18450 machine_frame_state::sp_realigned.
18451 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18452 (ix86_nsaved_sseregs): Likewise.
18453 (ix86_emit_save_regs): Likewise.
18454 (ix86_emit_save_regs_using_mov): Likewise.
18455 (ix86_emit_save_sse_regs_using_mov): Likewise.
18456 (get_scratch_register_on_entry): Likewise.
18457 (gen_frame_set): New function.
18458 (gen_frame_load): Likewise.
18459 (gen_frame_store): Likewise.
18460 (emit_outlined_ms2sysv_save): Likewise.
18461 (emit_outlined_ms2sysv_restore): Likewise.
18462 (ix86_expand_prologue): Modify stack re-alignment code and call
18463 emit_outlined_ms2sysv_save when appropriate.
18464 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18465 parameter rtx_insn *insn, which allows the function to be used to only
18466 generate the notes.
18467 (ix86_expand_epilogue): Modify validity checks of frame and stack
18468 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18469 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18470 * config/i386/predicates.md
18471 (save_multiple): New predicate.
18472 (restore_multiple): Likewise.
18473 * config/i386/sse.md
18474 (save_multiple<mode>): New pattern.
18475 (save_multiple_realign<mode>): Likewise.
18476 (restore_multiple<mode>): Likewise.
18477 (restore_multiple_and_return<mode>): Likewise.
18478 (restore_multiple_leave_return<mode>): Likewise.
18479 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18481 2017-05-14 Julia Koval <julia.koval@intel.com>
18483 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18484 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18485 (__builtin_ia32_xsetbv): New builtins.
18486 * config/i386/i386.c (ix86_expand_special_args_builtin):
18488 (ix86_expand_builtin): Special expand for new intrinsics.
18489 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18490 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18491 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18493 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18495 * cfganal.c (inverted_post_order_compute): Change argument type
18497 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18498 * df-core.c (rest_of_handle_df_initialize): Adjust.
18499 (rest_of_handle_df_finish): Likewise.
18500 (df_analyze_1): Likewise.
18501 (df_analyze): Likewise.
18502 (loop_inverted_post_order_compute): Change argument to be a vec *.
18503 (df_analyze_loop): Adjust.
18504 (df_get_n_blocks): Likewise.
18505 (df_get_postorder): Likewise.
18506 * df.h (struct df_d): Change field to be a vec.
18507 * lcm.c (compute_laterin): Adjust.
18508 (compute_available): Likewise.
18509 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18510 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18511 * tree-ssa-pre.c (compute_antic): Likewise.
18513 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18515 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18516 (depth_first_search::depth_first_search): Change structure init
18517 function to this constructor.
18518 (depth_first_search::add_bb): Rename function to this member.
18519 (depth_first_search::execute): Likewise.
18520 (flow_dfs_compute_reverse_finish): Adjust.
18522 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18524 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18525 (longest_simple_path): Likewise.
18526 * shrink-wrap.c (spread_components): Likewise.
18527 (disqualify_problematic_components): Likewise.
18528 (emit_common_heads_for_components): Likewise.
18529 (emit_common_tails_for_components): Likewise.
18530 (insert_prologue_epilogue_for_components): Likewise.
18532 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18534 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18537 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18539 * df-core.c (df_set_blocks): Start using auto_bitmap.
18540 (df_compact_blocks): Likewise.
18541 * df-problems.c (df_rd_confluence_n): Likewise.
18542 * df-scan.c (df_insn_rescan_all): Likewise.
18543 (df_process_deferred_rescans): Likewise.
18544 (df_update_entry_block_defs): Likewise.
18545 (df_update_exit_block_uses): Likewise.
18546 (df_entry_block_bitmap_verify): Likewise.
18547 (df_exit_block_bitmap_verify): Likewise.
18548 (df_scan_verify): Likewise.
18549 * lra-constraints.c (lra_constraints): Likewise.
18550 (undo_optional_reloads): Likewise.
18551 (lra_undo_inheritance): Likewise.
18552 * lra-remat.c (calculate_gen_cands): Likewise.
18553 (do_remat): Likewise.
18554 * lra-spills.c (assign_spill_hard_regs): Likewise.
18555 (spill_pseudos): Likewise.
18556 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18557 (bitmap_set_subtract_values): Likewise.
18559 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18561 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18562 management with auto_bitmap.
18563 (fix_inter_tick): Likewise.
18564 (fix_recovery_deps): Likewise.
18565 * ira.c (add_store_equivs): Likewise.
18566 (find_moveable_pseudos): Likewise.
18567 (split_live_ranges_for_shrink_wrap): Likewise.
18568 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18569 (rtx_reuse_manager::seen_def_p): Likewise.
18570 (rtx_reuse_manager::set_seen_def): Likewise.
18571 * print-rtl.h (class rtx_reuse_manager): Likewise.
18573 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18575 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18577 (migrate_btr_def): Likewise.
18578 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18579 * df-core.c (loop_post_order_compute): Likewise.
18580 (loop_inverted_post_order_compute): Likewise.
18581 * hsa-common.h: Likewise.
18582 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18583 * init-regs.c (initialize_uninitialized_regs): Likewise.
18584 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18585 (inline_small_functions): Likewise.
18586 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18587 * ira.c (combine_and_move_insns): Likewise.
18588 (build_insn_chain): Likewise.
18589 * loop-invariant.c (find_invariants): Likewise.
18590 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18591 * predict.c (tree_predict_by_opcode): Likewise.
18592 (predict_paths_leading_to): Likewise.
18593 (predict_paths_leading_to_edge): Likewise.
18594 (estimate_loops_at_level): Likewise.
18595 (estimate_loops): Likewise.
18596 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18597 (spread_components): Likewise.
18598 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18599 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18600 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18601 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18602 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18603 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18604 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18605 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18606 (mark_threaded_blocks): Likewise.
18607 (thread_through_all_blocks): Likewise.
18608 * tree-ssa.c (verify_ssa): Likewise.
18609 (execute_update_addresses_taken): Likewise.
18610 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18612 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18614 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18616 (post_order_compute): Likewise.
18617 (inverted_post_order_compute): Likewise.
18618 (pre_and_rev_post_order_compute_fn): Likewise.
18620 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18622 * genrecog.c (int_set::int_set): Explicitly construct our
18623 auto_vec base class.
18624 * vec.h (auto_vec::auto_vec): New constructor.
18626 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18628 * bitmap.h (class auto_bitmap): New constructor taking
18629 bitmap_obstack * argument.
18631 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18633 * bitmap.h (class auto_bitmap): Change type of m_bits to
18634 bitmap_head, and adjust ctor / dtor and member operators.
18636 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18638 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18639 when returned register mode doesn't match original mode.
18641 2017-05-12 Jeff Law <law@redhat.com>
18642 Jakub Jelinek <jakub@redhat.com>
18644 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18645 we look for cc setter after the compare-elim changes.
18646 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18647 within the vector to match what compare-elim now expects.
18648 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18649 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18651 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18652 after the compare-elim changes.
18653 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18654 the vector to match what compare-elim now expects.
18655 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18656 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18657 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18658 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18659 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18661 * config/visium/visium.c (single_set_and_flags): Fix where
18662 we look for cc setter after the compare-elim changes.
18663 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18664 with the vector to match what compare-elim now expects.
18665 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18666 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18667 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18668 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18669 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18670 (neg<mode>2_insn_set_overflow): Likewise.
18672 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18674 PR middle-end/79794
18675 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18676 maybe_expand_insn call, set ops[0].target. If still set after call,
18677 set alt_rtl. Add extra arg to recursive calls.
18678 (extract_bit_field): Add alt_rtl argument. Pass to
18680 * expmed.h (extract_bit_field): Fix prototype.
18681 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18682 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18683 to extract_bit_field_calls.
18684 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18685 Pass alt_rtl to extract_bit_field calls.
18686 * calls.c (store_unaligned_arguments_into_psuedos)
18687 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18688 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18690 * optabs.h (struct expand_operand): Add target bitfield.
18692 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18694 * compare-elim.c (try_eliminate_compare): Canonicalize
18695 operation with embedded compare to
18696 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18697 (set (reg) (operation)].
18699 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18701 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18704 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18705 cost of adding a carry flag for ADC instruction.
18706 [case MINUS]: Ignore the cost of subtracting a carry flag
18707 for SBB instruction.
18709 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18711 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18713 * config/rs6000/bmiintrin.h: New file.
18714 * config/rs6000/bmi2intrin.h: New file.
18715 * config/rs6000/x86intrin.h: New file.
18717 2017-05-12 Jeff Law <law@redhat.com>
18719 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18722 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18724 PR middle-end/80707
18725 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18727 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18729 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18730 early expansion of vector divide builtins.
18731 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18732 builtins identified as having unsigned arguments.
18734 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18736 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18737 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18738 expansion of vector logical operations (and, andc, or, xor,
18741 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18743 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18744 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18746 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18748 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18749 early GIMPLE expansion of vector multiplies.
18751 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18753 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18754 TARGET_HAVE_MOVT conditional.
18755 (movt splitter): Likewise.
18757 2017-05-12 Richard Biener <rguenther@suse.de>
18759 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18760 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18761 Fold all stmts not inplace.
18763 2017-05-12 Richard Biener <rguenther@suse.de>
18765 PR tree-optimization/80713
18766 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18767 inserted_exprs bit for not removed stmts.
18769 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18771 PR middle-end/69921
18772 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18773 parallelized" attribute for parallelized OpenACC kernels.
18774 * omp-offload.c (execute_oacc_device_lower): Use it.
18776 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18777 Set "oacc kernels" attribute.
18778 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18779 parameter. Adjust all users.
18780 (oacc_fn_attrib_kernels_p): Remove function.
18781 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18782 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18783 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18784 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18785 assert "oacc kernels" attribute is set.
18787 2017-05-11 Carl Love <cel@us.ibm.com>
18789 * config/rs6000/rs6000-c: Add support for built-in functions
18790 vector unsigned char vec_popcnt (vector signed char)
18791 vector unsigned char vec_popcnt (vector unsigned char)
18792 vector unsigned short vec_popcnt (vector signed short)
18793 vector unsigned short vec_popcnt (vector unsigned short)
18794 vector unsigned int vec_popcnt (vector signed int)
18795 vector unsigned int vec_popcnt (vector unsigned int)
18796 vector unsigned long long vec_popcnt (vector signed long long)
18797 vector unsigned long long vec_popcnt (vector unsigned long long)
18798 vector signed long long vec_slo (vector signed long long,
18799 vector signed char)
18800 vector signed long long vec_slo (vector signed long long,
18801 vector unsigned char)
18802 vector unsigned long long vec_slo (vector unsigned long long,
18803 vector signed char)
18804 vector unsigned long long vec_slo (vector unsigned long long,
18805 vector unsigned char)
18806 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18807 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18808 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18809 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18810 * doc/extend.texi: Update the built-in documentation file for the
18811 new built-in functions.
18813 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18815 * attribs.h (sorted_attr_string): Move machine independent
18816 functions for target clone support from the i386 port to common
18817 code. Rename ix86_function_versions to common_function_versions.
18818 Rename make_name to make_unique_name.
18819 (common_function_versions): Likewise.
18820 (make_unique_name): Likewise.
18821 (make_dispatcher_decl): Likewise.
18822 (is_function_default_version): Likewise.
18823 * attribs.c (attr_strcmp): Likewise.
18824 (sorted_attr_string): Likewise.
18825 (common_function_versions): Likewise.
18826 (make_unique_name): Likewise.
18827 (make_dispatcher_decl): Likewise.
18828 (is_function_default_version): Likewise.
18829 * config/i386/i386.c (attr_strcmp): Likewise.
18830 (sorted_attr_string): Likewise.
18831 (ix86_function_versions): Likewise.
18832 (make_name): Likewise.
18833 (make_dispatcher_decl): Likewise.
18834 (is_function_default_version): Likewise.
18835 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18837 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18840 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
18841 Account for direct move costs for vec_construct of integer
18844 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
18847 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
18848 (UNSPEC_STX_ATOMIC): Ditto.
18849 (loaddi_via_sse): New insn.
18850 (storedi_via_sse): Ditto.
18851 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
18852 Update corresponding peephole2 patterns.
18853 (atomic_storedi_fpu): Ditto.
18855 2017-05-11 Julia Koval <julia.koval@intel.com>
18857 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
18858 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
18860 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
18861 (__builtin_ia32_rsqrt14ss_mask): New builtins.
18862 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
18864 2017-05-11 Nathan Sidwell <nathan@acm.org>
18866 * graphite-poly.c: Include dumpfile.h.
18868 * dumpfle.h (dump_function): Declare here ...
18869 * tree-dump.h (dump_function): ... not here.
18870 * dumpfile.c: #include tree-cfg.h.
18871 (dump_function): Move here from ...
18872 * tree-dump.c (dump_function): ... here.
18873 * gimplify.c: #include splay-tree.h, not tree-dump.h.
18874 * graphite-poly.c: Don't include tree-dump.h.
18875 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
18876 * print-tree.c: Likewise.
18877 * stor-layout.c: Likewise.
18878 * tree-nested.c: Likewise.
18880 * dumpfile.c (dump_start): Use TDF_FLAGS.
18881 (dump_enable_all): Fix TDF_KIND check thinko.
18883 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18885 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18886 array entries to represent two legal parameterizations of the
18887 overloaded __builtin_cmpb function, as represented by the
18888 P6_OV_BUILTIN_CMPB constant.
18889 (altivec_resolve_overloaded_builtin): Add special case handling
18890 for the __builtin_cmpb function, as represented by the
18891 P6_OV_BUILTIN_CMPB constant.
18892 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
18893 (BU_P6_64BIT_2): New macro.
18894 (BU_P6_OVERLOAD_2): New macro
18895 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
18896 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
18897 (CMPB): Add overload support to represent both 32-bit and 64-bit
18898 compare-bytes function.
18899 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
18900 support for TARGET_CMPB.
18901 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
18902 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18903 documentation of the __builtin_cmpb overloaded built-in function.
18905 2017-05-11 Richard Biener <rguenther@suse.de>
18907 PR tree-optimization/80705
18908 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
18909 bases are not vectorizable.
18911 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18913 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
18914 when counting register pressure.
18916 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18918 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
18919 (iv_ca_more_deps): Renamed to ...
18920 (iv_ca_compare_deps): ... this.
18921 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
18923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18925 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
18927 (determine_group_iv_costs): ... here.
18928 (find_inv_vars_cb): Record inv var if it's not recorded before.
18930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18932 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
18933 (get_shiftadd_cost): Ditto.
18935 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18937 * tree-ssa-address.c: Include header file.
18938 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
18940 (add_to_parts): Refactor.
18941 (addr_to_parts): New parameter. Update use of move_hint_to_base.
18942 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
18945 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18947 PR tree-optimization/53090
18948 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
18950 (extract_cond_operands): Detect condition with IV on both sides
18951 and return COMP_IV_EXPR_2.
18952 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
18953 (rewrite_use_compare): Simplify by removing call to function
18954 extract_cond_operands.
18956 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18958 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
18959 (extract_cond_operands): Detect condition comparing against non-
18960 invariant bound and return appropriate enum value.
18961 (find_interesting_uses_cond): Update use of extract_cond_operands.
18962 Handle its return value accordingly.
18963 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
18965 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18967 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
18968 nonlinear iv_use computation in loop invariant sensitive way.
18970 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18972 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
18973 (find_iv_candidates): Call relate_compare_use_with_all_cands.
18975 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18977 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
18978 (dump_cand): Support iv_cand.inv_exprs.
18979 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
18981 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
18984 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18986 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
18988 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
18989 as local function. Include necessary header files.
18990 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
18992 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18994 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
18996 2017-05-11 Bin Cheng <bin.cheng@arm.com>
18998 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
18999 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19000 RSHIFT_EXPR and BIT_NOT_EXPR.
19002 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19004 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19005 (adjust_setup_cost): New parameter supporting round up adjustment.
19006 (struct address_cost_data): Delete.
19007 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19008 (split_address_cost, ptr_difference_cost): Delete.
19009 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19010 (struct ainc_cost_data): New struct.
19011 (get_address_cost_ainc): New function.
19012 (get_address_cost, get_computation_cost): Reimplement.
19013 (determine_group_iv_cost_address): Record inv_expr for all uses of
19015 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19016 (iv_ca_has_deps): Reimplemented to ...
19017 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19019 (iv_ca_extend): Call iv_ca_more_deps.
19021 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19023 * tree-ssa-address.c (struct mem_address): Move to header file.
19024 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19025 * tree-ssa-address.h (struct mem_address): Move from C file.
19026 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19028 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19030 * tree-affine.h (aff_combination_type): New interface.
19031 (aff_combination_zero_p): Remove static.
19032 (aff_combination_const_p): New interface.
19033 (aff_combination_singleton_var_p): New interfaces.
19035 2017-05-11 Richard Biener <rguenther@suse.de>
19037 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19038 Skip unreachable blocks and destinations.
19039 (eliminate): Move stmt removal and fixup ...
19040 (fini_eliminate): ... here. Skip inserted exprs.
19041 (pass_pre::execute): Move fini_pre after fini_eliminate.
19042 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19043 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19044 PRE to get rid of dead code that has invalid SSA form and
19045 split critical edges again.
19047 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19049 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19051 2017-05-11 Richard Biener <rguenther@suse.de>
19053 * passes.c (execute_function_todo): Verify loops if they are
19054 said to be up-to-date.
19055 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19056 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19058 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19061 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19062 handle calling assemble_external ourself.
19065 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19066 modes with zero size. Enhance comment.
19068 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19070 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19071 built-ins for vec_xl and vec_xst with short and char pointer
19074 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19076 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19077 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19078 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19079 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19080 (_mm_maskz_min_round_ss): New intrinsics.
19081 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19082 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19083 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19084 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19085 (__builtin_ia32_minss_mask_round): New builtins.
19086 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19087 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19088 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19090 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19091 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19093 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19096 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19098 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19099 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19100 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19101 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19102 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19103 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19104 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19105 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19106 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19107 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19108 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19109 (__builtin_ia32_mulss_mask_round): New builtins.
19110 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19111 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19112 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19114 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19115 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19117 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19120 2017-05-10 Julia Koval <julia.koval@intel.com>
19122 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19123 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19124 (_mm256_setr_m128i): New intrinsics.
19126 2017-05-10 Julia Koval <julia.koval@intel.com>
19128 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19129 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19130 (_mm_maskz_rcp14_ss): New intrinsics.
19131 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19132 (__builtin_ia32_rcp14ss_mask): New builtins.
19133 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19135 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19137 PR tree-optimization/51513
19138 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19139 (assert_unreachable_fallthru_edge_p): Use it.
19140 (group_case_labels_stmt): Likewise.
19141 * tree-cfg.h: Prototype it.
19142 * stmt.c: Include cfghooks.h and tree-cfg.h.
19143 (emit_case_dispatch_table) <gap_label>: New local variable.
19144 Use it to fill dispatch table gaps.
19145 Test for default_label before updating probabilities.
19146 (expand_case) <default_label>: Remove unneeded initialization.
19147 Test for unreachable default case statement and remove its edge.
19148 Set default_label accordingly.
19149 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19151 2017-05-10 Carl Love <cel@us.ibm.com>
19153 * config/rs6000/rs6000-c: Add support for built-in functions
19154 vector signed char vec_neg (vector signed char)
19155 vector signed short int vec_neg (vector short int)
19156 vector signed int vec_neg (vector signed int)
19157 vector signed long long vec_neg (vector signed long long)
19158 vector float vec_neg (vector float)
19159 vector double vec_neg (vector double)
19160 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19162 * config/rs6000/altivec.h: Add define for vec_neg
19163 * doc/extend.texi: Update the built-in documentation for the
19164 new built-in functions.
19166 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19168 PR tree-optimization/77644
19169 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19171 2017-05-10 Nathan Sidwell <nathan@acm.org>
19173 * dumpfile.h (TDI_lang_all): New.
19174 (TDF_KIND): New. Renumber others
19175 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19177 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19179 (get_dump_file_name): Adjust suffix generation.
19180 (dump_enable_all): Use TDF_KIND.
19181 * doc/invoke.texi (-fdump-lang-all): Document.
19183 * dumpfile.h: Tabify.
19185 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19188 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19189 Move member access before delete.
19191 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19193 * tree-inline.c (expand_call_inline): Split block at stmt
19196 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19199 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19200 are now unused after splitting mov{sf,sd}_hardfloat.
19201 (f32_lr2): Likewise.
19202 (f32_lm): Likewise.
19203 (f32_lm2): Likewise.
19204 (f32_li): Likewise.
19205 (f32_li2): Likewise.
19206 (f32_lv): Likewise.
19207 (f32_sr): Likewise.
19208 (f32_sr2): Likewise.
19209 (f32_sm): Likewise.
19210 (f32_sm2): Likewise.
19211 (f32_si): Likewise.
19212 (f32_si2): Likewise.
19213 (f32_sv): Likewise.
19214 (f32_dm): Likewise.
19215 (f32_vsx): Likewise.
19216 (f32_av): Likewise.
19217 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19218 For movsf, order stores so the VSX stores occur before the GPR
19219 store which encourages the register allocator to use a traditional
19220 FPR instead of a GPR. For movsd, order the stores so that the GPR
19221 store comes before the VSX stores to allow the power6 to work.
19222 This is due to the power6 not having a 32-bit integer store
19223 instruction from a FPR.
19224 (movsf_hardfloat): Likewise.
19225 (movsd_hardfloat): Likewise.
19227 2017-05-09 Martin Sebor <msebor@redhat.com>
19229 PR translation/80280
19230 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19233 PR translation/80280
19234 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19235 data member added in r247778.
19236 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19238 2017-05-09 Nathan Sidwell <nathan@acm.org>
19240 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19242 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19245 2017-05-09 Marek Polacek <polacek@redhat.com>
19247 * doc/invoke.texi: Fix typo.
19249 2017-05-09 Richard Biener <rguenther@suse.de>
19251 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19252 (vrp_val_is_min): Likewise.
19253 (set_value_range_to_value): Likewise.
19254 (set_value_range_to_nonnegative): Likewise.
19255 (gimple_assign_nonzero_p): Likewise.
19256 (gimple_stmt_nonzero_p): Likewise.
19257 (vrp_int_const_binop): Likewise. Remove unreachable case.
19258 (adjust_range_with_scev): Adjust comments.
19259 (compare_range_with_value): Likewise.
19260 (extract_range_from_phi_node): Likewise.
19261 (test_for_singularity): Likewise.
19263 2017-05-09 Richard Biener <rguenther@suse.de>
19265 * tree-vrp.c (get_single_symbol): Add assert that we don't
19266 get overflowed constants as invariant part.
19267 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19268 checks. Use wi::cmp instead of recursing for integer constants.
19269 (compare_values): Just ignore whether we assumed undefined
19270 overflow instead of failing the compare.
19271 (extract_range_for_var_from_comparison_expr): Add comment before the
19272 TREE_NO_WARNING sets.
19273 (test_for_singularity): Likewise.
19274 (extract_range_from_comparison): Do not disable optimization
19275 when we assumed undefined overflow.
19276 (extract_range_basic): Remove init of unused var.
19278 2017-05-09 Richard Biener <rguenther@suse.de>
19280 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19281 (extract_range_from_multiplicative_op_1): Adjust.
19282 (extract_range_from_binary_expr_1): Use int_const_binop.
19284 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19287 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19288 rs6000_store_data_bypass_p in seven define_bypass directives and
19289 in several comments.
19290 * config/rs6000/rs6000-protos.h: Add prototype for
19291 rs6000_store_data_bypass_p function.
19292 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19293 function implements slightly different (rs6000-specific) semantics
19294 than store_data_bypass_p, returning false rather than aborting
19295 with assertion error when arguments do not satisfy the
19296 requirements of store data bypass.
19297 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19298 rs6000_store_data_bypass_p.
19300 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19302 * config/xtensa/xtensa-protos.h
19303 (xtensa_initial_elimination_offset): New declaration.
19304 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19305 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19306 macro definition, add case for FRAME_POINTER_REGNUM when
19307 FRAME_GROWS_DOWNWARD.
19308 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19309 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19310 xtensa_initial_elimination_offset.
19312 2017-05-08 Nathan Sidwell <nathan@acm.org>
19314 * doc/invoke.texi: Alphabetize -fdump options.
19316 2017-05-08 Martin Sebor <msebor@redhat.com>
19318 PR translation/80280
19319 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19321 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19323 * target.def (compute_frame_layout): New optional target hook.
19324 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19325 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19326 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19328 * reload1.c (verify_initial_elim_offsets): Likewise.
19329 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19330 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19331 (arm_get_frame_offsets): Split up into this ...
19332 (arm_compute_frame_layout): ... and this function.
19334 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19336 * config/aarch64/constraints.md (Usa): New constraint.
19337 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19339 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19341 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19342 with_multilib_list after it has been checked.
19344 2017-05-08 Richard Biener <rguenther@suse.de>
19346 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19347 (bitmap_set_subtract_values): Likewise.
19349 2017-05-08 Richard Biener <rguenther@suse.de>
19351 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19352 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19354 (gimple_stmt_nonzero_warnv_p): Rename to ...
19355 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19357 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19358 (extract_range_basic): Adjust, do not disable propagation on
19359 strict overflow sensitive simplification.
19360 (vrp_visit_cond_stmt): Likewise.
19362 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19364 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19365 body size unconditionally.
19367 2017-05-07 Jeff Law <law@redhat.com>
19370 2017-05-06 Jeff Law <law@redhat.com>
19371 PR tree-optimization/78496
19372 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19375 PR tree-optimization/78496
19376 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19377 (simplify_stmt_using_ranges): Call it.
19378 (vrp_dom_walker::before_dom_children): Extract equivalences
19379 from an ASSERT_EXPR with an equality comparison against a
19382 2017-05-06 Jeff Law <law@redhat.com>
19384 PR tree-optimization/78496
19385 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19388 PR tree-optimization/78496
19389 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19390 (simplify_stmt_using_ranges): Call it.
19391 (vrp_dom_walker::before_dom_children): Extract equivalences
19392 from an ASSERT_EXPR with an equality comparison against a
19395 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19397 * lra-constraints.c (lra_copy_reg_equiv): New function.
19398 (split_reg): Use it to copy equivalence information from the
19399 original register to the spill register.
19401 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19403 PR rtl-optimization/75964
19404 * simplify-rtx.c (simplify_const_relational_operation): Remove
19405 invalid handling of comparisons of integer ABS.
19407 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19409 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19410 initialize to zero.
19411 (init_regs): Remove declaration.
19412 (function_arg_advance_32): Initialize error_p as boolean variable.
19414 2017-05-05 Nathan Sidwell <nathan@acm.org>
19416 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19417 lines. Use for (;;).
19419 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19421 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19422 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19423 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19424 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19425 VF=2 that require versioning.
19427 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19429 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19432 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19434 * diagnostic.h (diagnostic_override_option_index): Convert from
19435 macro to inline function.
19437 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19439 * diagnostic.c (last_module_changed_p): New function.
19440 (set_last_module): New function.
19441 (diagnostic_report_current_module): Convert macro usage to
19442 the above functions.
19443 * diagnostic.h (diagnostic_context::last_module): Strengthen
19444 from const line_map * to const line_map_ordinary *.
19445 (diagnostic_last_module_changed): Delete macro.
19446 (diagnostic_set_last_module): Delete macro.
19448 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19450 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19451 with diagnostic_report_diagnostic.
19452 (diagnostic_n_impl_richloc): Likewise.
19453 * diagnostic.h (report_diagnostic): Delete macro.
19454 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19455 with diagnostic_report_diagnostic.
19456 * substring-locations.c (format_warning_va): Likewise.
19458 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19460 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19461 save/restor of format_spec. Move option-printing code to...
19462 (print_option_information): ...this new function, and
19463 reimplement by simply printing to the pretty_printer,
19464 rather than appending to the format string.
19466 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19468 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19469 handling logic into...
19470 (update_effective_level_from_pragmas): ...this new function.
19472 2017-05-04 Andrew Waterman <andrew@sifive.com>
19474 * config/riscv/riscv.opt (mstrict-align): New option.
19475 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19476 (SLOW_UNALIGNED_ACCESS): Define.
19477 (riscv_slow_unaligned_access): Declare.
19478 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19480 (riscv_slow_unaligned_access): New variable.
19481 (rocket_tune_info): Set slow_unaligned_access to true.
19482 (optimize_size_tune_info): Set slow_unaligned_access to false.
19483 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19484 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19485 (riscv_option_override): Set riscv_slow_unaligned_access.
19486 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19488 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19490 * config/riscv/riscv.md: Unify indentation.
19492 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19497 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19499 (extendsi<mode>2): Add support for doing sign extension via
19500 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19501 don't have ISA 3.0 instructions.
19502 (extendsi<mode>2 splitter): Likewise.
19503 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19504 generate the normal insns since SImode can now go in vector
19505 registers. Disallow the special UNSPECs needed for previous
19506 machines to hide SImode being used. Add new insns
19507 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19508 (fix_trunc<mode>si2_stfiwx): Likewise.
19509 (fix_trunc<mode>si2_internal): Likewise.
19510 (fixuns_trunc<mode>si2): Likewise.
19511 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19512 (fctiw<u>z_<mode>_smallint): Likewise.
19513 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19514 of floating point to 32-bit integer from doing a direct move to
19515 the GPR registers to do a store.
19516 (fctiwz_<mode>): Break long line.
19518 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19520 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19521 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19522 (addr_list, addr_offset_valid_p): New.
19523 (split_address_groups): Check offset validity with above function.
19524 (gt-tree-ssa-loop-ivopts.h): Include header file.
19526 2017-05-05 Nathan Sidwell <nathan@acm.org>
19528 * config.gcc (arm*-*-*): Add missing 'fi'.
19530 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19532 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19533 included in -fopt-info does not matter.
19534 * doc/optinfo.texi (-fopt-info): Fix description of default
19535 behavour. Explicitly say order of options included in -fopt-info
19538 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19540 * config.gcc: Allow combinations of aprofile and rmprofile values for
19541 --with-multilib-list.
19542 * config/arm/t-multilib: New file.
19543 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19544 variables. Remove setting of ISA and floating-point ABI in
19545 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19546 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19547 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19548 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19550 * config/arm/t-rmprofile: Likewise except for the matches changes.
19551 * doc/install.texi (--with-multilib-list): Document the combination of
19552 aprofile and rmprofile values and warn about pitfalls in doing that.
19554 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19556 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19557 (movdi_aarch64): Likewise.
19559 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19561 PR tree-optimization/80632
19562 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19564 (build_arrays): Initialize it for virtual phis.
19565 (fix_phi_nodes): Use it for virtual phis.
19567 PR tree-optimization/80558
19568 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19569 [x, y] op z into [x op, y op z] for op & or | if conditions
19572 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19573 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19576 * config/arm/arm.md (use_literal_pool): Remove.
19577 (64-bit immediate split): No longer takes cost into consideration
19578 if arm_disable_literal_pool is enabled.
19579 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19580 used when arm_disable_literal_pool is enabled.
19581 (arm_max_const_double_inline_cost): Remove use of
19582 arm_disable_literal_pool.
19583 (push_minipool_fix): Add assert.
19584 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19585 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19586 (no_literal_pool_sf_immediate): New.
19588 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19590 PR tree-optimization/80613
19591 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19592 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19594 2017-05-05 Richard Biener <rguenther@suse.de>
19596 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19598 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19600 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19601 of this flag from insn conditions due to removal from r247495.
19603 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19605 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19607 (arm_early_store_addr_dep_ptr): Likewise.
19608 * config/arm/aarch-common-protos.h
19609 (arm_early_load_addr_dep_ptr): Add prototype.
19610 (arm_early_store_addr_dep_ptr): Likewise.
19611 * config/arm/cortex-a53.md: Add new bypasses.
19613 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19615 * tree.c (next_type_uid): Change type to unsigned.
19616 (type_hash_canon): Decrement back next_type_uid if
19617 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19618 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19621 2017-05-04 Martin Sebor <msebor@redhat.com>
19623 * builtins.c: Fix a trivial typo in a comment.
19625 PR middle-end/79234
19626 * builtins.c (check_sizes): Adjust to handle reading past the end.
19627 Avoid printing excessive upper bound of ranges. Use %E to print
19628 tree nodes instead of converting them to %wu.
19629 (expand_builtin_memchr): New function.
19630 (compute_dest_size): Rename...
19631 (compute_objsize): ...to this.
19632 (expand_builtin_memcpy): Adjust.
19633 (expand_builtin_mempcpy): Adjust.
19634 (expand_builtin_strcat): Adjust.
19635 (expand_builtin_strcpy): Adjust.
19636 (check_strncat_sizes): Adjust.
19637 (expand_builtin_strncat): Adjust.
19638 (expand_builtin_strncpy): Adjust and simplify.
19639 (expand_builtin_memset): Adjust.
19640 (expand_builtin_bzero): Adjust.
19641 (expand_builtin_memcmp): Adjust.
19642 (expand_builtin): Handle memcmp.
19643 (maybe_emit_chk_warning): Check strncat just once.
19645 2017-05-04 Martin Sebor <msebor@redhat.com>
19647 PR preprocessor/79214
19648 PR middle-end/79222
19649 PR middle-end/79223
19650 * builtins.c (check_sizes): Add inlining context and issue
19651 warnings even when -Wno-system-headers is set.
19652 (check_strncat_sizes): Same.
19653 (expand_builtin_strncat): Same.
19654 (expand_builtin_memmove): New function.
19655 (expand_builtin_stpncpy): Same.
19656 (expand_builtin): Handle memmove and stpncpy.
19658 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19660 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19661 which is not used any more.
19663 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19665 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19667 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19669 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19670 (cortexa53_tunings): Likewise.
19671 (cortexa57_tunings): Likewise.
19672 (cortexa72_tunings): Likewise.
19673 (cortexa73_tunings): Likewise.
19675 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19677 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19678 Set loop alignment to 8.
19680 2017-05-04 Martin Sebor <msebor@redhat.com>
19682 PR translation/80280
19683 * builtins.c (expand_builtin_object_size): Add missing quoting to
19684 %D and like directives.
19685 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19686 (hsa_type_for_tree_type): Same.
19687 (verify_function_arguments): Same.
19688 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19689 * varasm.c (get_section): Same.
19692 2017-05-04 Martin Sebor <msebor@redhat.com>
19694 PR translation/80280
19695 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19697 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19699 * config/aarch64/aarch64.c (generic_addrcost_table):
19700 Change HI/TI mode setting.
19702 2017-05-04 Martin Jambor <mjambor@suse.cz>
19704 PR tree-optimization/80622
19705 * tree-sra.c (comes_initialized_p): New function.
19706 (build_accesses_from_assign): Only set write lazily when
19707 comes_initialized_p is false.
19708 (analyze_access_subtree): Use comes_initialized_p.
19709 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19710 instead of testing for PARM_DECL.
19712 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19714 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19715 constraint on operand 0 to allow more general addressing modes.
19716 Adjust output template.
19717 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19719 * config/aarch64/aarch64-protos.h
19720 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19721 * config/aarch64/constraints.md (Dp): New address constraint.
19722 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19725 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19727 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19728 update use of estimate_ipcp_clone_size_and_time.
19729 (estimate_local_effects): Update use of
19730 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19731 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19732 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19733 Return nonspecialized time.
19735 2017-05-04 Richard Biener <rguenther@suse.de>
19737 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19738 for the last VUSE which def dominates the PHI. Directly call
19740 (get_continuation_for_phi_1): Remove.
19742 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19744 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19745 to explain the use of truncating division. Cap the number of
19746 iterations to the maximum given by nb_iterations_upper_bound,
19749 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19751 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19752 * configure: Regenerate.
19753 * config.in: Regenerate.
19754 * config/i386/driver-mingw32.c: new file.
19755 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19756 * config.host: Link driver-mingw32.o on MinGW host.
19757 * doc/install.texi: Document new --enable-mingw-wildcard configure
19760 2017-05-04 Marek Polacek <polacek@redhat.com>
19762 PR tree-optimization/80612
19763 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19765 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19766 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19768 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19769 (movt splitter): Likewise.
19770 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19771 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19772 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19773 block for Thumb-1 with MOVT.
19774 (thumb2_legitimate_address_p): Move code block ...
19775 (can_avoid_literal_pool_for_label_p): ... into this new function.
19776 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19778 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19779 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19780 "M-profile targets with the MOVT instruction".
19782 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19784 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19785 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19786 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19788 2017-05-04 Martin Liska <mliska@suse.cz>
19790 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19791 variable cond_code.
19793 2017-05-04 Richard Biener <rguenther@suse.de>
19795 * tree.c (array_at_struct_end_p): Handle arrays at struct
19796 end with flexarrays more conservatively. Refactor and treat
19797 arrays of arrays or aggregates more strict. Fix
19798 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19799 * tree.c (array_at_struct_end_p): Adjust prototype.
19800 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19801 * gimple-fold.c (get_range_strlen): Likewise.
19802 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19804 2017-05-04 Richard Biener <rguenther@suse.de>
19806 PR tree-optimization/31130
19807 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19809 (supports_overflow_infinity): Likewise.
19810 (is_negative_overflow_infinity): Likewise.
19811 (is_positive_overflow_infinity): Likewise.
19812 (is_overflow_infinity): Likewise.
19813 (stmt_overflow_infinity): Likewise.
19814 (overflow_infinity_range_p): Likewise.
19815 (usable_range_p): Remove as always returning true.
19816 (make_overflow_infinity): Remove.
19817 (negative_overflow_infinity): Likewise.
19818 (positive_overflow_infinity): Likewise.
19819 (avoid_overflow_infinity): Likewise.
19820 (set_value_range): Adjust accordingly.
19821 (set_value_range_to_nonnegative): Likewise, remove now unused
19822 overflow_infinity arg.
19823 (vrp_operand_equal_p): Adjust.
19824 (update_value_range): Likewise.
19825 (range_int_cst_singleton_p): Likewise.
19826 (operand_less_p): Likewise.
19827 (compare_values_warnv): Likewise.
19828 (extract_range_for_var_from_comparison_expr): Likewise.
19829 (vrp_int_const_binop): Likewise.
19830 (zero_nonzero_bits_from_vr): Likewise.
19831 (extract_range_from_multiplicative_op_1): Likewise.
19832 (extract_range_from_binary_expr_1): Likewise.
19833 (extract_range_from_unary_expr): Likewise.
19834 (extract_range_from_comparison): Likewise.
19835 (extract_range_basic): Likewise.
19836 (adjust_range_with_scev): Likewise.
19837 (compare_ranges): Likewise.
19838 (compare_range_with_value): Likewise.
19839 (dump_value_range): Likewise.
19840 (test_for_singularity): Likewise, remove strict_overflow_p parameter
19842 (simplify_cond_using_ranges): Adjust.
19844 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19846 * brig-builtins.def: Added a builtin for class_f64.
19847 * builtin-types.def: Added a builtin type needed by class_f64.
19849 2017-05-03 Jason Merrill <jason@redhat.com>
19851 * timevar.def: Add TV_CONSTEXPR.
19853 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19855 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
19857 2017-05-03 Martin Jambor <mjambor@suse.cz>
19859 * ipa-prop.c (ipa_update_after_lto_read): Removed.
19860 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
19861 * ipa-cp.c (ipcp_propagate_stage): Do not call
19862 ipa_update_after_lto_read.
19863 * ipa-inline.c (ipa_inline): Likewise.
19865 2017-05-03 Martin Jambor <mjambor@suse.cz>
19867 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
19868 tag. Added a default constructor and a destructor.
19869 (ipa_edge_args_sum_t): New class;
19870 (ipa_edge_args_sum): Declare.
19871 (ipa_edge_args_vector): Remove declaration.
19872 (IPA_EDGE_REF): Use ipa_edge_args_sum.
19873 (ipa_free_edge_args_substructures): Remove declaration.
19874 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
19875 (ipa_edge_args_info_available_for_edge_p): Likewise.
19876 * ipa-prop.c (ipa_edge_args_vector): Removed.
19877 (edge_removal_hook_holder): Likewise.
19878 (edge_duplication_hook_holder): Likewise.
19879 (ipa_edge_args_sum): New variable.
19880 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
19881 ipa_edge_args_vector.
19882 (ipa_free_edge_args_substructures): Likewise.
19883 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
19884 ipa_edge_args_vector.
19885 (ipa_edge_removal_hook): Turned into method
19886 ipa_edge_args_sum_t::remove.
19887 (ipa_edge_duplication_hook): Turned into method
19888 ipa_edge_args_sum_t::duplicate.
19889 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
19890 registering edge hooks.
19891 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
19892 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
19893 ipa_edge_args_sum instead of ipa_edge_args_vector.
19894 * ipa-profile.c (ipa_profile): Likewise.
19896 2017-05-03 Martin Jambor <mjambor@suse.cz>
19898 * symbol-summary.h (function_summary): New method exists.
19899 (function_summary::symtab_removal): Deallocate through release.
19900 (call_summary): New class.
19901 (gt_ggc_mx): New overload.
19902 (gt_pch_nx): Likewise.
19903 (gt_pch_nx): Likewise.
19905 2017-05-03 Jeff Law <law@redhat.com>
19907 PR tree-optimization/78496
19908 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
19909 from simplify_cond_using_ranges. Split off code to walk
19910 backwards through casts into ...
19911 (simplify_cond_using_ranges_2): New function.
19912 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
19913 (execute_vrp): After identifying jump threads, call
19914 simplify_cond_using_ranges_2.
19916 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
19919 * ipa-inline.h (inline_summary): Add ctor.
19920 (create_ggc): Do not use ggc_cleared_alloc.
19922 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
19923 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19925 * gcc.c (handle_braces): Support escaping in switch matching
19927 * doc/invoke.texi (Spec Files): Document it.
19928 Remove superfluous @code markup in items.
19930 2017-05-03 David Malcolm <dmalcolm@redhat.com>
19932 * diagnostic-show-locus.c (struct column_range): New struct.
19933 (get_affected_columns): New function.
19934 (get_printed_columns): New function.
19935 (struct correction): New struct.
19936 (correction::ensure_capacity): New function.
19937 (correction::ensure_terminated): New function.
19938 (struct line_corrections): New struct.
19939 (line_corrections::~line_corrections): New dtor.
19940 (line_corrections::add_hint): New function.
19941 (layout::print_trailing_fixits): Reimplement in terms of the new
19943 (selftest::test_overlapped_fixit_printing): New function.
19944 (selftest::diagnostic_show_locus_c_tests): Call it.
19946 2017-05-03 Nathan Sidwell <nathan@acm.org>
19948 Canonicalize canonical type hashing
19949 * tree.h (type_hash_canon_hash): Declare.
19950 * tree.c (type_hash_list, attribute_hash_list): Move into
19951 type_hash_canon_hash.
19952 (build_type_attribute_qual_variant): Break out hash code calc into
19953 type_hash_canon_hash.
19954 (type_hash_canon_hash): New. Generic type hash computation.
19955 (build_range_type_1, build_array_type_1, build_function_type,
19956 build_method_type_directly, build_offset_type, build_complex_type,
19957 make_vector_type): Call it.
19959 2017-05-03 Richard Biener <rguenther@suse.de>
19961 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
19962 When all DRs have unknown misaligned do not always peel
19963 when there is a store but apply the same costing model as if
19964 there were only loads.
19966 2017-05-03 Richard Biener <rguenther@suse.de>
19969 PR tree-optimization/80492
19970 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
19971 compare_base_decls returning dont-know properly.
19973 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
19975 * config/arm/iterators.md (CCSI): New mode iterator.
19976 (arch): New mode attribute.
19977 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
19978 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
19979 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
19980 code iterator for success result mode.
19981 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
19982 the corresponding new insn generators.
19984 2017-05-03 Bin Cheng <bin.cheng@arm.com>
19987 2017-05-02 Bin Cheng <bin.cheng@arm.com>
19988 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19990 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
19992 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
19993 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
19994 (DDR_A): Wrap DDR argument in brackets.
19995 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
19996 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
19997 (DDR_REVERSED_P): Likewise.
19999 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20001 PR tree-optimization/79472
20002 * tree-switch-conversion.c (struct switch_conv_info): Add
20003 contiguous_range and default_case_nonstandard fields.
20004 (collect_switch_conv_info): Compute contiguous_range and
20005 default_case_nonstandard fields, don't clear final_bb if
20006 contiguous_range and only the default case doesn't have the required
20008 (check_all_empty_except_final): Set default_case_nonstandard instead
20009 of failing if contiguous_range and the default case doesn't have empty
20011 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20012 and only the default case doesn't have the required constants. Skip
20014 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20015 if default_case_nonstandard.
20016 (build_constructors): Build constant 1 just once. Assert that default
20017 values aren't inserted in between cases if contiguous_range. Skip
20019 (build_arrays): Skip virtual phis.
20020 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20021 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20022 Handle virtual phis.
20023 (gen_inbound_check): Handle default_case_nonstandard case.
20024 (process_switch): Adjust check_final_bb caller. Call
20025 gather_default_values with the first non-default case instead of
20026 default case if default_case_nonstandard.
20028 2017-05-02 Nathan Sidwell <nathan@acm.org>
20030 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20031 check. Fix formatting.
20033 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20035 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20036 errors when comparing specialized and unspecialized times.
20038 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20040 * diagnostic-show-locus.c
20041 (layout::should_print_annotation_line_p): Make private.
20042 (layout::print_annotation_line): Make private.
20043 (layout::annotation_line_showed_range_p): Make private.
20044 (layout::show_ruler): Make private.
20045 (layout::print_source_line): Make private. Pass in line and
20046 line_width, rather than calling location_get_source_line. Drop
20048 (layout::print_leading_fixits): New method.
20049 (layout::print_any_fixits): Rename to...
20050 (layout::print_trailing_fixits): ...this, and make private.
20051 Don't print newline fixits.
20052 (diagnostic_show_locus): Move logic for printing one row into...
20053 (layout::print_line): ...this new function. Move the
20054 location_get_source_line call and error-handling from
20055 print_source_line to here. Call print_leading_fixits, and rename
20056 print_any_fixits to print_trailing_fixits.
20057 (selftest::test_fixit_insert_containing_newline): Update now that
20058 newlines are partially supported.
20059 (selftest::test_fixit_insert_containing_newline_2): New test.
20060 (selftest::test_fixit_replace_containing_newline): Update comments.
20061 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20062 * edit-context.c (class added_line): New class.
20063 (class edited_line): Describe newline handling in comment.
20064 (edited_line::actually_edited_p): New method.
20065 (edited_line::print_content): Delete redundant decl.
20066 (edited_line::m_predecessors): New field.
20067 (edited_file::print_content): Call edited_line::print_content.
20068 (edited_file::print_diff): Update to support newlines.
20069 (edited_file::print_diff_hunk): Likewise.
20070 (edited_file::print_run_of_changed_lines): New function.
20071 (edited_file::print_diff_line): Convert to...
20072 (print_diff_line): ...this.
20073 (edited_file::get_effective_line_count): New function.
20074 (edited_line::edited_line): Initialize new field m_predecessors.
20075 (edited_line::~edited_line): Clean up m_predecessors.
20076 (edited_line::apply_fixit): Handle newlines.
20077 (edited_line::get_effective_line_count): New function.
20078 (edited_line::print_content): New function.
20079 (edited_line::print_diff_lines): New function.
20080 (selftest::test_applying_fixits_insert_containing_newline): New
20082 (selftest::test_applying_fixits_replace_containing_newline): New
20084 (selftest::insert_line): New function.
20085 (selftest::test_applying_fixits_multiple_lines): Add example of
20087 (selftest::edit_context_c_tests): Call the new tests.
20089 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20091 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20092 parameter cand. Update dump information.
20093 (get_computation_cost): Update uses.
20095 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20097 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20098 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20099 (get_computation_at, rewrite_use_address): Update use of
20100 get_computation_aff.
20102 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20104 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20105 (get_computation): Delete.
20106 (get_computation_cost): Implement like get_computation_cost_at.
20107 Use get_computation_at.
20108 (get_computation_cost_at): Delete.
20109 (rewrite_use_nonlinear_expr): Use get_computation_at.
20110 (rewrite_use_compare, remove_unused_ivs): Ditto.
20112 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20114 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20116 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20118 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20119 (ivopts_global_cost_for_size): Rename parameter and update uses.
20120 (iv_ca_recount_cost): Update uses.
20121 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20122 candidates seperately in n_invs and n_cands.
20123 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20125 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20127 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20128 (find_inv_vars_cb): New.
20129 (find_depends): Renamed to ...
20130 (find_inv_vars): ... this.
20131 (add_candidate_1, force_var_cost): Call find_inv_vars.
20132 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20134 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20136 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20137 inv_vars. Add inv_exprs.
20138 (struct iv_cand): Rename depends_on to inv_vars.
20139 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20140 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20141 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20142 (dump_cand): Dump inv_vars.
20143 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20144 (record_invariant, find_depends, add_candidate_1): Ditto.
20145 (set_group_iv_cost, force_var_cost): Ditto.
20146 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20147 (get_computation_cost_at, get_computation_cost): Ditto.
20148 (determine_group_iv_cost_generic): Ditto.
20149 (determine_group_iv_cost_address): Ditto.
20150 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20151 (determine_group_iv_costs): Ditto.
20152 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20153 (iv_ca_set_remove_invariants): Renamed to ...
20154 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20155 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20156 (iv_ca_set_add_invariants): Renamed to ...
20157 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20158 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20159 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20160 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20161 (create_new_ivs): Remove useless dump.
20163 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20165 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20167 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20168 (iv_ca_set_no_cp, create_new_iv): Ditto.
20170 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20172 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20174 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20176 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20177 function tree_check2.
20179 2017-05-02 Martin Liska <mliska@suse.cz>
20181 * doc/gcov.texi: Add missing preposition.
20182 * gcov.c (function_info::function_info): Properly fill up
20183 all member variables.
20185 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20187 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20189 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20191 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20193 2017-05-02 Martin Liska <mliska@suse.cz>
20196 * lto-streamer-in.c (lto_read_tree_1): Remove
20197 LTO_STREAMER_DEBUG.
20198 * lto-streamer.c (struct tree_hash_entry): Likewise.
20199 (struct tree_entry_hasher): Likewise.
20200 (tree_entry_hasher::hash): Likewise.
20201 (tree_entry_hasher::equal): Likewise.
20202 (lto_streamer_init): Likewise.
20203 (lto_orig_address_map): Likewise.
20204 (lto_orig_address_get): Likewise.
20205 (lto_orig_address_remove): Likewise.
20206 * lto-streamer.h: Likewise.
20207 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20208 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20210 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20212 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20213 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20214 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20215 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20216 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20217 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20218 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20219 (mm_maskz_sub_ss): New intrinsics.
20220 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20221 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20222 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20223 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20224 (__builtin_ia32_subss_mask_round): New builtins.
20225 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20226 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20227 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20229 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20230 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20232 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20235 2017-05-02 Martin Jambor <mjambor@suse.cz>
20237 PR tree-optimization/78687
20238 * tree-sra.c (access): New field parent.
20239 (process_subtree_disqualification): New function.
20240 (disqualify_candidate): Call it.
20241 (build_accesses_from_assign): Reset write flag if creating an
20243 (build_access_subtree): Fill in parent field and also prpagate
20244 down grp_write flag.
20245 (create_artificial_child_access): New parameter set_grp_write, set
20246 grp_write to its value.
20247 (propagate_subaccesses_across_link): Also propagate grp_write flag
20249 (propagate_all_subaccesses): Push the closest parent back to work
20250 queue if add_access_to_work_queue returned true.
20252 2017-05-02 Richard Biener <rguenther@suse.de>
20254 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20255 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20256 -fstrict-overflow documentation.
20257 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20258 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20259 flag_strict_overflow.
20260 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20261 * lto-opts.c (lto_write_options): Do not stream it.
20262 * lto-wrapper.c (merge_and_complain): Do not handle it.
20263 * opts.c (default_options_table): Do not set -fstrict-overflow.
20264 (finish_options): Likewise do not clear it when sanitizing.
20265 * simplify-rtx.c (simplify_const_relational_operation): Do not
20266 test flag_strict_overflow.
20268 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20270 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20271 using enabled attribute.
20272 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20273 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20274 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20275 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20276 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20277 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20278 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20279 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20280 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20281 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20283 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20285 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20287 2017-05-02 Richard Biener <rguenther@suse.de>
20289 PR tree-optimization/80591
20291 2017-04-10 Richard Biener <rguenther@suse.de>
20293 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20296 2017-05-02 Richard Biener <rguenther@suse.de>
20298 PR tree-optimization/80549
20299 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20300 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20301 headers if they do not have a preheader.
20303 2017-05-02 Martin Liska <mliska@suse.cz>
20306 * common.opt: Fix typo.
20307 * doc/invoke.texi: Likewise.
20309 2017-05-01 Jan Beulich <jbeulich@suse.com>
20311 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20312 swapping, add (x,x,m,x,n) alternative.
20314 2017-05-01 Nathan Sidwell <nathan@acm.org>
20316 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20317 unnecessary unadjusted_alignment check.
20319 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20322 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20324 * gimplify.c (gimplify_cilk_detach): New function.
20325 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20326 * tree-core.h: Document EXPR_CILK_SPAWN.
20327 * tree.h (EXPR_CILK_SPAWN): Define.
20329 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20331 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20332 to use new fixit_hint representation, using the "replace" logic.
20333 (get_line_span_for_fixit_hint): Likewise.
20334 (layout::print_any_fixits): Likewise.
20335 (selftest::test_one_liner_many_fixits): Rename to...
20336 (selftest::test_one_liner_many_fixits_1): ...this, and update
20337 comment and expected output to reflect that the multiple fix-it
20338 hints are now consolidated into one insertion.
20339 (selftest::test_one_liner_many_fixits_2): New test.
20340 (selftest::test_diagnostic_show_locus_one_liner): Update for
20342 (selftest::test_fixit_consolidation): Update for fix-it API
20344 * diagnostic.c (print_parseable_fixits): Likewise.
20345 * edit-context.c (edited_line::m_line_events): Convert from
20346 auto_vec <line_event *> to auto_vec <line_event>.
20347 (class line_event): Convert from abstract base class to a concrete
20348 class, taking over the role of replace_event.
20349 (class insert_event): Delete.
20350 (class replace_event): Rename to class line_event. Convert to
20352 (edit_context::add_fixits): Reimplement.
20353 (edit_context::apply_insert): Delete.
20354 (edit_context::apply_replace): Rename to...
20355 (edit_context::apply_fixit): ...this. Convert to half-open range.
20356 (edited_file::apply_insert): Delete.
20357 (edited_file::apply_replace): Rename to...
20358 (edited_file::apply_fixit): ...this.
20359 (edited_line::~edited_line): Drop deletion of events.
20360 (edited_line::apply_insert): Delete.
20361 (edited_line::apply_replace): Rename to...
20362 (edited_line::apply_fixit): ...this. Convert to half-open range.
20363 Update for change to type of m_line_events.
20364 * edit-context.h (edit_context::apply_insert): Delete.
20365 (edit_context::apply_replace): Rename to...
20366 (edit_context::apply_fixit): ...this.
20368 2017-05-01 Martin Sebor <msebor@redhat.com>
20370 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20373 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20376 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20377 __get_cpuid_max returns 0.
20378 (__get_cpuid_count): Ditto.
20380 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20382 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20383 replacement expression is another instance of one of its arguments.
20385 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20388 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20389 check for stack push/pop autoinc.
20390 * config/i386/i386.c (ix86_agi_dependent): Return false
20391 if the only reason why modified_in_p returned true is that
20392 addr is SP based and set_insn is a push or pop.
20394 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20396 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20399 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20402 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20403 (account_size_time): Use two predicates - exec_pred and
20405 (evaluate_conditions_for_known_args): Compute both clause and
20407 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20408 (inline_summary_t::duplicate): Update.
20409 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20411 (compute_inline_parameters): Likewise.
20412 (estimate_edge_size_and_time): Update caluclation of time.
20413 (estimate_node_size_and_time): Compute both time and nonspecialized
20415 (estimate_ipcp_clone_size_and_time): Update.
20416 (inline_merge_summary): Update.
20417 (do_estimate_edge_time): Update.
20418 (do_estimate_edge_size): Update.
20419 (do_estimate_edge_hints): Update.
20420 (inline_read_section, inline_write_summary): Stream both new predicates.
20421 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20423 (compute_inlined_call_time): Cleanup.
20424 (big_speedup_p): Update.
20425 (edge_badness): Update.
20426 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20427 (size_time_entry): Replace predicate by exec_predicate and
20428 nonconst_predicate.
20429 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20430 (estimate_edge_time): Return also nonspec_time.
20431 (reset_edge_growth_cache): Update.
20433 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20435 PR rtl-optimization/80491
20436 * ifcvt.c (noce_process_if_block): When looking for x setter
20437 with missing else_bb, don't check only the insn right before
20438 cond_earliest, but look for the last insn that x is modified in
20439 within the same bb.
20441 PR rtl-optimization/80491
20442 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20444 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20446 PR tree-optimization/80487
20447 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20449 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20451 PR tree-optimization/79697
20452 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20453 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20454 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20456 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20457 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20459 2017-04-28 Martin Sebor <msebor@redhat.com>
20461 PR tree-optimization/80523
20462 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20463 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20465 (maybe_warn, format_directive, parse_directive): Use new functions.
20466 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20468 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20470 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20472 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20474 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20475 target_header_dir): Set correctly.
20476 * configure: Regenerated.
20477 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20478 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20479 instead of SYSTEM_HEADER_DIR.
20481 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20483 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20484 (estimate_local_effects): Likewise.
20485 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20486 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20487 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20488 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20489 do_estimate_edge_time, estimate_edge_time): Likewise.
20490 * ipa-inline-analysis.c (estimate_node_size_and_time,
20491 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20492 (estimate_time_after_inlining): Remove.
20494 2017-04-28 Martin Liska <mliska@suse.cz>
20496 * doc/gcov.texi: Enhance documentation of gcov.
20498 2017-04-28 Martin Liska <mliska@suse.cz>
20500 * doc/gcov.texi: Sort options in alphabetic order.
20501 * doc/gcov-dump.texi: Likewise.
20502 * doc/gcov-tool.texi: Likewise.
20503 * gcov.c (print_usage): Likewise.
20504 * gcov-dump.c (print_usage): Likewise.
20505 * gcov-tool.c (print_merge_usage_message): Likewise.
20506 (print_rewrite_usage_message): Likewise.
20507 (print_overlap_usage_message): Likewise.
20509 2017-04-28 Martin Liska <mliska@suse.cz>
20511 PR gcov-profile/53915
20512 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20514 2017-04-28 Martin Liska <mliska@suse.cz>
20516 PR gcov-profile/79891
20517 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20518 is marked by compiler as living on a line.
20519 (get_cycles_count): Remove usage of the union.
20520 (output_intermediate_file): Likewise.
20521 (find_source): Fix GNU coding style.
20522 (accumulate_line_counts): Remove old non-all block mode.
20523 (output_lines): Remove usage of the union.
20524 * profile.c (output_location): Include all BBs, even if
20525 belonging to a same line (and file) as a previous BB.
20527 2017-04-28 Martin Liska <mliska@suse.cz>
20529 * gcov.c (process_args): Handle new argument 'w'.
20530 (read_graph_file): Assign ID to BBs.
20531 (output_branch_count): Display BB # if verbose flag is set.
20532 (output_lines): Likewise for arcs.
20533 (print_usage): Add '--verbose' option help.
20534 * doc/gcov.texi: Document --verbose (-w) option.
20536 2017-04-28 Martin Liska <mliska@suse.cz>
20538 * gcov.c (struct block_location_info): New struct.
20539 (process_file): Fill up the new structure.
20540 (read_graph_file): Replace usage of encoding by the newly added
20542 (add_line_counts): Likewise.
20543 (accumulate_line_counts): Remove usage of the union.
20544 (function_info::function_info): New function.
20545 (function_info::~function_info): Likewise.
20546 (process_file): Call delete instead of release_function.
20547 (release_function): Release the function.
20548 (release_structures): Call delete instead of release_function.
20549 (solve_flow_graph): Replace usage of num_blocks.
20550 (find_exception_blocks): Likewise.
20551 (output_lines): Fix GNU coding style.
20553 2017-04-28 Martin Liska <mliska@suse.cz>
20556 * coverage.c (coverage_remove_note_file): New function.
20557 * coverage.h: Declare the function.
20558 * toplev.c (finalize): Clean if an error has been seen.
20560 2017-04-28 Martin Liska <mliska@suse.cz>
20562 PR gcov-profile/80031
20563 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20564 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20565 * gcov.c (read_graph_file): Read just number of blocks.
20566 * profile.c (branch_prob): Do not stream 0 flags per a basic
20569 2017-04-28 Martin Liska <mliska@suse.cz>
20571 * gcov-dump.c (tag_*): Add new argument to declarations.
20572 (dump_gcov_file): Likewise.
20573 (tag_blocks): Add and use new argument depth.
20574 (tag_arcs): Likewise.
20575 (tag_lines): Likewise.
20576 (tag_counters): Likewise.
20577 (tag_summary): Likewise.
20578 (dump_working_sets): Use depth to do a proper indentation.
20580 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20583 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20584 * symtab.c (symtab_node::debug_symtab): Move definition here.
20586 2017-04-28 Richard Biener <rguenther@suse.de>
20588 * lto-streamer.h (LTO_major_version): Bump to 7.
20590 2017-04-28 Richard Biener <rguenther@suse.de>
20592 * tree-vrp.c (assert_info): New struct.
20593 (add_assert_info): New helper.
20594 (register_edge_assert_for_2): Refactor to add asserts to a vector
20596 (register_edge_assert_for_1): Likewise.
20597 (register_edge_assert_for): Likewise.
20598 (finish_register_edge_assert_for): New helper actually registering
20599 asserts where live on edge.
20600 (find_conditional_asserts): Adjust.
20601 (find_switch_asserts): Likewise.
20602 (evrp_dom_walker::try_find_new_range): Generalize.
20603 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20605 2017-04-27 Marek Polacek <polacek@redhat.com>
20608 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20609 arg10 and arg11 to itype.
20611 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20613 * doc/extend.texi (Object Size Checking): Improve grammar.
20615 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20618 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20619 that the logic for permitting reciprocal estimates matches that
20622 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20625 * tree.c (type_cache_hasher::equal): Only compare
20626 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20627 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20628 non-aggregate element types.
20629 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20630 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20632 2017-04-27 Richard Biener <rguenther@suse.de>
20634 PR middle-end/80533
20635 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20636 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20637 keeping a reference to a trailing array.
20639 2017-04-27 Richard Biener <rguenther@suse.de>
20641 PR middle-end/80539
20642 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20643 being in loop-closed SSA form conservatively.
20644 (chrec_fold_multiply_poly_poly): Likewise.
20646 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20648 PR middle-end/79665
20649 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20650 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20652 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20655 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20656 (aarch64_function_arg_alignment): Return unsigned int again, but still
20657 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20658 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20659 Don't emit -Wpsabi note.
20660 (aarch64_function_arg_boundary): Likewise.
20661 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20664 2017-04-26 Nathan Sidwell <nathan@acm.org>
20666 * tree.h (crc32_unsigned_n): Declare.
20667 (crc32_unsigned, crc32_unsigned): Make inline.
20668 * tree.c (crc32_unsigned_bits): Replace with ...
20669 (crc32_unsigned_n): ... this.
20670 (crc32_unsigned, crc32_byte): Remove.
20671 (crc32_string): Remove unnecessary braces.
20673 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20675 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20676 * ipa-inline-analysis.c (MAX_TIME): Remove.
20677 (account_size_time): Use sreal for time.
20678 (dump_inline_summary): Update.
20679 (estimate_function_body_sizes): Update.
20680 (estimate_edge_size_and_time): Update.
20681 (estimate_calls_size_and_time): Update.
20682 (estimate_node_size_and_time): Update.
20683 (inline_merge_summary): Update.
20684 (inline_update_overall_summary): Update.
20685 (estimate_time_after_inlining): Update.
20686 (inline_read_section): Update.
20687 (inline_write_summary): Update.
20688 * ipa-inline.c (compute_uninlined_call_time): Update.
20689 (compute_inlined_call_time): Update.
20690 (recursive_inlining): Update.
20691 (inline_small_functions): Update.
20692 (dump_overall_stats): Update.
20693 * ipa-inline.h: Include sreal.h.
20694 (size_time_entry): Turn time to sreal.
20695 (inline_summary): Turn self_time nad time to sreal.
20697 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20699 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20701 (sreal::stream_out, sreal::stream_in): New.
20702 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20704 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20706 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20709 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20712 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20714 (dimode_scalar_chain::compute_convert_gain): Ditto.
20715 (dimode_scalar_chain::make_vector_copies): Ditto.
20716 (dimode_scalar_chain::convert_reg): Ditto.
20717 (dimode_scalar_chain::convert_insn): Ditto.
20718 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20719 (VI248_AVX512BW_1): New mode iterator.
20720 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20721 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20724 2017-04-25 Martin Sebor <msebor@redhat.com>
20726 PR tree-optimization/80497
20727 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20728 constants are representable in HOST_WIDE_INT.
20729 (parse_directive): Ditto.
20731 2017-04-25 Martin Sebor <msebor@redhat.com>
20734 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20735 (new_zero_array): Adjust signature.
20736 (dom_info::dom_init): Used unsigned rather that size_t.
20737 (dom_info::dom_info): Same.
20739 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20740 Jakub Jelinek <jakub@redhat.com>
20743 * config/arm/arm.c: Include gimple.h.
20744 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20745 returns negative, increment ncrn only if it returned positive.
20746 (arm_needs_doubleword_align): Return int instead of bool,
20747 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20748 members, but if there is any such non-FIELD_DECL
20749 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20750 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20751 returns negative, increment nregs only if it returned positive.
20752 (arm_setup_incoming_varargs): Likewise.
20753 (arm_function_arg_boundary): Emit -Wpsabi note if
20754 arm_needs_doubleword_align returns negative, return
20755 DOUBLEWORD_ALIGNMENT only if it returned positive.
20757 2017-04-25 Marek Polacek <polacek@redhat.com>
20760 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20761 first argument to type.
20763 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20766 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20767 type checks to test for compatibility instead of equality.
20769 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20770 Jakub Jelinek <jakub@redhat.com>
20773 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20775 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20776 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20777 the alignment computation, but return their maximum in warn_alignment.
20778 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20779 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20781 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20782 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20785 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20787 * config/arc/simdext.md (dmpyh): Fix typo.
20789 2017-04-25 Richard Biener <rguenther@suse.de>
20791 PR tree-optimization/80492
20792 * alias.c (compare_base_decls): Handle registers with asm
20793 specification conservatively.
20794 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20795 compare_base_decls returning dont-know properly.
20797 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20799 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20800 (legitimate_offset_address_p): New function.
20801 (arc_legitimate_address_p): Use above function.
20803 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20805 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20807 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20809 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20810 ACCH registers whenever they are available.
20812 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20814 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20815 double regs fix when not used.
20817 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20819 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20821 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20822 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20824 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20826 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20827 when emitting short ADD instructions.
20829 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20831 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20833 (cmpsi_cc_c_insn): Likewise.
20834 (cbranchsi4_scratch): Compute proper instruction length using
20835 compact_hreg_operand.
20836 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20838 2017-04-25 Richard Biener <rguenther@suse.de>
20840 PR middle-end/80509
20841 * passes.c (pass_manager::pass_manager): Initialize
20842 m_name_to_pass_map.
20844 2017-04-25 Richard Biener <rguenther@suse.de>
20846 PR tree-optimization/79201
20847 * tree-ssa-sink.c (statement_sink_location): Handle calls.
20849 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20852 * config/s390/vector.md: Split MEM->GPR vector moves for
20853 non-s_operand addresses.
20855 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20858 * config/s390/predicates.md (reload_const_wide_int_operand): New
20860 * config/s390/s390.md ("movti"): Remove d/P alternative.
20861 ("movti_bigconst"): New pattern definition.
20863 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20866 * s390-protos.h (s390_expand_cs_hqi): Removed.
20867 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
20868 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
20869 modes as well as CCZ1mode and CCZmode.
20870 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
20871 signature of s390_emit_compare_and_swap.
20872 (s390_expand_cs_hqi): Likewise, make static.
20873 (s390_expand_cs_tdsi): Generate an explicit compare before trying
20874 compare-and-swap, in some cases.
20875 (s390_expand_cs): Wrapper function.
20876 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
20878 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
20879 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
20880 patterns for small and large integers. Forbid symref memory operands.
20881 Move expander to s390.c. Require cc register.
20882 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
20883 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
20884 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
20885 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
20886 symref memory operands. Remove CC mode and call s390_match_ccmode
20888 ("atomic_exchange<mode>"): Allow and implement all integer modes.
20890 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20892 * config/s390/s390.md (define_peephole2): New peephole to help
20893 combining the load-and-test pattern with volatile memory.
20895 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
20897 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
20898 with CCZmode for TARGET_Z196.
20900 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20902 PR rtl-optimization/80501
20903 * combine.c (make_compound_operation_int): Set subreg_code to SET
20904 even for AND with mask of the sign bit of mode.
20906 PR rtl-optimization/80500
20907 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
20908 sum's initial value.
20910 2017-04-25 Julian Brown <julian@codesourcery.com>
20911 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20913 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
20915 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
20917 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
20919 2017-04-25 Julian Brown <julian@codesourcery.com>
20920 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20922 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
20923 (thunderx2t99_sha): New Reservation.
20925 2017-04-25 Julian Brown <julian@codesourcery.com>
20926 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20928 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
20929 type for 1-element load.
20931 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
20933 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
20935 2017-04-24 Martin Jambor <mjambor@suse.cz>
20937 PR tree-optimization/80293
20938 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
20939 char arrays not totally scalarizable if it is false.
20940 (analyze_all_variable_accesses): Pass correct value in the new
20941 parameter. Add a statistics counter.
20943 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
20945 PR middle-end/79931
20946 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
20948 2017-04-24 Richard Biener <rguenther@suse.de>
20950 PR tree-optimization/80494
20951 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
20952 out for complex types.
20954 2017-04-24 Richard Biener <rguenther@suse.de>
20956 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
20957 * tree-ssa-sccvn.c (print_scc): Print SCC size.
20958 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
20959 (DFS): Adjust and never fail.
20960 (sccvn_dom_walker::fail): Remove.
20961 (sccvn_dom_walker::before_dom_children): Adjust.
20962 (run_scc_vn): Likewise and never fail.
20963 * tree-ssa-pre.c (pass_pre::execute): Adjust.
20964 (pass_fre::execute): Likewise.
20966 2017-04-24 Richard Biener <rguenther@suse.de>
20968 PR tree-optimization/79725
20969 * tree-ssa-sink.c (statement_sink_location): Return whether
20970 failure reason was zero uses. Move that check later.
20971 (sink_code_in_bb): Deal with zero uses by removing the stmt
20974 2017-04-24 Richard Biener <rguenther@suse.de>
20977 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
20978 pointer-based references.
20980 2017-04-24 Richard Biener <rguenther@suse.de>
20983 * pass_manager.h (pass_manager::operator new): Remove.
20984 (pass_manager::operator delete): Likewise.
20985 * passes.c (pass_manager::operator new): Remove.
20986 (pass_manager::operator delete): Likewise.
20987 (pass_manager::pass_manager): Zero individual pass members.
20989 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
20992 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
20993 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
20994 Check "XEXP (src, 1)" operand here.
20995 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
20996 Check "XEXP (src, 1)" operand here.
20997 (dimode_scalar_chain::make_vector_copies): Detect count register
20998 of a shift instruction. Zero extend count register from QImode
20999 to DImode to satisfy vector shift pattern count operand predicate.
21000 Substitute vector shift count operand with a DImode copy.
21001 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21004 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21006 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21007 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21008 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21009 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21010 (UNSPEC_NOREX_MEM): Remove definition.
21012 2017-04-21 Richard Biener <rguenther@suse.de>
21014 PR tree-optimization/79547
21015 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21016 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21017 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21018 without any constraints.
21020 2017-04-21 Richard Biener <rguenther@suse.de>
21022 PR tree-optimization/78847
21023 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21025 2017-04-21 Richard Biener <rguenther@suse.de>
21027 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21028 (build_distinct_type_copy): Likewise.
21029 (build_variant_type_copy): Likewise.
21030 * tree.c (build_qualified_type): Pass down mem-stat info.
21031 (build_distinct_type_copy): Likewise.
21032 (build_variant_type_copy): Likewise.
21034 2017-04-21 Richard Biener <rguenther@suse.de>
21036 PR tree-optimization/80237
21037 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21039 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21040 for a simplified result.
21042 2016-04-21 Richard Biener <rguenther@suse.de>
21044 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21045 sth as strict as a simple_iv but a chrec without symbols and an
21046 operand defined in the loop we are peeling (and not some subloop).
21047 (propagate_constants_for_unrolling): Propagate all constants.
21049 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21052 * config/i386/i386.c (print_reg): Remove assert for disalowed
21053 regno values, call output_operand_lossage instead.
21055 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21058 * config/i386/constraints.md (Yc): New register constraint.
21059 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21060 Use Yc constraint for alternative 2 of operand 0. Remove
21061 preferred_for_speed attribute.
21063 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21065 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21066 lastprivate clauses in SIMT case.
21068 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21070 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21072 2017-04-20 Richard Biener <rguenther@suse.de>
21074 PR tree-optimization/57796
21075 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21076 as N scalar stores.
21077 (vect_model_load_cost): Cost gathers as N scalar loads.
21079 2017-04-20 Richard Biener <rguenther@suse.de>
21081 * ggc-page.c (ggc_allocated_p): Rename to ...
21082 (safe_lookup_page_table_entry): ... this and return the lookup
21084 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21086 2017-04-20 Richard Biener <rguenther@suse.de>
21088 PR tree-optimization/80453
21089 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21090 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21091 from the conditions.
21092 (vn_phi_eq): Pass them down.
21093 (vn_phi_lookup): Record them.
21094 (vn_phi_insert): Likewise.
21096 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21098 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21099 uninitialized variable warning to avoid buffer overrun.
21101 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21104 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21105 is suppressed for '{ 0 }' in C.
21107 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21109 * BASE-VER: Set to 8.0.0.
21111 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21113 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21114 priority .init_array and .fini_array section with SECTION_NOTYPE
21117 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21119 PR middle-end/80423
21120 * tree.h (build_array_type): Add typeless_storage default argument.
21121 * tree.c (type_cache_hasher::equal): Also compare
21122 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21123 (build_array_type): Add typeless_storage argument, set
21124 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21126 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21127 (build_array_type): Likewise. Add typeless_storage argument.
21129 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21130 Jakub Jelinek <jakub@redhat.com>
21132 PR tree-optimization/80426
21133 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21134 operation on symbolic operands, also compute the overflow for the
21135 invariant part when the operation degenerates into a negation.
21137 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21140 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21141 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21144 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21146 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21149 * config/avr/avr.c (tree.h): Include it.
21150 (cgraph.h): Include it.
21151 (avr_encode_section_info): Don't warn for uninitialized progmem
21152 variable if it's just an alias.
21154 2017-04-19 Richard Biener <rguenther@suse.de>
21157 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21158 when needed by AutoPGO.
21160 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21163 * doc/lto.texi: Remove an extra 'that'.
21165 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21167 PR rtl-optimization/80429
21168 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21169 are only used in debug insns.
21171 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21172 Vladimir Makarov <vmakarov@redhat.com>
21174 * config/sparc/predicates.md (input_operand): Add comment. Return
21175 true for any memory operand when LRA is in progress.
21176 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21178 2017-04-18 Jeff Law <law@redhat.com>
21181 * mips.md ({return,simple_return}_internal): Do not overwrite
21184 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21186 PR tree-optimization/80443
21187 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21188 instead of adding 1, subtract -1 and similarly instead of subtracting
21191 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21193 PR rtl-optimization/80357
21194 * haifa-sched.c (tmp_bitmap): New variable.
21195 (model_recompute): Handle duplicate use records.
21196 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21197 (free_global_sched_pressure_data): Free it.
21199 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21202 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21203 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21204 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21205 instead of SYSTEM_HEADER_DIR.
21207 2017-04-18 Jeff Law <law@redhat.com>
21209 PR middle-end/80422
21210 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21211 predecessors after walking up the insn chain.
21213 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21216 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21217 sizetype type into debug info.
21219 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21222 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21223 unneeded test for TARGET_UPPER_REGS_SF.
21224 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21226 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21229 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21230 instead of gsi_after_labels.
21232 2017-04-18 Jeff Law <law@redhat.com>
21234 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21238 2017-04-13 Jeff Law <law@redhat.com>
21239 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21240 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21242 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21245 * config/avr/avr.c (intl.h): Include it.
21246 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21248 2017-04-18 Martin Liska <mliska@suse.cz>
21250 PR gcov-profile/78783
21251 * gcov-tool.c (gcov_output_files): Validate that destination
21252 file is either removed by the tool or by a user.
21254 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21255 Guy Benyei <guybe@mellanox.com>
21257 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21258 block, and do not negate it, the stored id is already negative.
21260 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21262 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21264 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21267 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21268 masks of options that should be turned off if the VSX vector
21269 options are turned off.
21270 (OTHER_P8_VECTOR_MASKS): Likewise.
21271 (OTHER_VSX_VECTOR_MASKS): Likewise.
21272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21273 rs6000_disable_incompatible_switches to validate no type switches
21275 (rs6000_incompatible_switch): New function to disallow turning on
21276 other vector options if -mno-vsx, -mno-power8-vector, or
21277 -mno-power9-vector are specified.
21279 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21281 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21283 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21285 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21286 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21287 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21288 (ARG_POINTER_CFA_OFFSET): Likewise.
21290 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21292 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21293 conditions to take advantage of various optimizations.
21295 2017-04-13 Jeff Law <law@redhat.com>
21297 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21298 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21299 (zero_extendsidi2_dext): Likewise.
21301 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21304 * fold-const.c (fold_ternary_loc): Revert
21305 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21308 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21310 PR rtl-optimization/80343
21311 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21312 new scratch pseudo.
21314 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21317 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21318 to ubsan_encode_value.
21320 2017-04-13 Jeff Law <law@redhat.com>
21322 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21323 appearing in DEBUG_INSNs.
21325 2017-04-13 Martin Liska <mliska@suse.cz>
21327 PR gcov-profile/80413
21328 * gcov-io.c (gcov_write_string): Copy to buffer just when
21329 allocated size is greater than zero.
21331 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21334 * dwarf2out.c (decls_for_scope): Ignore declarations of
21335 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21337 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21340 * ipa-visibility.c (non_local_p): Fix typos.
21341 (localize_node): When localizing symbol in same comdat group,
21342 dissolve the group only when we know external symbols are going
21344 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21346 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21348 PR tree-optimization/79390
21349 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21350 order does not result in usable sequence, retry with reversed operand
21356 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21357 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21358 op0 instead of fold_convert_loc (loc, type, arg0).
21360 2017-04-12 Jeff Law <law@redhat.com>
21362 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21363 has a delay slot in the generated code.
21365 * config/cris/cris.md (cris_preferred_reload_class): Return
21366 GENNONACR_REGS rather than GENERAL_REGS.
21368 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21371 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21372 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21373 signedness of the result type.
21375 2017-04-12 Richard Biener <rguenther@suse.de>
21376 Jeff Law <law@redhat.com>
21378 PR tree-optimization/80359
21379 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21380 trim stores to TARGET_MEM_REFs.
21382 2017-04-12 Richard Biener <rguenther@suse.de>
21384 PR tree-optimization/79390
21385 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21386 threading case even more.
21388 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21391 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21392 for quad_address_p for TImode, instead of just not indexed_address.
21394 2017-04-12 Richard Biener <rguenther@suse.de>
21395 Bernd Edlinger <bernd.edlinger@hotmail.de>
21397 PR middle-end/79671
21398 * alias.c (component_uses_parent_alias_set_from): Handle
21399 TYPE_TYPELESS_STORAGE.
21400 (get_alias_set): Likewise.
21401 * tree-core.h (tree_type_common): Add typeless_storage flag.
21402 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21403 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21404 for types containing members with TYPE_TYPELESS_STORAGE.
21405 (place_field): Likewise.
21406 (layout_type): Likewise for ARRAY_TYPE.
21407 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21408 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21409 TYPE_TYPELESS_STORAGE.
21410 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21412 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21415 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21416 first argument to type.
21418 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21422 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21423 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21424 (rs6000_expand_binop_builtin): Likewise.
21425 (rs6000_expand_ternop_builtin): Likewise; also add missing
21426 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21427 vshasigma built-ins.
21428 * doc/extend.texi: Document that vec_xxpermdi's third argument
21429 must be a constant.
21431 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21433 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21434 Use shift_const cost parameter when calculating gain of STV shifts.
21436 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21438 PR rtl-optimization/70478
21439 * lra-constraints.c (process_alt_operands): Check memory for
21440 disfavoring memory insn operand.
21442 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21444 PR middle-end/80100
21445 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21446 left shift in unsigned HOST_WIDE_INT type.
21448 PR rtl-optimization/80385
21449 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21450 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21453 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21454 if they have any depend clauses.
21456 2017-04-11 Martin Liska <mliska@suse.cz>
21459 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21460 * ipa-split.c (split_function): Create a local comdat symbol
21461 if caller is in a comdat group.
21463 2017-04-11 Martin Liska <mliska@suse.cz>
21466 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21469 2017-04-11 Martin Sebor <msebor@redhat.com>
21471 PR middle-end/80364
21472 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21473 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21475 (directive::set_width, directive::set_precision, format_character):
21477 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21480 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21483 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21484 conflict, set target->arch_name instead of target->cpu_name.
21486 2017-04-11 Richard Biener <rguenther@suse.de>
21488 PR tree-optimization/80374
21489 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21490 build_zero_cst, remove fold_convertible_p check again.
21492 2017-04-11 Martin Liska <mliska@suse.cz>
21495 * ubsan.c (instrument_object_size): Do not instrument register
21498 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21501 * config/i386/i386-builtin-types.def
21502 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21503 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21504 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21505 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21506 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21507 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21508 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21509 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21510 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21511 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21512 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21513 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21514 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21515 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21516 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21517 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21518 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21519 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21520 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21521 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21522 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21523 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21524 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21525 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21526 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21527 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21528 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21529 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21530 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21531 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21532 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21533 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21534 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21535 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21536 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21537 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21538 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21539 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21540 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21541 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21542 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21543 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21544 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21545 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21546 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21547 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21549 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21550 flag to second_arg_count, handle 4 argument function type _COUNT
21551 aliases, handle second_arg_count on second argument rather than last.
21553 2017-04-10 Jeff Law <law@redhat.com>
21555 PR tree-optimization/80374
21556 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21557 record anything if we can not convert integer_zero_node to the
21560 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21563 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21564 Enhance special handling given to the TARGET_P9_MINMAX option in
21565 relation to certain other options.
21567 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21569 PR tree-optimization/80153
21570 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21571 remove POINTER_PLUS_EXPR's base part directly, rather than through
21574 2017-04-10 Richard Biener <rguenther@suse.de>
21575 Bin Cheng <bin.cheng@arm.com>
21577 PR tree-optimization/80153
21578 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21579 the first element of pointer type aff_tree. Build result expr in
21581 (add_elt_to_tree): Convert to type unconditionally. Remove other
21582 fold_convert calls.
21583 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21584 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21586 2017-04-10 Richard Biener <rguenther@suse.de>
21588 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21591 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21593 PR rtl-optimization/70478
21594 * lra-constraints.c (curr_small_class_check): New.
21595 (update_and_check_small_class_inputs): New.
21596 (process_alt_operands): Update curr_small_class_check. Disfavor
21597 alternative insn memory operands. Check available regs for small
21600 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21603 * config/mips/mips.opt (-mvirt): Update description.
21604 * doc/invoke.texi (-mvirt): Likewise.
21606 2017-04-10 Richard Biener <rguenther@suse.de>
21608 PR middle-end/80362
21609 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21610 looking for NEGATE_EXPR in -A / -B to A / B folding.
21612 2017-04-10 Martin Liska <mliska@suse.cz>
21614 PR gcov-profile/80224
21615 * gcov.c (print_usage): Fix usage string.
21616 (get_gcov_intermediate_filename): Remove.
21617 (output_gcov_file): Use both for normal and intermediate format.
21618 (generate_results): Do not initialize special file for
21619 intermediate format.
21621 2017-04-10 Richard Biener <rguenther@suse.de>
21623 PR tree-optimization/80304
21624 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21627 2017-04-10 Nathan Sidwell <nathan@acm.org>
21630 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21631 (rs6000_init_builtins): Use it.
21633 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21635 * config/arm/arm.md (<mrc>): Add mode to SET source.
21636 (<mrrc>): Likewise.
21638 2017-04-10 Richard Biener <rguenther@suse.de>
21640 PR middle-end/80344
21641 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21643 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21646 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21647 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21648 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21649 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21650 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21651 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21652 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21653 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21654 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21655 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21656 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21657 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21658 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21659 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21660 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21661 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21662 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21663 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21664 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21665 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21666 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21667 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21668 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21670 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21672 PR rtl-optimization/70478
21673 * lra-constraints.c: Reverse the last patch.
21675 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21677 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21678 Add comment for WCHAR_T.
21680 2017-04-08 Martin Liska <mliska@suse.cz>
21683 2017-04-07 Martin Liska <mliska@suse.cz>
21686 * ipa-split.c (split_function): Add function part to a same comdat
21689 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21692 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21694 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21696 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21697 (vec_pairing): ...to this.
21698 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21699 (rs6000_sched_init): Adjust for name change.
21700 (struct rs6000_sched_context): Likewise.
21701 (rs6000_init_sched_context): Likewise.
21702 (rs6000_set_sched_context): Likewise.
21704 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21710 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21712 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21713 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21714 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21716 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21718 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21720 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21722 PR rtl-optimization/70703
21723 * ira-color.c (update_conflict_hard_regno_costs): Use
21724 int64_t instead of HOST_WIDE_INT.
21726 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21728 PR rtl-optimization/70478
21729 * lra-constraints.c (process_alt_operands): Disfavor alternative
21730 insn memory operands.
21732 2017-04-07 Jeff Law <law@redhat.com>
21734 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21735 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21737 2017-04-07 Martin Liska <mliska@suse.cz>
21740 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21741 Show error message instead of an ICE.
21743 2017-04-07 Martin Liska <mliska@suse.cz>
21746 * ipa-split.c (split_function): Add function part to a same comdat
21749 2017-04-07 Richard Biener <rguenther@suse.de>
21751 PR middle-end/80341
21752 * tree.c (get_unwidened): Also handle ! for_type case for
21754 * convert.c (do_narrow): Split out from ...
21755 (convert_to_integer_1): ... here. Do not pass final truncation
21756 type to get_unwidened for TRUNC_DIV_EXPR.
21758 2017-04-07 Richard Biener <rguenther@suse.de>
21760 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21762 (aff_combination_const): Adjust.
21763 (aff_combination_scale): Likewise.
21764 (aff_combination_add_elt): Likewise.
21765 (aff_combination_add_cst): Likewise.
21766 (aff_combination_convert): Likewise.
21767 (add_elt_to_tree): Likewise. Remove unused argument.
21768 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21770 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21772 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21774 * config/arm/arm.c (arm_default_short_enums): Use
21775 ARM_DEFAULT_SHORT_ENUMS.
21776 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21778 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21781 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21782 members with redundant out-of-class redeclaration.
21784 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21787 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21788 * config/i386/i386.md (*zero_extendsidi2):
21789 Add (?*x,*x) and (?*v,*v) alternatives.
21791 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21794 * config/i386/i386.c (ix86_expand_builtin)
21795 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21796 mode from insn data. Convert operands to insn operand mode.
21797 Copy operands that don't satisfy insn predicate to a register.
21799 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21801 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21804 2017-04-06 Richard Biener <rguenther@suse.de>
21806 PR tree-optimization/80334
21807 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21808 preserve alignment of accesses.
21810 2017-04-06 Richard Biener <rguenther@suse.de>
21812 PR tree-optimization/80262
21813 * tree-sra.c (build_ref_for_offset): Preserve address-space
21815 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21816 Drop useless address-space information on MEM_REF offsets.
21818 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21820 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21822 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21824 PR rtl-optimization/70703
21825 * ira-color.c (update_conflict_hard_regno_costs): Use
21826 HOST_WIDE_INT instead of long.
21828 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21831 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21832 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21834 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21835 for x86_64 target. Handle -m3dnowa option.
21837 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21839 PR rtl-optimization/70703
21840 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
21841 (update_conflict_hard_regno_costs): Use long instead of unsigned
21842 arithmetic for cost calculation.
21844 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21845 Bernd Edlinger <bernd.edlinger@hotmail.de>
21848 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
21851 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
21854 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
21855 ptr_mode with Pmode throughout.
21856 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
21857 into probe_stack_range and use DImode.
21859 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
21862 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
21863 call_eh_return is true.
21865 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21867 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
21868 Initialize last_match_fntype_index.
21870 2017-04-05 Jakub Jelinek <jakub@redhat.com>
21873 * tree-nvr.c: Include internal-fn.h.
21874 (pass_return_slot::execute): Ignore internal calls without
21877 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21878 Richard Biener <rguenther@suse.de>
21881 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
21882 captures used multiple times, except for the last use.
21883 * generic-match-head.c: Include gimplify.h.
21885 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21887 PR tree-optimization/79390
21888 * target.h (struct noce_if_info): Declare.
21889 * targhooks.h (default_noce_conversion_profitable_p): Declare.
21890 * target.def (noce_conversion_profitable_p): New target hook.
21891 * ifcvt.h (struct noce_if_info): New type, moved from ...
21892 * ifcvt.c (struct noce_if_info): ... here.
21893 (noce_conversion_profitable_p): Renamed to ...
21894 (default_noce_conversion_profitable_p): ... this. No longer
21896 (noce_try_store_flag_constants, noce_try_addcc,
21897 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
21898 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
21899 instead of noce_conversion_profitable_p.
21900 * config/i386/i386.c: Include ifcvt.h.
21901 (ix86_option_override_internal): Don't override
21902 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
21903 (ix86_noce_conversion_profitable_p): New function.
21904 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
21905 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
21906 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
21907 * doc/tm.texi: Regenerated.
21909 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21911 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
21914 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21917 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
21918 instructions for small multiply cores.
21920 2017-04-04 Jeff Law <law@redhat.com>
21922 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
21924 (mips_expand_vec_perm_const): Initialize elements in orig_perm
21925 that are not set by the loop over the elements.
21927 2017-04-04 Jakub Jelinek <jakub@redhat.com>
21930 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
21931 int mode, convert_modes it to mode as unsigned, otherwise use
21932 lowpart_subreg to mode rather than SImode.
21933 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
21934 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
21935 Use DImode instead of SImode for the shift count operand.
21936 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
21939 2017-04-04 Richard Biener <rguenther@suse.de>
21941 PR middle-end/80281
21942 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
21943 arithmetic done for the negate or the plus. Simplify.
21944 (A - (-B) -> A + B): Likewise.
21945 * fold-const.c (split_tree): Make sure to not negate pointers.
21947 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
21949 PR rtl-optimization/60818
21950 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
21951 a compare of comparisons with the thing compared if this results
21952 in a different machine mode.
21954 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
21956 * alias.c (base_alias_check): Fix typo in comment.
21957 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
21958 * cgraphunit.c (symbol_table::compile): Likewise.
21959 * collect2.c (maybe_run_lto_and_relink): Likewise.
21960 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
21961 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
21962 * config/avr/avr.c (avr_map_op_t): Likewise.
21963 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
21964 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
21965 * config/epiphany/epiphany.md (movcc): Likewise.
21966 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
21967 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
21969 * config/mips/mips.c (mips_save_restore_reg): Likewise.
21970 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
21971 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
21972 * config/sh/sh.c (sh_rtx_costs): Likewise.
21973 * fold-const.c (fold_truth_andor): Likewise.
21974 * genautomata.c (collapse_flag): Likewise.
21975 * gengtype.h (struct type::u::s): Likewise.
21976 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
21977 * input.c (FORMAT_AMOUNT): Likewise.
21978 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
21979 (known_aggs_to_agg_replacement_list): Likewise.
21980 * ipa-inline-analysis.c: Likewise.
21981 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
21982 * ipa-polymorphic-call.c
21983 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
21984 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
21985 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
21987 * modulo-sched.c (apply_reg_moves): Likewise.
21988 * omp-expand.c (build_omp_regions_1): Likewise.
21989 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
21990 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
21991 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
21992 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
21993 * value-prof.c: Likewise.
21994 * var-tracking.c (val_reset): Likewise.
21996 2017-04-03 Richard Biener <rguenther@suse.de>
21998 PR tree-optimization/80275
21999 * fold-const.c (split_address_to_core_and_offset): Handle
22002 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22004 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22005 descriptors is at least equal to that of functions.
22007 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22009 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22011 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22014 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22015 (mov<IMOD4:mode>): New expander.
22016 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22018 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22020 PR rtl-optimization/79405
22021 * fwprop.c (propagations_left): New variable.
22022 (forward_propagate_into): Decrement it.
22023 (fwprop_init): Initialize it.
22024 (fw_prop): If the variable has reached zero, stop propagating.
22025 (fwprop_addr): Ditto.
22027 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22030 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22031 a FUNCTION_DECL, pass it as decl instead of origin to
22034 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22036 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22039 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22042 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22043 TARGET_VSX_SMALL_INTEGER.
22045 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22047 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22048 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22050 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22052 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22053 extraction from odd-numbered MSA register.
22055 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22057 PR middle-end/80173
22058 * expmed.c (store_bit_field_1): Don't attempt to create
22059 a word subreg out of hard registers wider than word if they
22060 have HARD_REGNO_NREGS of 1 for their mode.
22062 PR middle-end/80163
22063 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22064 conversions to integer types wider than word and pointer.
22067 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22068 (rtx_equal_for_cselib_p): Pass 0 to it.
22069 * cselib.c (cselib_hasher::equal): Likewise.
22070 (rtx_equal_for_cselib_1): Add depth argument. If depth
22071 is 128, don't look up VALUE locs and punt. Increment
22072 depth in recursive calls when walking VALUE locs.
22074 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22076 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22077 (make_gcov_file_name): Use the canonical path name for generating
22079 (read_line): Fix handling of files with ascii null bytes.
22081 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22083 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22084 to initialise a vector register instead
22085 of using a const_int.
22087 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22089 PR translation/80189
22090 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22091 diagnostic messages.
22093 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22096 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22097 (dfp_diex_<mode>): Update mode of operand 1.
22098 * doc/extend.texi (dxex, dxexq): Document change to return type.
22099 (diex, diexq): Document change to argument type.
22101 2017-03-30 Martin Jambor <mjambor@suse.cz>
22104 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22105 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22106 it reflects the signature changes performed at the callee side.
22107 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22108 to cgraph_build_function_type_skip_args.
22109 (build_function_decl_skip_args): Adjust call to the above function.
22111 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22114 * config/i386/sse.md
22115 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22116 register as dest whenever it is a MEM not rtx_equal_p to the
22117 corresponding dup operand, and when forcing into reg move the
22118 reg into the memory afterwards.
22119 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22120 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22121 for the force_reg mode.
22122 (avx512vl_vextractf128<mode>): Use register as dest either
22123 always when a MEM, or when it is a MEM not rtx_equal_p to the
22124 corresponding dup operand, or even not when it is a CONST_VECTOR
22125 depending on the mode and lo vs. hi.
22126 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22128 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22129 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22130 Likewise. Require that operands[2] is even.
22131 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22132 Remove extraneous parens. Require that operands[2] is a multiple
22134 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22135 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22136 disallow memory then.
22138 2017-03-30 Richard Biener <rguenther@suse.de>
22140 PR tree-optimization/77498
22141 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22142 to non-constants over backedges.
22144 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22146 PR rtl-optimization/80233
22147 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22148 as last_combined_insn. Do not test for BARRIER_P separately.
22150 2017-03-29 Andreas Schwab <schwab@suse.de>
22153 * calls.c (prepare_call_address): Convert funexp to Pmode before
22154 copying to temp reg.
22156 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22158 PR tree-optimization/80158
22159 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22160 Handle possible future case of more than one alternate
22162 (replace_rhs_if_not_dup): Likewise.
22163 (replace_one_candidate): Likewise.
22165 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22167 PR rtl-optimization/80193
22168 * ira.c (ira): Do not check allocation for LRA.
22170 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22172 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22173 (nvptx_output_simt_exit): Declare.
22174 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22175 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22176 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22177 (nvptx_declare_function_name): ...here. Emit declaration of local
22178 memory space buffer for omp_simt_enter insn.
22179 (nvptx_output_unisimt_switch): New.
22180 (nvptx_output_softstack_switch): New.
22181 (nvptx_output_simt_enter): New.
22182 (nvptx_output_simt_exit): New.
22183 * config/nvptx/nvptx.h (struct machine_function): New fields
22184 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22185 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22186 (UNSPECV_SIMT_EXIT): Ditto.
22187 (omp_simt_enter_insn): New insn.
22188 (omp_simt_enter): New expansion.
22189 (omp_simt_exit): New insn.
22190 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22192 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22193 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22194 (expand_GOMP_SIMT_EXIT): New.
22195 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22196 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22197 (GOMP_SIMT_EXIT): Ditto.
22198 * target-insns.def (omp_simt_enter): New insn.
22199 (omp_simt_exit): Ditto.
22200 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22202 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22203 (lower_rec_input_clauses): Likewise.
22204 (lower_lastprivate_clauses): Handle SIMT privatization.
22206 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22207 (ompdevlow_adjust_simt_enter): New.
22208 (find_simtpriv_var_op): New.
22209 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22210 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22212 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22213 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22214 (copy_decl_for_dup_finish): Ditto.
22216 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22218 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22221 * config/i386/i386.c (ix86_option_override_internal): Always
22222 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22224 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22226 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22228 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22230 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22231 mark new edge's irreducible flag accordign to it.
22232 (vect_do_peeling): Check loop preheader edge's irreducible flag
22233 and pass it to function slpeel_add_loop_guard.
22235 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22237 PR tree-optimization/80218
22238 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22239 Update block frequencies and counts.
22241 2017-03-28 Richard Biener <rguenther@suse.de>
22243 PR tree-optimization/78644
22244 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22245 of a simplification result we may not use it at all.
22247 2017-03-28 Richard Biener <rguenther@suse.de>
22250 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22251 without arguments, generate default definition of a SSA name.
22253 2017-03-28 Richard Biener <rguenther@suse.de>
22255 PR middle-end/80222
22256 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22257 TYPE_REF_CAN_ALIAS_ALL references.
22258 * fold-const.c (fold_indirect_ref_1): Likewise.
22260 2017-03-28 Martin Liska <mliska@suse.cz>
22263 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22264 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22266 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22267 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22269 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22270 (EXTRA_SPECS): Define.
22271 (SUBTARGET_EXTRA_SPECS): Likewise.
22272 (SUBTARGET_CPP_SPEC): Likewise.
22273 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22274 SUBTARGET_EXTRA_SPECS.
22275 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22277 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22279 * config/arc/simdext.md (vst64_insn): Update pattern.
22280 (vld32wh_insn): Likewise.
22281 (vld32wl_insn): Likewise.
22282 (vld64_insn): Likewise.
22283 (vld32_insn): Likewise.
22285 2017-03-28 Marek Polacek <polacek@redhat.com>
22288 * fold-const.c (fold_comparison): Use protected_set_expr_location
22289 instead of SET_EXPR_LOCATION.
22291 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22293 * tree.c (add_expr): Avoid name lookup warning.
22295 2017-03-27 Jeff Law <law@redhat.com>
22297 PR tree-optimization/80216
22298 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22299 function name. Limit recursion depth.
22300 (record_temporary_equivalences): Corresponding changes.
22302 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22304 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22307 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22310 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22312 * cfgcleanup.c (reg_note_cfa_p): New array.
22313 (insns_have_identical_cfa_notes): New function.
22314 (old_insns_match_p): Don't cross-jump in between /f
22315 and non-/f instructions. If both i1 and i2 are frame related,
22316 verify all CFA notes, their order and content.
22318 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22321 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22322 HImode and SImode with zero extend to DImode to one insn.
22323 (bswap<mode>2_extenddi): Likewise.
22324 (bswapsi2_extenddi): Likewise.
22325 (bswaphi2_extendsi): Likewise.
22326 (bswaphi2): Combine bswap HImode and SImode into one insn.
22327 Separate memory insns from swapping register.
22328 (bswapsi2): Likewise.
22329 (bswap<mode>2): Likewise.
22330 (bswaphi2_internal): Delete, no longer used.
22331 (bswapsi2_internal): Likewise.
22332 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22333 store, and gpr<-gpr swap insns.
22334 (bswap<mode>2_store): Likewise.
22335 (bswaphi2_reg): Register only splitter, combine with the splitter.
22336 (bswaphi2 splitter): Likewise.
22337 (bswapsi2_reg): Likewise.
22338 (bswapsi2 splitter): Likewise.
22339 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22340 the insns into load, store, and register/register insns.
22341 (bswapdi2_ldbrx): Likewise.
22342 (bswapdi2_load): Likewise.
22343 (bswapdi2_store): Likewise.
22344 (bswapdi2_reg): Likewise.
22346 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22348 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22349 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22351 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22354 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22357 special handling for target option conflicts between dform
22358 options (-mpower9-dform, -mpower9-dform-vector,
22359 -mpower9-dform-scalar) and -mno-direct-move.
22361 2017-03-27 Richard Biener <rguenther@suse.de>
22363 PR tree-optimization/80181
22364 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22366 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22368 * config/arc/predicates.md (move_double_src_operand): Replace the
22369 call to move_double_src_operand with a call to address_operand.
22371 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22373 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22374 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22375 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22377 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22379 * config/arc/predicates.md (long_immediate_loadstore_operand):
22380 Consider scaled addresses cases.
22382 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22384 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22385 restored when in interrupt.
22386 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22387 doesn't have delay slot.
22389 2017-03-27 Richard Biener <rguenther@suse.de>
22392 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22393 inlined thunk clones.
22395 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22398 * asan.c (instrument_derefs): Copy over last operand from
22399 original COMPONENT_REF to the new COMPONENT_REF with
22400 DECL_BIT_FIELD_REPRESENTATIVE.
22401 * ubsan.c (instrument_object_size): Likewise.
22403 2017-03-27 Richard Biener <rguenther@suse.de>
22405 PR tree-optimization/80170
22406 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22407 sure DR/SCEV didnt fold in constants we do not see when looking
22408 at the reference base alignment.
22410 2017-03-27 Richard Biener <rguenther@suse.de>
22412 PR middle-end/80171
22413 * gimple-fold.c (fold_ctor_reference): Properly guard against
22414 NULL return value from canonicalize_constructor_val.
22416 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22419 * config/i386/i386.c (ix86_expand_builtin)
22420 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22421 flags reg setting and flags reg using instructions.
22422 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22423 clobbering instructions to zero extend op2.
22425 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22427 * doc/install.texi (Configuration) <--with-aix-soname>:
22428 Update link to AIX ld.
22430 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22432 PR rtl-optimization/80160
22433 PR rtl-optimization/80159
22434 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22435 reg_alternate_class into account.
22437 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22440 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22441 to consider in curr_insn_transform.
22443 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22445 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22446 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22447 and emit_mode_inner.
22449 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22451 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22452 argument to the overloaded builtin variants. Use the new flag to
22453 deprecate certain builtin variants.
22454 * config/s390/s390-builtin-types.def: Add new builtin types.
22455 * config/s390/s390-builtins.h: Support new flags field for
22456 overloaded builtins.
22457 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22458 (s390_macro_to_expand): Enable vector float data type.
22459 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22460 builtins by incrementing the __VEC__ version number.
22461 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22463 (s390_resolve_overloaded_builtin): Emit error messages depending
22464 on the builtin flags.
22465 * config/s390/s390.c (s390_expand_builtin): Support additional
22466 flags argument. Change error message to match the messages
22467 emitted in s390-c.c.
22468 * config/s390/s390.md: New UNSPEC_* constants.
22469 (op_type): Add new instruction types.
22470 * config/s390/vecintrin.h: Add new builtins and test data class
22472 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22473 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22474 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22475 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22476 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22477 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22479 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22480 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22481 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22482 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22484 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22485 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22486 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22487 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22489 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22490 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22491 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22492 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22493 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22494 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22495 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22497 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22498 ("vec_scatter_element<V_HW_4:mode>_DI")
22499 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22500 ("vec_fpint<mode>", "vflls")
22501 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22502 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22503 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22504 ("*vec_cmphe<mode>_cc"): ... these.
22506 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22507 mode constant instead of magic value.
22509 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22511 * config/s390/s390.c (s390_expand_vec_compare): Support other
22512 vector floating point modes than just V2DF.
22513 (s390_expand_vcond): Likewise.
22514 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22515 (s390_cannot_change_mode_class): Prevent mode changes between TF
22516 and V1TF in vector registers.
22517 * config/s390/s390.md (DF, SF): New mode attributes.
22518 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22519 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22520 SFmode support for VRs.
22521 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22523 (VFT, VF_HW): New mode iterators.
22524 (vw, sdx): New mode attributes.
22525 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22526 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22527 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22528 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22529 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22530 also the new vector floating point modes. Renaming to ...
22532 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22533 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22534 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22535 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22536 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22537 ("vec_unordered<mode>"): ... these.
22539 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22540 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22541 ("*vec_extendv2df"): New insn definitions.
22543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22545 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22546 ("mulditi3_2", "*muldi3_sign"): New patterns.
22547 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22548 rename the pattern definition.
22550 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22552 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22554 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22556 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22558 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22559 instruction if possible.
22560 * config/s390/vector.md (vec_halfnumelts): New mode
22562 ("*vec_vllezlf<mode>"): New pattern.
22564 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22566 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22567 ("popcountv4si2", "popcountv2di2"): Rename to ...
22568 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22569 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22571 ("popcount<mode>2_vxe"): New pattern.
22573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22575 * common/config/s390/s390-common.c (processor_flags_table): Add
22577 * config.gcc: Add arch12.
22578 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22579 Default to arch12 for unknown CPU model numbers.
22580 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22581 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22582 PROCESSOR_max sanity check.
22583 * config/s390/s390-opts.h (enum processor_type): Add
22585 * config/s390/s390.c (processor_table): Add arch12.
22586 (s390_expand_builtin): Add check for B_VXE flag.
22587 (s390_issue_rate): Add PROCESSOR_ARCH12.
22588 (s390_get_sched_attrmask): Likewise.
22589 (s390_get_unit_mask): Likewise.
22590 (s390_sched_score): Enable z13 scheduling for arch12.
22591 (s390_sched_reorder): Likewise.
22592 (s390_sched_variable_issue): Likewise.
22593 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22595 (s390_tune_attr): Use z13 scheduling also for arch12.
22596 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22597 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22598 (TARGET_VXE_P): New macros.
22599 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22600 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22601 * config/s390/s390.opt: Add arch12 as processor_type.
22603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22605 * config/s390/s390.md
22606 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22607 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22608 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22610 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22611 Rename expanders to ...
22613 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22614 ("fixuns_truncdddi2_emu"): ... these.
22616 ("fixuns_trunc<mode>si2_emu"): New expander.
22618 ("*fixuns_truncdfdi2_z13"): Rename to ...
22619 ("*fixuns_truncdfdi2_vx"): ... this.
22621 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22623 * config/s390/2964.md: Remove the single element vector compare
22624 instructions which are no longer used.
22625 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22627 (s390_canonicalize_comparison): Remove handling of DFmode
22629 (s390_expand_vec_compare_scalar): Remove function.
22630 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22631 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22633 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22635 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22637 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22639 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22640 will anyway by matched by mov<mode>_64dfp.
22642 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22644 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22645 vlef/vstef. Add missing operand to vleif.
22647 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22649 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22650 pair for all vector types with 64 bit elements.
22651 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22652 * config/s390/vector.md (V_HW_64): ... here.
22653 (V_128_NOSINGLE): New mode iterator.
22654 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22655 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22656 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22657 ("*vec_load_pairv2di"): Change to ...
22658 ("*vec_load_pair<mode>"): ... this one.
22660 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22662 * config/s390/constraints.md: Add comments.
22663 (jKK): Reject element sizes > 8 bytes.
22664 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22666 * config/s390/s390.md: Add the s_operand checks formerly in
22667 s390_split_ok_p to various splitters where they are still
22669 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22670 for 128 bit vectors. Plus two splitters.
22672 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22674 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22680 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22681 error if the boundary argument is not constant.
22683 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22685 PR rtl-optimization/80112
22686 * loop-doloop.c (doloop_condition_get): Don't check condition
22687 if cmp isn't SET with IF_THEN_ELSE src.
22689 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22691 PR tree-optimization/80158
22692 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22693 replacing a candidate statement, also replace it for the
22694 candidate's alternate interpretation.
22695 (replace_rhs_if_not_dup): Likewise.
22696 (replace_one_candidate): Likewise.
22698 2017-03-24 Richard Biener <rguenther@suse.de>
22700 PR tree-optimization/80167
22701 * graphite-isl-ast-to-gimple.c
22702 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22704 (translate_isl_ast_to_gimple::get_rename): Likewise.
22706 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22708 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22709 handling of certain combinations of target options, including the
22710 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22711 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22713 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22716 * config/arm/arm.md (*load_multiple): Add reload_completed to
22717 matching condition.
22719 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22720 Richard Biener <rguenth@suse.de>
22722 PR tree-optimization/79908
22723 PR tree-optimization/80136
22724 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22725 been cast away, gimplify_and_add suffices.
22727 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22729 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22731 2017-03-23 Richard Biener <rguenther@suse.de>
22733 PR tree-optimization/80032
22734 * gimplify.c (gimple_push_cleanup): Forced unconditional
22735 cleanups still have to go to the conditional_cleanups
22738 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22740 PR tree-optimization/80072
22741 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22743 (next_operand_entry_id): Change type to unsigned int.
22744 (sort_by_operand_rank): Make sure to return the right return value
22745 even if unsigned fields are bigger than INT_MAX.
22746 (struct oecount): Change cnt and id type to unsigned int.
22747 (oecount_hasher::equal): Formatting fix.
22748 (oecount_cmp): Make sure to return the right return value
22749 even if unsigned fields are bigger than INT_MAX.
22750 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22753 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22754 TREE_READONLY on result if writing it more than once.
22757 * doc/invoke.texi (-fsanitize=thread): Document that with
22758 -fnon-call-exceptions atomics are not able to throw
22762 * tsan.c: Include tree-eh.h.
22763 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22764 maybe_clean_or_replace_eh_stmt where needed.
22765 (instrument_memory_accesses): Add cfg_changed argument.
22766 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22767 if it returned true.
22768 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22770 PR rtl-optimization/63191
22771 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22772 wrapper function, moved the whole old content into ...
22773 (ix86_delegitimize_address_1): ... this. New inline function.
22774 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22775 true as last argument instead of ix86_delegitimize_address.
22777 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22779 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22780 cortexa57_branch_cost.
22782 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22784 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22786 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22789 * doc/md.texi (Constraints): Document wA constraint.
22790 * config/rs6000/constraints.md (wA): New.
22791 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22792 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22793 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22794 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22796 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22799 * gimplify.c (is_oacc_declared): New function.
22800 (oacc_default_clause): Use it to set default flags for acc declared
22801 variables inside parallel regions.
22802 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22803 declared variables.
22804 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22805 declare attribute to any decl as necessary.
22807 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22810 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22811 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22812 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22813 (arm_arch_lpae): This.
22814 * config/arm/arm.c (arm_arch7ve): Rename into ...
22815 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22816 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22819 2017-03-22 Martin Liska <mliska@suse.cz>
22822 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22823 error message instead of an ICE.
22825 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22827 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22829 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22831 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22833 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22836 2017-03-21 Martin Sebor <msebor@redhat.com>
22838 * doc/extend.texi: Use "cannot" instead of "can't."
22839 * doc/hostconfig.texi: Same.
22840 * doc/install.texi: Same.
22841 * doc/invoke.texi: Same.
22842 * doc/loop.texi: Same.
22843 * doc/md.texi: Same.
22844 * doc/objc.texi: Same.
22845 * doc/rtl.texi: Same.
22846 * doc/tm.texi: Same.
22847 * doc/tm.texi.in: Same.
22848 * doc/trouble.texi: Same.
22850 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
22853 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
22854 (collect_checksum_attributes): Set it.
22855 (die_checksum_ordered): Use it.
22857 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22859 PR tree-optimization/79908
22860 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
22861 change: For a VA_ARG whose LHS has been cast away, use
22862 force_gimple_operand to construct the side effects.
22864 2017-03-21 David Malcolm <dmalcolm@redhat.com>
22866 PR translation/80001
22867 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
22868 more amenable to translation.
22869 (oacc_loop_auto_partitions): Likewise.
22871 2017-03-21 Marek Polacek <polacek@redhat.com>
22872 Martin Sebor <msebor@redhat.com>
22874 PR tree-optimization/80109
22875 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
22876 on INTEGRAL_TYPE_P.
22878 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22879 Segher Boessenkool <segher@kernel.crashing.org>
22882 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
22883 check reg_used_between_p between insn and one of succ or succ2
22884 depending on if succ is artificial insn not inserted into insn
22887 2017-03-21 Martin Liska <mliska@suse.cz>
22889 PR gcov-profile/80081
22890 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
22891 * doc/gcc.texi: Include gcov-dump stuff.
22892 * doc/gcov-dump.texi: New file.
22894 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
22896 PR rtl-optimization/79150
22897 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
22898 conditional jump, if the jump is the last insn of the loop.
22900 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22901 Richard Biener <rguenth@suse.de>
22903 PR tree-optimization/79908
22904 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22905 been cast away, use force_gimple_operand to construct the side
22908 2017-03-21 Martin Liska <mliska@suse.cz>
22910 PR libfortran/79956
22911 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
22914 2017-03-21 Brad Spengler <spender@grsecurity.net>
22917 * plugin.c (htab_hash_plugin): New function.
22918 (add_new_plugin): Use it and adjust.
22919 (parse_plugin_arg_opt): Adjust.
22920 (init_one_plugin): Likewise.
22922 2017-03-21 Richard Biener <rguenther@suse.de>
22924 PR tree-optimization/80032
22925 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
22926 if set force the cleanup to happen unconditionally.
22927 (gimplify_target_expr): Push inserted clobbers with force_uncond
22928 to avoid them being removed by control-dependent DCE.
22930 2017-03-21 Richard Biener <rguenther@suse.de>
22932 PR tree-optimization/80122
22933 * tree-inline.c (copy_bb): Do not expans va-arg packs or
22934 va_arg_pack_len when the inlined call stmt requires pack
22936 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
22938 2017-03-21 Jakub Jelinek <jakub@redhat.com>
22941 * tsan.c (instrument_builtin_call): If the memory model argument
22942 is not a constant, assume it is valid.
22945 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
22948 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22950 PR rtl-optimization/79910
22951 * combine.c (can_combine_p): Do not allow combining an I0 or I1
22952 if its dest is used by an insn before I2 (other than the combined
22953 insns themselves, which are properly handled already).
22955 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
22958 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
22960 * combine.c (record_used_regs): New static function.
22961 (try_combine): Handle situations where there is an additional
22962 instruction between I2 and I3 which needs to have a LOG_LINK
22966 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
22968 * combine.c (try_combine): Delete redundant i1 test. Call
22969 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
22971 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22974 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
22975 alternatives 13/14.
22977 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22979 PR tree-optimization/80054
22980 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
22981 the optimization if a PHI or any of its arguments is not dominated
22982 by the candidate's basis. Use gphi* rather than gimple* as
22984 (replace_profitable_candidates): Clean up a gimple* variable that
22985 should be a gphi* variable.
22987 2017-03-20 Martin Sebor <msebor@redhat.com>
22990 * doc/extend.texi (attribute constructor): Document present limitation.
22992 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
22995 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
22996 __POWER9_VECTOR__ #ifdef control, change template definition to
22997 use Power9-specific built-in function.
22998 (vec_any_eq): Likewise.
22999 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23000 to control outcomes from this test.
23001 (vector_ae_<mode>p): For VEC_F modes, likewise.
23003 2017-03-20 Ian Lance Taylor <iant@google.com>
23005 * config/i386/i386.c (ix86_function_regparm): Save an extra
23006 register for -fsplit-stack with DECL_STATIC_CHAIN.
23008 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23011 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23012 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23014 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23016 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23018 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23021 2017-03-20 Marek Polacek <polacek@redhat.com>
23024 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23026 2017-03-20 Richard Biener <rguenther@suse.de>
23028 PR tree-optimization/80113
23029 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23030 allocate extra SSA name for PHI def.
23031 (add_close_phis_to_outer_loops): Likewise.
23032 (add_close_phis_to_merge_points): Likewise.
23033 (copy_loop_close_phi_args): Likewise.
23034 (copy_cond_phi_nodes): Likewise.
23036 2017-03-20 Martin Liska <mliska@suse.cz>
23038 PR middle-end/79753
23039 * tree-chkp.c (chkp_build_returned_bound): Do not build
23040 returned bounds for a LHS that's not a BOUNDED_P type.
23042 2017-03-20 Martin Liska <mliska@suse.cz>
23046 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23047 COMPLEX_CST and VECTOR_CST.
23049 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23052 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23053 target operand. A new splitter adds the clobber statement in case
23054 the target operand is dead anyway.
23056 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23058 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23059 to age-old versions of binutils and glibc.
23061 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23063 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23065 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23067 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23069 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23071 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23072 requirement for binutils 2.13.
23074 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23076 * combine.c (try_combine): Delete redundant i1 test. Call
23077 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23079 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23081 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23082 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23084 <riscv64-*-elf>: Re-arrange section
23085 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23086 <riscv32-*-linux>: Likewise.
23087 <riscv64-*-elf>: Likewise
23088 <riscv64-*-linux>: Likewise.
23090 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23093 * aarch64.opt(verbose-cost-dump): Fix typo.
23095 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23098 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23099 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23101 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23103 * reload.c (find_reloads): When reloading a nonoffsettable address,
23104 use RELOAD_OTHER for it and its address reloads.
23106 PR rtl-optimization/79910
23107 * combine.c (record_used_regs): New static function.
23108 (try_combine): Handle situations where there is an additional
23109 instruction between I2 and I3 which needs to have a LOG_LINK
23112 2017-03-17 Jeff Law <law@redhat.com>
23114 PR tree-optimization/71437
23115 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23116 conditional in the hash table first.
23117 (vrp_dom_walker::before_dom_children): Extract condition from
23118 ASSERT_EXPR. Record condition, its inverion and any implied
23119 conditions as well.
23121 2017-03-17 Marek Polacek <polacek@redhat.com>
23122 Markus Trippelsdorf <markus@trippelsdorf.de>
23124 PR tree-optimization/80079
23125 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23128 2017-03-17 Richard Biener <rguenther@suse.de>
23130 PR middle-end/80075
23131 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23132 Properly verify the LHS before the RHS possibly claims to be
23134 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23137 2017-03-17 Martin Jambor <mjambor@suse.cz>
23139 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23140 (List of -O2 options): Likewise.
23141 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23144 2017-03-17 Tom de Vries <tom@codesourcery.com>
23146 * gcov-dump.c (print_usage): Print bug_report_url.
23148 2017-03-17 Richard Biener <rguenther@suse.de>
23150 PR middle-end/80050
23151 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23152 (parser::peek): Likewise.
23154 2017-03-17 Richard Biener <rguenther@suse.de>
23156 PR tree-optimization/80048
23157 * sese.c (free_sese_info): Properly release rename_map and
23158 copied_bb_map elements.
23160 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23162 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23163 Add linked-list forward and backlinks. Insert on
23164 construction, remove on destruction.
23165 (class pass_store_merging): Add m_stores_head field.
23166 (pass_store_merging::terminate_and_process_all_chains):
23167 Iterate over m_stores_head list.
23168 (pass_store_merging::terminate_all_aliasing_chains):
23170 (pass_store_merging::execute): Check for debug stmts first.
23171 Push new chains onto the m_stores_head stack.
23173 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23176 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23177 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23178 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23180 2017-03-16 Jeff Law <law@redhat.com>
23182 PR tree-optimization/71437
23183 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23184 member function. Implementation moved into after_dom_children
23185 member function and into the threader's thread_outgoing_edges
23187 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23188 some code into new thread_outgoing_edges.
23189 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23190 definition. Simplify marker handling (do it here). Assume we always
23191 have the available expression and the const/copies tables.
23192 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23194 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23195 * tree-vrp.c (equiv_stack): No longer file scoped.
23196 (vrp_dom_walker): New class.
23197 (vrp_dom_walker::before_dom_children): New member function.
23198 (vrp_dom_walker::after_dom_children): Likewise.
23199 (identify_jump_threads): Setup domwalker. Use it rather than
23200 walking edges in a random order by hand. Simplify setup/finalization.
23201 (finalize_jump_threads): Remove.
23202 (vrp_finalize): Do not call identify_jump_threads here.
23203 (execute_vrp): Do it here instead and call thread_through_all_blocks
23206 PR tree-optimization/71437
23207 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23209 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23211 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23212 (dom_opt_dom_walker::thread_across_edge): Remove
23213 handle_dominating_asserts argument. All callers changed.
23214 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23215 changes. Remove calls to lhs_of_dominating_assert. Other
23216 uses of handle_dominating_asserts turn into unconditional code
23217 (simplify_control_stmt_condition_1): Likewise.
23218 (simplify_control_stmt_condition): Likewise.
23219 (thread_through_normal_block, thread_across_edge): Likewise.
23220 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23221 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23222 object if it is not an SSA_NAME.
23223 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23224 before calling into the VRP specific simplifiers.
23225 (identify_jump_threads): Remove handle_dominating_asserts
23228 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23231 * tree-diagnostic.c (default_tree_printer): No longer static.
23232 * tree-diagnostic.h (default_tree_printer): New prototype.
23234 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23236 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23237 Change ins into fmov.
23239 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23241 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23242 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23243 Use h_con constraint for operand 1.
23244 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23245 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23247 2017-03-15 Jeff Law <law@redhat.com>
23249 PR tree-optimization/71437
23250 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23251 (record_temporary_equivalences): Use it.
23253 PR tree-optimization/71437
23254 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23255 tree-ssa-scopedtables.
23256 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23257 (record_conditions, record_cond, vuse_eq): Likewise.
23258 (record_edge_info): Adjust to API tweak of record_conditions.
23259 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23260 (record_temporary_equivalences, optimize_stmt): Likewise.
23261 (eliminate_redundant_computations): Likewise.
23262 (record_equivalences_from_stmt): Likewise.
23263 * tree-ssa-scopedtables.c: Include options.h and params.h.
23264 (vuse_eq): New function, moved from tree-ssa-dom.c
23265 (build_and_record_new_cond): Likewise.
23266 (record_conditions): Likewise. Accept vector of conditions rather
23267 than edge_equivalence structure for first argument.
23268 for the first argument.
23269 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23270 from tree-ssa-dom.c.
23271 (avail_exprs_stack::record_cond): Likewise.
23272 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23273 from tree-ssa-dom.c.
23274 (avail_exprs_stack): Add new member functions lookup_avail_expr
23276 (record_conditions): Declare.
23278 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23281 * lra-constraints.c (process_alt_operands): Increase reject for
23282 reloading an input/output operand.
23284 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23287 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23288 insns to convert from signed/unsigned char/short to IEEE 128-bit
23290 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23292 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23295 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23296 subreg of inner mode for values already in registers.
23298 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23300 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23301 iteration reg is used after the loop.
23303 2017-03-14 Martin Sebor <msebor@redhat.com>
23305 PR tree-optimization/79800
23306 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23307 precision in negative-positive range.
23308 (format_floating): Call non-const overload with adjusted precision.
23310 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23313 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23316 2017-03-14 Martin Sebor <msebor@redhat.com>
23318 PR middle-end/80020
23319 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23320 * builtins.def (aligned_alloc): Use it.
23323 * Makefile.in (GTFILES): Add calls.c.
23324 * calls.c: Include "gt-calls.h".
23326 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23328 PR rtl-optimization/79728
23329 * regs.h (struct target_regs): New field
23330 x_contains_allocatable_regs_of_mode.
23331 (contains_allocatable_regs_of_mode): New macro.
23332 * reginfo.c (init_reg_sets_1): Initialize it, and change
23333 contains_reg_of_mode so it includes global regs as well.
23334 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23335 rather than contains_regs_of_mode.
23337 2017-03-14 Martin Liska <mliska@suse.cz>
23339 * doc/invoke.texi: Document options that can't be combined with
23340 -fcheck-pointer-bounds.
23342 2017-03-14 Martin Liska <mliska@suse.cz>
23344 PR middle-end/79831
23345 * doc/invoke.texi (-Wchkp): Document the option.
23347 2017-03-14 Martin Liska <mliska@suse.cz>
23349 * Makefile.in: Install gcov-dump.
23351 2017-03-14 Martin Liska <mliska@suse.cz>
23353 * multiple_target.c (expand_target_clones): Bail out for
23354 an invalid attribute.
23356 2017-03-14 Richard Biener <rguenther@suse.de>
23358 * alias.c (struct alias_set_entry): Pack properly.
23359 * cfgloop.h (struct loop): Likewise.
23360 * cse.c (struct set): Likewise.
23361 * ipa-utils.c (struct searchc_env): Likewise.
23362 * loop-invariant.c (struct invariant): Likewise.
23363 * lra-remat.c (struct cand): Likewise.
23364 * recog.c (struct change_t): Likewise.
23365 * rtl.h (struct address_info): Likewise.
23366 * symbol-summary.h (function_summary): Likewise.
23367 * tree-loop-distribution.c (struct partition): Likewise.
23368 * tree-object-size.c (struct object_size_info): Likewise.
23369 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23370 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23371 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23372 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23373 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23374 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23375 (struct _stmt_vec_info): Likewise.
23377 2017-03-14 Martin Liska <mliska@suse.cz>
23380 * multiple_target.c (create_dispatcher_calls): Check that
23381 a target can create a function dispatcher.
23383 2017-03-14 Martin Liska <mliska@suse.cz>
23386 * multiple_target.c (expand_target_clones): Drop local.local
23387 flag for default implementation.
23389 2017-03-14 Richard Biener <rguenther@suse.de>
23391 PR tree-optimization/80030
23392 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23394 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23396 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23397 gcc_fallthrough() instead of __attribute__((fallthrough));
23399 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23401 * doc/gcc.texi: Remove "up" link to (DIR).
23402 * doc/gccint.texi: Ditto.
23404 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23406 * doc/install.texi (Specific) <avr>: Remove reference to
23409 2017-03-13 Jeff Law <law@redhat.com>
23411 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23412 attribute rather than comments.
23414 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23415 match_scratch operand is highest.
23417 2017-03-13 Martin Liska <mliska@suse.cz>
23419 PR middle-end/78339
23420 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23421 is a CHKP clone, use original declaration.
23423 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23425 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23426 (arc_conditional_register_usage): Use a different allocation order
23427 when optimizing for size.
23428 * common/config/arc/arc-common.c (arc_option_optimization_table):
23429 Section anchors default on when optimizing for size.
23431 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23433 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23435 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23437 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23438 * config/arc/arc.md (cpu_facility): Add cd variant.
23439 (*movqi_insn): Add code density variant.
23440 (*movhi_insn): Likewise.
23441 (*movqi_insn): Likewise.
23442 (*addsi3_mixed): Likewise.
23443 (subsi3_insn): Likewise.
23445 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23447 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23449 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23451 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23452 expressions with MINUS and UNARY ops.
23454 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23457 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23459 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23460 between vec_select and vector argument.
23461 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23462 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23463 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23464 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23465 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23466 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23468 2017-03-13 Richard Biener <rguenther@suse.de>
23471 * params.def (vect-max-peeling-for-alignment): Fix typo.
23473 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23475 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23476 issue that only occurred with binutils below 2.18.
23478 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23480 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23481 refer to binutils 2.11/2.12 minimum.
23483 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23485 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23486 ftp.kernel.org and simplify binutils requirement.
23488 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23490 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23492 (Optimize Options): Ditto. Also remove redundancy.
23494 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23496 PR translation/79848
23497 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23499 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23500 to G_ to avoid double translation.
23502 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23504 PR translation/79923
23505 * auto-profile.c (get_combined_location): Convert leading
23506 character of diagnostics to lower case and remove trailing period.
23507 (read_profile): Likewise for various diagnostics.
23508 * config/arm/arm.c (arm_option_override): Remove trailing period
23509 from various diagnostics.
23510 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23511 (msp430_expand_delay_cycles): Likewise.
23513 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23516 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23517 full command-line argument, rather than just "str".
23518 (aarch64_validate_march): Likewise.
23519 (aarch64_validate_mtune): Likewise.
23521 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23523 PR rtl-optimization/78911
23524 * lra-assigns.c (must_not_spill_p): New function.
23525 (spill_for): Use it.
23527 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23529 PR tree-optimization/79981
23530 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23531 ATOMIC_COMPARE_EXCHANGE ifn result.
23532 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23533 IFN_ATOMIC_COMPARE_EXCHANGE.
23535 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23538 * opts.c (parse_sanitizer_options): Add missing question mark to
23539 "did you mean" message.
23541 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23543 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23545 (VMULEUH_UNS): Likewise.
23546 (VMULOUB_UNS): Likewise.
23547 (VMULOUH_UNS): Likewise.
23548 * config/rs6000/rs6000.c (builtin_function_type): Remove
23549 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23551 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23554 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23555 x with result of extra_parsing_for_operand_code_0.
23556 (function_reader::extra_parsing_for_operand_code_0): Convert
23557 return type from void to rtx, returning x. When reading
23558 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23559 larger size containing struct block_symbol.
23561 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23563 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23564 -mfloat128-hardware without -m64.
23566 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23569 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23570 entries to the case statement that marks unsigned arguments to
23571 overloaded functions.
23573 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23575 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23576 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23578 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23581 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23582 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23584 2017-03-10 Martin Liska <mliska@suse.cz>
23588 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23589 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23592 2017-03-10 Olivier Hainque <hainque@adacore.com>
23594 * tree-switch-conversion (array_value_type): Start by resetting
23595 candidate type to it's main variant.
23597 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23599 PR rtl-optimization/79909
23600 * combine.c (try_combine): Use simplify_replace_rtx on individual
23601 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23602 of the whole CALL_INSN_FUNCTION_USAGE.
23604 PR tree-optimization/79972
23605 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23606 get_range_info on SSA_NAMEs. Formatting fixes.
23608 2017-03-10 Richard Biener <rguenther@suse.de>
23609 Jakub Jelinek <jakub@redhat.com>
23611 PR tree-optimization/77975
23612 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23613 edge to be constant.
23614 (get_val_for): For constant x return it. Formatting fix.
23615 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23616 would use the same bases as the current one.
23618 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23620 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23621 instead of vec_select for V1TImode.
23622 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23624 (VSX_LE_128): Add V1TI to this mode iterator.
23625 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23626 (*vsx_le_perm_store_<mode>): Likewise.
23627 (pre-reload splitter for VSX stores): Likewise.
23628 (post-reload splitter for VSX stores): Likewise.
23629 (*vsx_xxpermdi2_le_<mode>): Likewise.
23630 (*vsx_lxvd2x2_le_<mode>): Likewise.
23631 (*vsx_stxvd2x2_le_<mode>): Likewise.
23633 2017-03-09 Michael Eager <eager@eagercon.com>
23635 Correct failures with --enable-checking=yes,rtl.
23637 * config/microblaze/microblaze.c (microblaze_expand_shift):
23638 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23639 test for const0_rtx.
23640 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23641 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23643 2017-03-09 Richard Biener <rguenther@suse.de>
23645 PR tree-optimization/79977
23646 * graphite-scop-detection.c (scop_detection::merge_sese):
23647 Handle the case of extra exits to blocks dominating the entry.
23649 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23651 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23654 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23656 PR rtl-optimization/79949
23657 * lra-constraints.c (process_alt_operands): Check memory when
23658 trying to predict a cycle. Print about the overall increase.
23660 2017-03-09 Richard Biener <rguenther@suse.de>
23662 PR middle-end/79971
23663 * gimple-expr.c (useless_type_conversion_p): Preserve
23664 TYPE_SATURATING for fixed-point types.
23666 2017-03-09 Richard Biener <rguenther@suse.de>
23669 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23670 alignment of BLKmode params.
23672 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23675 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23676 (VALL_NO_V2Q): Likewise.
23678 * config/aarch64/aarch64-simd.md
23679 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23681 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23682 VALL_NO_V2Q mode iterator.
23683 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23685 2017-03-09 Martin Liska <mliska@suse.cz>
23687 PR tree-optimization/79631
23688 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23689 tree_int_cst_sign_bit just for INTEGER constants.
23691 2017-03-09 Martin Liska <mliska@suse.cz>
23695 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23698 2017-03-09 Marek Polacek <polacek@redhat.com>
23701 * tree.c (inchash::add_expr): Handle TREE_VEC.
23703 2017-03-09 Martin Liska <mliska@suse.cz>
23706 (chkp_narrow_size_and_offset): New function.
23707 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23708 (void chkp_parse_bit_field_ref): New function.
23709 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23710 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23712 2017-03-09 Martin Liska <mliska@suse.cz>
23715 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23716 (chkp_find_bounds_1): Remove gcc_unreachable.
23718 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23721 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23722 BUILT_IN_SYNC*, determine the access type from the size suffix and
23723 always build a MEM_REF with that type. Handle forgotten
23724 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23727 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23728 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23729 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23730 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23731 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23732 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23733 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23734 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23735 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23736 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23737 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23738 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23739 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23740 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23741 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23742 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23743 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23744 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23745 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23746 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23747 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23748 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23749 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23750 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23751 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23752 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23753 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23754 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23755 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23756 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23757 definitions outside of __OPTIMIZE__ guarded section.
23760 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23761 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23762 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23763 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23766 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23768 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23769 ("vfenez<mode>"): Add missing constraints.
23771 2017-03-08 Martin Sebor <msebor@redhat.com>
23774 * config/nds32/nds32.c (nds32_option_override):
23775 Fix misspelled diagnostic.
23777 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23780 * gimplify.c (gimplify_omp_for): Replace index var in outer
23781 taskloop statement with an artificial variable and add
23782 OMP_CLAUSE_PRIVATE clause for it.
23784 2017-03-08 Richard Biener <rguenther@suse.de>
23786 PR tree-optimization/79955
23787 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23788 for accesses that are completely outside of the variable.
23790 2017-03-08 Andrew Haley <aph@redhat.com>
23792 PR tree-optimization/79943
23793 * tree-ssa-loop-split.c (compute_new_first_bound): When
23794 calculating the new upper bound, (END-BEG) should be added, not
23797 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23799 * config/avr/avr.md (setmemhi): Make sure match_dup
23800 operand number comes before match_scratch.
23802 2017-03-08 Richard Biener <rguenther@suse.de>
23804 PR tree-optimization/79920
23805 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23806 with ncopies == 1 to ...
23807 (vect_transform_slp_perm_load): ... here. Properly compute
23808 all element loads by iterating VF times over the group. Do
23809 not handle ncopies (computed in a broken way) in
23810 vect_create_mask_and_perm.
23812 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23815 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23816 is a uniform vector, use uniform_vector_p return value instead of
23817 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23819 2017-03-07 Marek Polacek <polacek@redhat.com>
23821 PR middle-end/79809
23822 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23823 (alloca_call_type): Likewise.
23825 2017-03-07 Martin Liska <mliska@suse.cz>
23827 * gcov.c (process_args): Put comment to correct location.
23829 2017-03-07 Martin Liska <mliska@suse.cz>
23831 PR middle-end/68270
23832 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23833 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23834 (chkp_narrow_bounds_for_field): Likewise.
23835 (chkp_parse_array_and_component_ref): Pass one more argument to
23838 2017-03-07 Richard Biener <rguenther@suse.de>
23840 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
23843 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23845 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
23846 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
23848 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23851 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
23852 to end of description.
23853 (PARAM_MAX_STORES_TO_MERGE): Likewise.
23855 2017-03-07 Jakub Jelinek <jakub@redhat.com>
23857 PR rtl-optimization/79901
23858 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
23860 (*avx512f_<code><mode>3<mask_name>): ... this.
23861 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
23862 iterator instead of VI8_AVX2_AVX512BW.
23864 PR rtl-optimization/79901
23865 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
23866 min/max expander, expand it using expand_vec_cond_expr.
23869 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
23872 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23875 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
23876 to void * for PCH reasons.
23877 * dwarf2out.c (output_loc_operands, output_die): Cast
23878 v.val_vec.array to unsigned char *.
23880 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
23883 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
23886 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
23888 PR rtl-optimization/79571
23889 * lra-constraints.c (process_alt_operands): Calculate static
23890 reject and subtract it from overall when only addresses will be
23893 2017-03-06 Julia Koval <julia.koval@intel.com>
23896 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
23897 incoming stack boundary to 128 for 64-bit targets.
23899 2017-03-06 Richard Biener <rguenther@suse.de>
23901 PR tree-optimization/79894
23902 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
23903 to NULL after folding it.
23905 2017-03-06 Richard Biener <rguenther@suse.de>
23907 PR tree-optimization/79824
23908 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
23909 check disabling peeling for gaps.
23911 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
23913 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
23914 attributes): Document gettimeofday.
23916 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23918 * config/s390/s390.c (s390_option_override_internal): Set
23919 PARAM_MIN_VECT_LOOP_BOUND
23921 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
23923 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
23924 * config/s390/s390.md: Likewise.
23926 2017-03-06 Jakub Jelinek <jakub@redhat.com>
23929 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
23930 (<avx2_avx512>_perm<mode>): Rename to ...
23931 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
23933 (<avx512>_perm<mode>_mask): Rename to ...
23934 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
23936 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
23937 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
23938 instead of VI8F_256_512.
23939 (avx512f_perm<mode>): New define_expand.
23940 (avx512f_perm<mode>_mask): Likewise.
23941 (avx512f_perm<mode>_1<mask_name>): New define_insn.
23942 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
23944 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23946 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
23947 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
23949 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
23951 2017-03-06 Martin Liska <mliska@suse.cz>
23954 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
23955 when having a SSA NAME w/o VAR_DECL assigned to it.
23957 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23959 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
23960 msa_dpsub_<su>_d): Fix MODE for vec_select.
23962 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
23964 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
23966 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
23968 2017-03-06 Richard Biener <rguenther@suse.de>
23970 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
23971 * plugin.c (register_plugin_info): Likewise.
23972 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
23974 2017-03-05 Jakub Jelinek <jakub@redhat.com>
23976 * config/i386/sse.md (sse_storehps, sse_storelps,
23977 avx_<castmode><avxsizesuffix>_<castmode>,
23978 avx512f_<castmode><avxsizesuffix>_<castmode>,
23979 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
23980 in condition that at least one operand is not a MEM.
23982 2017-03-03 Jakub Jelinek <jakub@redhat.com>
23984 PR middle-end/79805
23985 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
23986 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
23988 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
23989 gimple_call_nothrow_p flag based on whether original builtin can throw.
23990 If it can, emit following stmts on the fallthrough edge.
23991 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
23992 don't create new bb if inserting just debug stmts on the edge, try to
23993 insert them on the fallthru bb or just reset debug stmts.
23995 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
23998 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
23999 restore recog_data (including the operand rtxes inside it) around
24000 the call to get_insn_template.
24002 2017-03-03 Martin Sebor <msebor@redhat.com>
24004 PR tree-optimization/79699
24005 * context.c (context::~context): Free MPFR caches to avoid
24006 a memory leak on program exit.
24008 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24010 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24011 Use wide_int::ulow () instead of .elt (0).
24013 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24015 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24016 (*pushxf): Limit oF constraint to 32bit targets and add oC
24017 constraint for 64bit targets.
24018 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24019 (*pushdf): Change rmF constraint to rmC.
24021 2017-03-03 Martin Liska <mliska@suse.cz>
24023 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24024 Remove unused variable.
24026 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24029 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24030 is a memory operand, increase num_memory.
24031 (ix86_expand_args_builtin): Likewise.
24033 2017-03-03 Jan Hubicka <jh@suse.cz>
24036 * ipa-devirt.c (maybe_record_node): Properly handle
24037 __cxa_pure_virtual visibility.
24039 2017-03-03 Martin Liska <mliska@suse.cz>
24041 PR tree-optimization/79803
24042 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24044 (pass_loop_prefetch::execute): Disabled optimization if an
24045 assumption about L1 cache size is not met.
24047 2017-03-03 Martin Liska <mliska@suse.cz>
24049 PR rtl-optimization/79574
24050 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24051 (hash_scan_set): Likewise.
24052 (dump_hash_table): Likewise.
24053 (hoist_code): Likewise.
24055 2017-03-03 Richard Biener <rguenther@suse.de>
24057 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24058 in place of uhigh ().
24059 (fixed_convert_from_real): Likewise.
24061 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24064 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24066 2017-03-03 Richard Biener <rguenther@suse.de>
24068 PR middle-end/79818
24069 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24070 TYPE_OVERFLOW_UNDEFINED check.
24072 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24074 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24076 (vector_ae_<mode>_p): Likewise.
24077 (vector_nez_<mode>_p): Likewise.
24078 (vector_ne_v2di_p): Likewise.
24079 (vector_ae_v2di_p): Likewise.
24080 (vector_ne_<mode>_p): Likewise.
24081 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24083 (vsx_tsqrt<mode>2_fe): Likewise.
24085 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24088 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24090 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24092 PR rtl-optimization/79780
24093 * cprop.c (one_cprop_pass): When second and further conditional trap
24094 in a single basic block is turned into an unconditional trap, turn it
24095 into a deleted note to avoid RTL verification failures.
24097 2017-03-02 Richard Biener <rguenther@suse.de>
24099 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24101 2017-03-02 Richard Biener <rguenther@suse.de>
24103 PR tree-optimization/79345
24105 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24106 param and abort the walk, returning -1 if it is hit.
24107 (walk_aliased_vdefs): Take a limit param and pass it on.
24108 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24109 defaulting to 0 and return a signed int.
24110 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24111 (check_defs): New helper.
24112 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24113 about uninitialized memory.
24114 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24115 bogus uninitialized warning.
24116 (fixed_convert_from_real): Likewise.
24118 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24120 PR tree-optimization/66768
24121 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24122 iv_use if base object can't be determined.
24124 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24126 PR tree-optimization/79345
24127 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24128 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24129 (get_pattern_stats): Initialize it.
24130 * genemit.c (gen_expand): Verify match_scratch numbers come after
24131 match_operand/match_dup numbers.
24132 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24133 match_scratch numbers.
24134 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24136 * config/s390/s390.md (trunctdsd2): Likewise.
24138 2017-03-02 Richard Biener <rguenther@suse.de>
24140 * wide-int.h (wide_int_storage::operator=): Implement in terms
24143 2017-03-02 Richard Biener <rguenther@suse.de>
24145 PR tree-optimization/79777
24146 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24147 the to insert expression to sth existing.
24149 2017-03-01 Martin Sebor <msebor@redhat.com>
24151 PR middle-end/79692
24152 * gimple-ssa-sprintf.c
24153 (directive::known_width_and_precision): New function.
24154 (format_integer): Use it.
24155 (get_mpfr_format_length): Consider the full range of precision
24156 when computing %g output with the # flag. Set the likely byte
24157 count to 3 rather than 1 when precision is indeterminate.
24158 (format_floating): Correct the lower bound of precision.
24160 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24162 * doc/invoke.texi: Document default code model for 64-bit Linux.
24164 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24167 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24168 udiv rather than div since input pattern is unsigned.
24170 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24172 * config/i386/i386.c (print_reg): Warn for values of
24173 unsupported size in integer register.
24175 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24178 * config/rs6000/predicates.md (current_file_function_operand): Do
24179 not allow self calls to be local if the function is replaceable.
24181 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24184 * config/rs6000/altivec.h (vec_ctz and others): Change the
24185 preprocessor macro that controls conditional compilation from
24186 _ARCH_PWR9 to __POWER9_VECTOR__.
24187 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24188 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24189 control (instead of _ARCH_PWR9 control) so that template
24190 definition uses power9-specific function.
24191 (vec_any_eq): Likewise.
24192 (vec_all_ne): Change macro definition to use a power9-specific
24193 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24194 _ARCH_PWR9 control).
24195 (vec_any_eq) Likewise.
24196 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24197 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24198 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24199 support for xvcmpnedp instruction.
24200 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24201 macro expansion so that Power9 implementation of vec_all_ne does
24202 not use the AltiVec predicate framework.
24203 (VCMPNEH_P): Likewise.
24204 (VCMPNEW_P): Likewise.
24205 (VCMPNED_P): Likewise.
24206 (VCMPNEFP_P): Likewise.
24207 (VCMPNEDP_P): Likewise.
24208 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24209 implementation of vec_any_eq to not use AltiVec predicate
24211 (VCMPAEH_P): Likewise.
24212 (VCMPAEW_P): Likewise.
24213 (VCMPAED_P): Likewise.
24214 (VCMPAEFP_P): Likewise.
24215 (VCMPAEDP_P): Likewise.
24216 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24217 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24218 not use the AltiVec predicate framework.
24219 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24220 of vec_any_eq to not use AltiVec predicate framework.
24221 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24222 support for predefined __POWER9_VECTOR__ macro to indicate that
24223 Power9 instruction selection is enabled.
24224 (altivec_overloaded_builtins): Remove extraneous
24225 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24226 function argument types RS6000_BTI_bool_V16QI and
24227 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24228 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24229 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24230 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24231 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24232 Power9 for implementations of vec_cmpne. Change the signature for
24233 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24234 (representing vec_all_ne) to remove the previously described first
24235 argument of type RS6000_BTI_INTSI, as this was an artifact of
24236 reliance on the AltiVec predicate framework, which is no longer
24237 used in the implementation of these functions. Add
24238 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24239 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24240 since, unlike the AltiVec predicate framework implementation, we
24241 do not share function descriptors between vec_alle and vec_anyeq.
24242 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24243 set of modes that receive special treatment even when
24244 TARGET_P9_VECTOR is true. The special treatment emits code that
24245 does not depend on Power9 instructions.
24246 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24247 define_expand to not rely on AltiVec predicate framework.
24248 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24250 (vector_ne_v2di_p): Change this define_expand to not rely on
24251 AltiVec predicate framework.
24252 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24254 (vector_ne_<mode>_p): Change this define_expand to not rely on
24255 AltiVec predicate framework.
24256 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24258 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24259 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24260 define_insn pattern.
24261 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24262 define_insn pattern because the xvcmpne<VSs>. instruction is not
24264 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24265 instruction is not supported.
24267 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24269 * config/nvptx/nvptx.c: Include intl.h.
24271 2017-03-01 Martin Jambor <mjambor@suse.cz>
24274 * ipa-prop.h (ipa_bits): Removed field known.
24275 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24276 to pointers. Adjusted their comments to warn about their sharing.
24277 (ipcp_transformation_summary): Change bits to a vector of pointers.
24278 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24279 (ipa_get_ipa_bits_for_value): Declare.
24280 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24281 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24282 (ipa_bits_hash_table): Likewise.
24283 (ipa_vr_ggc_hash_traits): Likewise.
24284 (ipa_vr_hash_table): Likewise.
24285 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24286 being pointers and vr_known being removed.
24287 (ipa_set_jf_unknown): Likewise.
24288 (ipa_get_ipa_bits_for_value): New function.
24289 (ipa_set_jfunc_bits): Likewise.
24290 (ipa_get_value_range): New overloaded functions.
24291 (ipa_set_jfunc_vr): Likewise.
24292 (ipa_compute_jump_functions_for_edge): Use the above functions to
24293 construct bits and vr parts of jump functions.
24294 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24295 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24297 (ipcp_grow_transformations_if_necessary): Also allocate
24298 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24300 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24301 them. Fix too long lines.
24302 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24303 vr_known being removed.
24304 (ipa_read_jump_function): Use new setter functions to construct bits
24305 and vr parts of jump functions or set them to NULL.
24306 (write_ipcp_transformation_info): Adjust for bits being pointers.
24307 (read_ipcp_transformation_info): Likewise.
24308 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24310 Include gt-ipa-prop.h.
24311 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24313 (ipcp_store_bits_results): Likewise.
24314 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24315 Do not write to existing jump functions but use a temporary instead.
24317 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24320 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24321 attempt to use its first operand as BIT_FIELD_REF base.
24323 2017-03-01 Richard Biener <rguenther@suse.de>
24325 PR middle-end/79721
24326 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24327 interpolating formula in wrapping arithmetic.
24328 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24330 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24332 PR tree-optimization/79734
24333 * tree-vect-generic.c (expand_vector_condition): Optimize
24334 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24335 Handle VEC_COND_EXPR where comparison has different inner width from
24336 type's inner width.
24338 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24340 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24341 markup, and similar issues. Remove @opindex entries for things
24342 that aren't options. Add missing -mmpy-option entries.
24344 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24346 PR tree-optimization/79737
24347 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24348 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24349 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24350 instead of byte_size. Formatting fix.
24351 (shift_bytes_in_array_right): Formatting fix.
24353 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24356 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24357 condition on optimize for the leaf function test.
24359 2017-02-28 Martin Liska <mliska@suse.cz>
24362 * read-rtl-function.c (function_reader::handle_unknown_directive):
24363 Bail out when one uses -flto.
24365 2017-02-28 Martin Liska <mliska@suse.cz>
24367 * common.opt: Replace space with tabular for options of <number>
24369 * config/i386/i386.opt: Show <number> value for
24370 -mlarge-data-threshold.
24371 * opts.c (print_filtered_help): Do not display number in hexadecimal
24374 2017-02-28 Martin Liska <mliska@suse.cz>
24376 * common.opt: Fix --help=option -Q for options which are of
24379 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24381 * config/i386/i386.c (print_reg): Error out for values
24382 of 8-bit size in invalid integer register.
24384 2017-02-28 Martin Sebor <msebor@redhat.com>
24386 PR tree-optimization/79691
24387 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24389 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24392 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24393 gcc_unreachable with output_operand_lossage.
24395 2017-02-28 Richard Biener <rguenther@suse.de>
24397 PR tree-optimization/79740
24398 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24400 (visit_nary_op): Insert the nary into the hashtable if we
24401 pattern-matched sth.
24402 * tree-ssa-pre.c (eliminate_insert): Robustify.
24404 2017-02-28 Richard Biener <rguenther@suse.de>
24406 PR middle-end/79731
24407 * fold-const.c (decode_field_reference): Reject out-of-bound
24410 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24412 * config/i386/i386.c: Include intl.h.
24413 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24414 instead of just cond ? "..." : "...".
24415 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24416 * coverage.c (read_counts_file): Likewise.
24417 * omp-offload.c: Include intl.h.
24418 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24419 of just cond ? "..." : "...".
24420 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24421 of just cond ? "..." : "...".
24423 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24426 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24428 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24429 'tune for' CPU name.
24430 * config/arm/arm-cpu-data.h: Regenerated.
24432 2017-02-28 Richard Biener <rguenther@suse.de>
24434 PR tree-optimization/79732
24435 * tree-inline.c (expand_call_inline): Do not shadow var.
24437 2017-02-28 Richard Biener <rguenther@suse.de>
24439 PR tree-optimization/79723
24440 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24441 address-space properly.
24443 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24445 * doc/optinfo.texi (Optimization groups): Fix option used for
24447 * doc/invoke.texi (-fopt-info): Document "omp".
24448 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24449 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24450 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24451 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24452 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24454 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24456 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24457 OPTGROUP_OMP with "omp".
24459 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24462 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24463 for arithmetic shift of unsigned V2DI.
24465 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24467 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24468 arc/linux.h headers.
24469 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24470 (LINK_SPEC): Likewise.
24471 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24472 (EXTRA_SPECS): Likewise.
24473 (STARTFILE_SPEC): Likewise.
24474 (ENDFILE_SPEC): Likewise.
24475 (LIB_SPEC): Likewise.
24476 (TARGET_SDATA_DEFAULT): Likewise.
24477 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24478 (MULTILIB_DEFAULTS): Likewise.
24479 (DWARF2_UNWIND_INFO): Likewise.
24480 * config/arc/big.h: New file.
24481 * config/arc/elf.h: Likewise.
24482 * config/arc/linux.h: Likewise.
24483 * config/arc/t-uClibc: Remove.
24485 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24487 PR tree-optimization/77536
24488 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24489 (tree_transform_and_unroll_loop): Use above function to compute the
24490 estimated niter of unrolled loop and use it when scaling profile.
24491 Also use count info rather than frequency if it's non-zero.
24492 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24493 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24494 (vect_transform_loop): Call above function.
24496 2017-02-27 Richard Biener <rguenther@suse.de>
24498 PR tree-optimization/45397
24499 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24500 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24501 (visit_nary_op): Add pattern matching for CSEing sign-changed
24502 or truncated operations with wider ones.
24504 2017-02-27 Richard Biener <rguenther@suse.de>
24506 PR tree-optimization/79690
24507 * tree-vect-stmts.c (vectorizable_store): Use vector type
24508 built from the DR with address-space.
24510 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24512 * doc/invoke.texi (Optimize Options): Refine the description
24513 of asan-use-after-return.
24515 2017-02-25 Alan Modra <amodra@gmail.com>
24517 PR rtl-optimization/79584
24518 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24519 base, not ad->base_term, the reg within base. Remove assertion
24520 that ad->base == ad->base_term. Replace gen_int_mode using
24521 bogus mode with const0_rtx.
24523 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24525 PR middle-end/79396
24526 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24527 FMA_EXPR like tcc_binary or tcc_unary.
24529 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24532 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24534 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24535 (output_loc_operands): Handle DW_OP_call_ref and
24536 DW_OP_GNU_variable_value.
24537 (struct variable_value_struct): New type.
24538 (struct variable_value_hasher): Likewise.
24539 (variable_value_hash): New variable.
24540 (string_types): Remove.
24541 (copy_loc_descr): New function.
24542 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24543 (prepend_loc_descr_to_each): New function.
24544 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24545 instead of add_loc_descr_to_each if the first argument is single
24546 location list and the second has multiple.
24547 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24548 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24549 when looking for variable value which doesn't have other location info.
24550 (loc_list_from_tree): Formatting fix.
24551 (gen_array_type_die): Simplify DW_AT_string_length handling.
24552 (adjust_string_types): Remove.
24553 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24554 string_types. Call resolve_variable_values.
24555 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24556 (resolve_addr_in_expr): Likewise. Add A argument.
24557 (copy_deref_exprloc): Remove deref argument. Adjust for the
24558 original expression being DW_OP_GNU_variable_value with optionally
24559 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24560 optionally after it.
24561 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24562 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24563 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24564 (variable_value_hasher::hash, variable_value_hasher::equal): New
24566 (resolve_variable_value_in_expr, resolve_variable_value,
24567 resolve_variable_values, note_variable_value_in_expr,
24568 note_variable_value): New functions.
24569 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24572 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24575 * opts.h (handle_generated_option): Add GENERATED_P argument.
24576 * opts-common.c (handle_option): Adjust function comment.
24577 (handle_generated_option): Add GENERATED_P argument, pass it to
24579 (control_warning_option): Pass false to handle_generated_option
24581 * opts.c (maybe_default_option): Pass true to handle_generated_option
24583 * optc-gen.awk: Likewise.
24585 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24587 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24588 a REG, look at the REG it is a SUBREG of.
24589 (splitter for cmpeqsi_t): Ditto.
24591 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24593 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24594 the special USEs with the pattern of the insn, not the insn itself.
24596 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24599 * doc/invoke.texi: Document -mload-store-pairs.
24601 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24602 Sandra Loosemore <sandra@codesourcery.com>
24604 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24605 argument isn't a CONST_INT.
24606 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24607 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24608 (nios2_validate_compare): Bypass alternate compare logic if *op2
24609 is not a CONST_INT.
24610 (ldstwm_operation_p): Return false if first_base is not a REG or
24611 if first_offset is not a CONST_INT.
24613 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24615 * config/cris/cris.md: Use correct operand in a define_peephole2.
24617 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24619 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24621 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24623 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24624 this_insn if it is an INSN or JUMP_INSN.
24625 (force_offsettable): Look at base, not at addr.
24626 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24627 on things that aren't necessarily CONST_INTs.
24629 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24631 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24632 -mfpmath=sse is the default also for x86-32 targets with SSE2
24633 instruction set when @option{-ffast-math} is enabled
24635 2017-02-24 Jeff Law <law@redhat.com>
24637 PR rtl-optimizatoin/79286
24638 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24641 2017-02-24 Richard Biener <rguenther@suse.de>
24643 PR tree-optimization/79389
24644 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24647 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24649 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24650 function comment to reflect reality.
24651 (loop_exits_before_overflow): Fix typo in function description.
24653 2017-02-24 Richard Biener <rguenther@suse.de>
24655 PR tree-optimization/79389
24656 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24657 properly that a threading opportunity exists. Detect conditional
24658 copy/constant propagation opportunities.
24660 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24662 * config/visium/visium.md (type): Add trap.
24663 (b): New mode attribute.
24664 (*btst): Rename into...
24665 (*btst<mode>): ...this and adjust.
24666 (*cbranchsi4_btst_insn): Rename into...
24667 (*cbranch<mode>4_btst_insn): ...this and adjust.
24668 (trap): New define_insn.
24670 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24672 PR tree-optimization/79389
24673 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24674 (noce_reversed_cond_code): New function.
24675 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24676 reversed_comparison_code. Formatting fix.
24677 (noce_try_store_flag): Test rev_cond != NULL in addition to
24678 reversed_comparison_code.
24679 (noce_try_store_flag_constants): Likewise.
24680 (noce_try_store_flag_mask): Likewise.
24681 (noce_try_addcc): Use rev_cond if non-NULL instead of
24682 reversed_comparison_code.
24683 (noce_try_cmove_arith): Likewise. Formatting fixes.
24684 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24685 (noce_find_if_block): Initialize rev_cond.
24686 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24687 instead of false as last argument never attempt to reverse it
24690 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24692 PR tree-optimization/79663
24693 * tree-predcom.c (combine_chains): Process refs in reverse order
24694 only for ZERO length chains, and add explaining comment.
24696 2017-02-23 Jeff Law <law@redhat.com>
24698 PR tree-optimization/79578
24699 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24700 in call to operand_equal_p.
24702 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24705 * config/i386/cpuid.h: Fix another undefined behavior.
24707 2017-02-23 Richard Biener <rguenther@suse.de>
24709 PR tree-optimization/79683
24710 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24711 vector types for data-refs.
24713 2017-02-23 Martin Liska <mliska@suse.cz>
24715 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24717 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24719 PR middle-end/79665
24720 * internal-fn.c (get_range_pos_neg): Moved to ...
24721 * tree.c (get_range_pos_neg): ... here. No longer static.
24722 * tree.h (get_range_pos_neg): New prototype.
24723 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24724 are known to be in between 0 and signed maximum inclusive, try to
24725 expand both unsigned and signed divmod and use the cheaper one from
24728 2017-02-22 Jeff Law <law@redhat.com>
24730 PR tree-optimization/79578
24731 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24732 to compare base operands.
24734 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24737 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24738 gpc_reg_operand instead of fpr_reg_operand.
24740 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24742 * config/mips/mips.c (mips_return_in_memory): Force FP
24743 vector types to be returned in memory for o32 ABI.
24745 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24747 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24748 instead of DW_TAG_member for static data member declarations and don't
24749 set no_linkage_name for static inline data members.
24750 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24753 2017-02-22 Martin Liska <mliska@suse.cz>
24755 * doc/invoke.texi: Replace inequality signs with square brackets
24758 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24760 PR tree-optimization/68644
24761 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24763 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24766 * lra-constraints.c (simplify_operand_subreg): Handle
24767 WORD_REGISTER_OPERATIONS targets.
24769 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24772 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24773 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24774 elimination by swapping fld*.
24776 2017-02-22 Richard Biener <rguenther@suse.de>
24778 PR tree-optimization/79673
24779 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24780 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24781 irrelevant address-space qualifiers and avoiding a
24782 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24784 2017-02-22 Richard Biener <rguenther@suse.de>
24786 PR tree-optimization/79666
24787 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24788 to not symbolically negate if that may introduce undefined
24791 2017-02-22 Martin Liska <mliska@suse.cz>
24794 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24795 * data-streamer-out.c (streamer_write_gcov_count_stream):
24797 * value-prof.c (stream_out_histogram_value): Make assert more
24798 precise based on type of counter.
24800 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24803 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24804 Use nonimmediate_operand instead of memory_operand for operand 1.
24805 (float-extend standard_x87sse_constant_load splitter): Ditto.
24807 2017-02-21 Jeff Law <law@redhat.com>
24809 PR tree-optimization/79621
24810 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24811 blocks with edges to themselves.
24813 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24816 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24817 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24818 Use gimple_call_builtin_p.
24821 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24822 on temporarily removed DEBUG_INSNs.
24824 PR tree-optimization/79649
24825 * tree-loop-distribution.c (classify_partition): Give up on
24826 non-generic address space loads/stores.
24828 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24830 * doc/loop.texi (Loop manipulation): Remove nonexistent
24831 tree_ssa_loop_version from the documentation.
24832 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24834 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24837 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24838 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24839 * config/rs6000/rs6000.c: Include except.h.
24840 (rs6000_expand_split_stack_prologue): Call
24841 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
24843 2017-02-21 Martin Jambor <mjambor@suse.cz>
24846 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
24847 have been analyzed.
24849 2017-02-21 Martin Jambor <mjambor@suse.cz>
24851 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
24852 for backward compatibility only.
24853 * doc/invoke.texi (Option Summary): Remove all references to
24854 -fipa-cp-alignment.
24856 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
24860 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24862 * lra-constraints.c (curr_insn_transform): Handle
24863 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24865 2017-02-21 Martin Liska <mliska@suse.cz>
24867 * config/i386/i386.opt: Replace -masm-dialect with -masm.
24869 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
24871 PR translation/79638
24872 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
24874 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
24877 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
24878 (arm_function_ok_for_sibcall): Return false for an indirect call by
24879 descriptor if all the argument registers are used.
24880 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
24881 alignment of the function.
24883 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24885 PR tree-optimization/61441
24886 * simplify-rtx.c (simplify_const_unary_operation): For
24887 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
24888 the sNaN unmodified.
24890 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24892 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24893 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24894 instead of SYSTEM_HEADER_DIR.
24896 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
24897 Martin Liška <mliska@suse.cz>
24899 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
24900 Fix typos and grammar, use active voice, and clarify.
24902 2017-02-20 Marek Polacek <polacek@redhat.com>
24904 PR middle-end/79537
24905 * gimplify.c (gimplify_expr): Handle unused *&&L;.
24908 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
24910 2017-02-20 Jakub Jelinek <jakub@redhat.com>
24913 * config/i386/i386.c (ix86_expand_builtin): Handle
24914 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
24915 ix86_builtins_isa[fcode].isa as a requirement of those
24916 flags and any other flag in the bitmask.
24917 (ix86_init_mmx_sse_builtins): Use 0 instead of
24918 ~OPTION_MASK_ISA_64BIT as mask.
24919 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
24920 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
24921 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
24922 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
24924 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24927 * lra-constraints.c (split_reg): Check requested split mode
24928 is supported by the register.
24930 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24932 * lra-constraints.c (simplify_operand_subreg): Remove early
24935 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24938 * lra-constraints.c (curr_insn_transform): Tighten condition
24939 for converting SUBREG reloads from OP_OUT to OP_INOUT.
24941 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
24944 * lra-constraints.c (curr_insn_transform): Handle
24945 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
24947 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
24950 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
24952 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
24954 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
24957 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
24960 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24962 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
24965 2017-02-18 Jakub Jelinek <jakub@redhat.com>
24967 * final.c (last_columnnum, override_columnnum): New variables.
24968 (final_start_function): Set last_columnnum, pass it to begin_prologue
24969 hook and pass 0 to dwarf2out_begin_prologue.
24970 (final_scan_insn): Update override_columnnum. Pass last_columnnum
24971 to source_line debug hook.
24972 (notice_source_line): Compute last_columnnum and for debug_column_info
24973 return true on column changes.
24974 * debug.h (struct gcc_debug_hooks): Add column argument to
24975 source_line and begin_prologue hooks.
24976 (debug_nothing_int_charstar_int_bool): Remove prototype.
24977 (debug_nothing_int_int_charstar,
24978 debug_nothing_int_int_charstar_int_bool): New prototypes.
24979 (dwarf2out_begin_prologue): Add column argument.
24980 * debug.c (do_nothing_debug_hooks): Adjust source_line and
24981 begin_prologue hooks.
24982 (debug_nothing_int_charstar_int_bool): Remove.
24983 (debug_nothing_int_int_charstar,
24984 debug_nothing_int_int_charstar_int_bool): New functions.
24985 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
24986 through to dwarf2out_source_line.
24987 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
24988 (dwarf2out_source_line): Add column argument, emit it if requested.
24989 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
24991 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24992 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
24993 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
24994 through to dwarf2out_begin_prologue.
24995 (vmsdbgout_source_line): Add column argument, pass it through to
24996 dwarf2out_source_line.
24997 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
24998 dbxout_source_line caller.
24999 (dbxout_source_line): Add column argument.
25001 * common.opt (gno-column-info, gcolumn-info): New options.
25002 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25003 (check_die): Also test for multiple DW_AT_decl_column attributes.
25004 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25005 DW_AT_decl_column if requested.
25006 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25008 (gen_variable_die): Likewise.
25009 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25010 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25013 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25014 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25015 (ix86_handle_option): Handle OPT_m3dnowa.
25016 * doc/invoke.texi (-m3dnowa): Document.
25017 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25018 -m3dnowa instead of -m3dnow -march=athlon.
25021 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25022 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25024 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25027 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25028 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25029 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25030 generator for vsx_xxpermdi_<mode>_be.
25031 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25032 force big-endian semantics.
25033 (vsx_xxpermdi_<mode>_be): New define_expand with same
25034 implementation as previous version of vsx_xxpermdi_<mode>.
25036 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25038 PR tree-optimization/79327
25039 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25040 variable, its initialization and use.
25042 2017-02-17 Julia Koval <julia.koval@intel.com>
25044 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25045 (OPTION_MASK_ISA_PKU_UNSET): New.
25046 (ix86_handle_option): Handle -mrdpid.
25047 * config/i386/cpuid.h (bit_RDPID): New.
25048 * config/i386/driver-i386.c (host_detect_local_cpu):
25049 Detect RDPID feature.
25050 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25051 * config/i386/i386-c.c (ix86_target_macros_internal):
25053 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25054 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25055 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25056 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25057 * config/i386/i386.md (define_insn "rdpid"): New.
25058 * config/i386/i386.opt Add -mrdpid.
25059 * config/i386/immintrin.h (_rdpid_u32): New.
25061 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25063 PR rtl-optimization/79541
25064 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25065 instead of transforming it into USE.
25067 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25069 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25070 If HONOR_SNANS (SFmode) force the input to a register.
25071 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25072 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25073 an frsp or similar insn.
25075 2017-02-17 Martin Liska <mliska@suse.cz>
25077 PR rtl-optimization/79577
25078 * params.def (selsched-max-sched-times): Increase minimum to 1.
25080 2017-02-17 Martin Liska <mliska@suse.cz>
25082 PR rtl-optimization/79574
25083 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25085 2017-02-17 Martin Liska <mliska@suse.cz>
25087 PR tree-optimization/79529
25088 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25089 ssa_defined_default_def_p to handle cases which are implicitly
25091 * tree-ssa.c (ssa_defined_default_def_p): New function.
25092 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25093 which are implicitly defined.
25094 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25096 2017-02-17 Richard Biener <rguenther@suse.de>
25098 PR middle-end/79576
25099 * params.def (max-ssa-name-query-depth): Limit to 10.
25101 2017-02-17 Richard Biener <rguenther@suse.de>
25103 PR tree-optimization/79552
25104 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25107 2017-02-17 Richard Biener <rguenther@suse.de>
25110 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25112 2017-02-17 Marek Polacek <polacek@redhat.com>
25114 PR middle-end/79536
25115 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25116 (fold_negate_expr): New wrapper.
25118 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25120 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25121 Correct terminology and de-emphasize pre-standard behavior.
25123 2017-02-16 Alan Modra <amodra@gmail.com>
25125 PR rtl-optimization/79286
25126 * ira.c (def_dominates_uses): New function.
25127 (update_equiv_regs): Don't create an equivalence for insns that
25128 may trap where the register def does not dominate the use.
25130 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25132 PR rtl-optimization/78127
25133 * lra.c (lra): Call lra_eliminate before finish the loop after
25136 2017-02-16 Richard Biener <rguenther@suse.de>
25138 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25140 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25141 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25142 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25143 (isl_val_int_from_wi): New function.
25144 (extract_affine_gmp): Rename to ...
25145 (extract_affine_wi): ... this, take a widest_int.
25146 (extract_affine_int): Just wrap extract_affine_wi.
25147 (add_param_constraints): Use isl_val_int_from_wi.
25148 (add_loop_constraints): Likewise, and extract_affine_wi.
25150 2017-02-15 Jeff Law <law@redhat.com>
25152 PR middle-end/79521
25153 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25154 ira_init_register_move_cost_if_necessary.
25156 2017-02-15 Martin Sebor <msebor@redhat.com>
25158 PR middle-end/32003
25159 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25160 removed in a prior commit.
25162 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25164 PR tree-optimization/79347
25165 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25166 counters during peeling.
25168 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25170 * Makefile.in (site.exp): Remove "set ISLVER".
25172 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25175 * real.c (real_from_integer): Call real_convert even for decimal.
25177 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25180 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25182 2017-02-14 Andrew Pinski <apinski@cavium.com>
25184 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25185 cores and change the partno/implementer to be correct.
25186 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25187 the 'B" as the implementer.
25188 * config/aarch64/aarch64-tune.md: Regenerate.
25190 2017-02-14 Carl Love <cel@us.ibm.com>
25192 * config/rs6000/rs6000.c: Add case statement entry to make the
25193 xvcvuxdsp built-in argument unsigned.
25194 * config/rs6000/vsx.md: Fix the source and return operand types so they
25195 match the instruction definitions from the ISA document. Fix typo
25196 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25199 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25202 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25203 member early_clobber_alts.
25204 * lra-lives.c (reg_early_clobber_p): New.
25205 (process_bb_lives): Use it.
25206 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25207 (debug_operand_data): Initialize early_clobber_alts.
25208 (setup_operand_alternative): Set up early_clobber_alts.
25209 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25210 alternatives to new_insn_reg.
25211 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25213 (lra_update_insn_regno_info): Pass the new arg.
25215 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25217 PR middle-end/79505
25218 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25219 (new_oacc_loop_raw): Don't clear already cleared fields.
25222 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25223 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25224 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25226 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25229 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25231 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25234 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25235 the extra instruction to the right place to store 128-bit constant
25238 2017-02-14 Martin Sebor <msebor@redhat.com>
25240 PR middle-end/79448
25241 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25242 warning for strings of unknown length.
25244 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25246 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25248 2017-02-14 Jeff Law <law@redhat.com>
25251 * ira-costs.c (scan_one_insn): Initialize register move costs
25252 for pseudos seen in USE/CLOBBER insns.
25254 PR tree-optimization/79095
25255 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25256 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25257 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25258 if the operands are known to be not equal, then the resulting range
25260 (intersect_ranges): If the new range is ~[0,0] and the old range is
25261 wide, then prefer ~[0,0].
25262 * tree-vrp.c (overflow_comparison_p_1): New function.
25263 (overflow_comparison_p): New function.
25264 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25265 if NAME is used in an overflow test.
25266 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25267 overflow check that can be expressed as an equality test, then adjust
25268 ops to be that equality test.
25270 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25272 * config/s390/s390-builtin-types.def: Remove flags argument.
25273 * config/s390/s390.c (s390_init_builtins): Likewise.
25275 2017-02-14 Martin Liska <mliska@suse.cz>
25277 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25278 vector. Fix trailing white spaces.
25280 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25282 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25285 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25287 PR rtl-optimization/68664
25288 * config/arm/arm.c (arm_sched_can_speculate_insn):
25289 New function. Declare prototype.
25290 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25292 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25294 PR rtl-optimization/68664
25295 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25297 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25299 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25301 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25302 max skip bytes for function, loop and jump.
25304 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25306 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25307 ABS_EXPR for gimple dump.
25309 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25312 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25314 PR tree-optimization/79408
25315 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25316 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25317 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25318 also if rhs1 is INTEGER_CST.
25320 2017-02-14 Richard Biener <rguenther@suse.de>
25322 PR middle-end/79432
25323 * tree-into-ssa.c (insert_phi_nodes): When the function can
25324 have abnormal edges rewrite SSA names with broken use-def
25325 dominance out of SSA and register them for PHI insertion.
25327 2017-02-13 Martin Sebor <msebor@redhat.com>
25329 PR middle-end/79496
25330 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25331 clearing info.nowrite flag when snprintf size argument is a range.
25333 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25335 * cprop.c (cprop_jump): Add missing space in string literal.
25336 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25337 (get_constraint_for_component_ref): Likewise.
25338 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25339 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25340 * lra-constraints.c (process_alt_operands): Likewise.
25341 * ipa-inline.c (inline_small_functions): Likewise.
25342 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25343 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25344 * trans-mem.c (diagnose_tm_1_op): Likewise.
25345 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25346 (grid_parallel_clauses_gridifiable): Likewise.
25348 * config/nvptx/mkoffload.c (process): Add space in between
25351 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25352 "MOD4_SSE_REGS" and "ALL_REGS".
25354 * spellcheck.c (test_data): Add , in between "foo" and "food".
25356 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25359 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25360 boundary crossing check and subsequent code generation agree.
25362 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25364 * config/aarch64/aarch64.c (has_memory_op): Delete.
25365 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25368 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25370 PR rtl-optimization/79388
25371 PR rtl-optimization/79450
25372 * combine.c (distribute_notes): When removing TEM_INSN for which
25373 corresponding dest has last value recorded, invalidate that last
25376 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25378 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25379 of explicit '@'. Add missing assembly comment marker on branch costs
25382 2017-02-13 Nathan Sidwell <nathan@acm.org>
25384 * gengtype-lex.l (<in_struct>): Add '/'.
25386 2017-02-13 Martin Liska <mliska@suse.cz>
25389 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25391 2017-02-13 Richard Biener <rguenther@suse.de>
25393 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25395 * configure: Re-generate.
25396 * config.in: Likewise.
25397 * graphite-dependences.c: Simplify as if
25398 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25399 * graphite-isl-ast-to-gimple.c: Likewise.
25400 * graphite-optimize-isl.c: Likewise.
25401 * graphite-poly.c: Likewise.
25402 * graphite-sese-to-poly.c: Likewise.
25403 * graphite.h: Likewise.
25404 * toplev.c: Include isl/version.h and use isl_version () for
25405 printing the ISL version.
25406 * doc/install.texi: Update ISL requirement.
25408 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25410 * doc/standards.texi (Standards): Update reference to
25413 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25415 * doc/extend.texi (Named Address Spaces): sourceware.org now
25417 * doc/install.texi (Binaries): Ditto.
25420 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25422 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25423 standard terminology "stringize"/"stringizing" throughout.
25424 * doc/cppinternals.texi: Likewise.
25426 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25428 * doc/extend.texi: Fix some spelling mistakes and typos.
25429 * doc/invoke.texi: Likewise.
25431 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25434 * params.def (inline-min-speedup) Change from 10 to 8.
25436 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25438 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25441 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25444 * ipa-inline-analysis.c (get_minimal_bb): New function.
25445 (record_modified): Use it.
25446 (remap_edge_change_prob): Handle also ancestor functions.
25448 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25450 * doc/contrib.texi (Contributors): Remove broken link into
25451 the Mauve CVS repository.
25453 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25455 PR middle-end/79454
25456 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25457 result computation whenever lhs doesn't have vector mode, not
25458 just when it has BLKmode.
25460 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25462 * doc/makefile.texi (profiledbootstrap): Refer to the
25463 installation instructions only in textual form.
25465 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25468 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25470 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25472 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25473 (Specific): Update mingw-w64 reference.
25475 (Specific): Remove broken link to Renesas RX processor.
25477 2017-02-10 Richard Biener <rguenther@suse.de>
25479 * toplev.c (process_options): Do not mention obsolete graphite
25480 options when printing sorry message about missing graphite support.
25481 Mention -floop-nest-optimize.
25483 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25485 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25486 (vtst_p16): Likewise.
25487 (vtstq_p8): Likewise.
25488 (vtstq_p16): Likewise.
25490 (vtstq_p64): Likewise.
25491 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25492 (vset_lane_p64): New.
25493 (vsetq_lane_p64): New.
25495 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25497 PR tree-optimization/79411
25498 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25499 stmt operands are SSA_NAMEs used in abnormal phis.
25500 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25503 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25506 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25509 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25511 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25513 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25515 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25519 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25520 not arbitrary TREE_CONSTANT.
25523 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25524 "omp declare target link" attribute unless is_global_var.
25525 * omp-offload.c (find_link_var_op): Likewise.
25527 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25528 Chung-Lin Tang <cltang@codesourcery.com>
25530 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25532 (gimplify_adjust_omp_clauses): Don't delete TILE.
25533 (gimplify_omp_for): Deal with TILE.
25534 * internal-fn.c (expand_GOACC_TILE): New function.
25535 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25537 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25538 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25540 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25541 avoid DIV for outermost collapse var.
25542 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25543 Remove out of date comments, fix whitespace.
25544 * omp-general.c (omp_extract_for_data): Deal with tiling.
25545 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25546 adjust OLF_DIM_BASE value.
25547 (struct omp_for_data): Add tiling field.
25548 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25549 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25550 for auto loops. Remove default auto determining, moved to
25551 oacc_loop_fixed_partitions.
25552 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25553 stmts, add e_mask field.
25554 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25555 (oacc_thread_numbers): Use oacc_dim_call.
25556 (oacc_xform_tile): New.
25557 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25558 (finish_oacc_loop): Adjust for ifns vector.
25559 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25560 add case for GOACC_TILE fns.
25561 (oacc_loop_xform_loop): Delete.
25562 (oacc_loop_process): Iterate over call list directly, and add
25563 handling for GOACC_TILE fns.
25564 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25566 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25567 vector partitioning to outer loops. Assign 2 partitions to loops
25568 when available. Add TILE handling.
25569 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25570 (execite_oacc_device_lower): Process GOACC_TILE fns,
25571 ignore unknown specs.
25572 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25573 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25574 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25576 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25578 * configure.ac (ACX_BUGURL): Update.
25579 * configure: Regenerate.
25581 2017-02-09 Richard Biener <rguenther@suse.de>
25583 PR tree-optimization/69823
25584 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25585 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25587 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25589 * config/arc/arc-c.def: Add __NPS400__ definition.
25590 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25591 (TARGET_NPS400): Define.
25593 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25595 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25597 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25598 pointer, arch_info.
25599 (arc_cpu_types): Fill the arch_info field with a pointer into the
25600 arc_arch_types table.
25601 (arc_selected_cpu): Declare.
25602 * config/arc/arc.c (arc_selected_cpu): Make global.
25603 (arc_selected_arch): Delete.
25604 (arc_base_cpu): Delete.
25605 (arc_override_options): Remove references to deleted variables,
25606 update access to arch information.
25607 (ARC_OPT): Update access to arch information.
25608 (ARC_OPTX): Likewise.
25609 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25610 (TARGET_ARC600): Update access to arch information.
25611 (TARGET_ARC601): Likewise.
25612 (TARGET_ARC700): Likewise.
25613 (TARGET_EM): Likewise.
25614 (TARGET_HS): Likewise.
25615 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25618 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25621 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25622 condition/operands for integer GE/LE/GEU/LEU operations.
25624 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25626 PR translation/79397
25627 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25630 2017-02-08 Martin Jambor <mjambor@suse.cz>
25633 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25634 whether allocation happened.
25635 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25636 nothing was allocated.
25638 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25640 PR tree-optimization/79408
25641 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25642 constant, but SSA_NAME with a known integer range, use the minimum
25643 of that range instead of op1 to determine if modulo can be replaced
25644 with its first operand.
25646 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25648 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25650 2017-02-08 Richard Biener <rguenther@suse.de>
25652 PR tree-optimization/71824
25653 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25654 Check all loops contained in the merged region.
25656 2017-02-07 Andrew Pinski <apinski@cavium.com>
25658 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25660 2017-02-07 Andrew Pinski <apinski@cavium.com>
25662 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25663 (thunderxt88): Likewise.
25664 (thunderxt81): Disable LSE and change v8.1 to v8.
25665 (thunderxt83): Likewise.
25667 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25668 Richard Biener <rguenther@suse.de>
25670 PR middle-end/79399
25671 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25672 type from int to size_t.
25673 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25675 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25677 PR rtl-optimization/79386
25678 * cprop.c (bypass_conditional_jumps): Initialize
25679 bypass_last_basic_block already before splitting bbs after
25680 unconditional traps...
25681 (bypass_conditional_jumps): ... rather than here.
25684 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25685 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25686 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25687 fix -masm=intel patterns.
25689 2017-02-07 Richard Biener <rguenther@suse.de>
25691 PR tree-optimization/79256
25692 PR middle-end/79278
25693 * builtins.c (get_object_alignment_2): Use min_align_of_type
25694 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25695 and ADJUST_FIELD_ALIGN.
25697 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25699 * doc/tm.texi: Regenerate.
25700 * stor-layout.c (layout_decl): Adjust.
25701 (update_alignment_for_field): Likewise.
25702 (place_field): Likewise.
25703 (min_align_of_type): Likewise.
25704 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25705 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25706 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25707 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25708 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25709 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25710 * config/i386/i386.c (x86_field_alignment): Likewise.
25711 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25712 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25713 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25714 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25715 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25716 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25720 2017-01-30 Richard Biener <rguenther@suse.de>
25722 PR tree-optimization/79256
25723 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25724 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25727 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25729 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25730 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25731 builtins to SImode and emit a zero-extend, if necessary.
25733 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25735 * docs/invoke.texi (RISC-V Options): Alphabetize.
25737 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25739 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25742 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25744 * config/riscv/riscv.c: New file.
25745 * common/config/riscv/riscv-common.c: Likewise.
25746 * config.gcc: Likewise.
25747 * config/riscv/constraints.md: Likewise.
25748 * config/riscv/elf.h: Likewise.
25749 * config/riscv/generic.md: Likewise.
25750 * config/riscv/linux.h: Likewise.
25751 * config/riscv/multilib-generator: Likewise.
25752 * config/riscv/peephole.md: Likewise.
25753 * config/riscv/pic.md: Likewise.
25754 * config/riscv/predicates.md: Likewise.
25755 * config/riscv/riscv-builtins.c: Likewise.
25756 * config/riscv/riscv-c.c: Likewise.
25757 * config/riscv/riscv-ftypes.def: Likewise.
25758 * config/riscv/riscv-modes.def: Likewise.
25759 * config/riscv/riscv-opts.h: Likewise.
25760 * config/riscv/riscv-protos.h: Likewise.
25761 * config/riscv/riscv.h: Likewise.
25762 * config/riscv/riscv.md: Likewise.
25763 * config/riscv/riscv.opt: Likewise.
25764 * config/riscv/sync.md: Likewise.
25765 * config/riscv/t-elf-multilib: Likewise.
25766 * config/riscv/t-linux: Likewise.
25767 * config/riscv/t-linux-multilib: Likewise.
25768 * config/riscv/t-riscv: Likewise.
25769 * configure.ac: Likewise.
25770 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25771 Waterman as RISC-V maintainers.
25772 * doc/install.texi: Add RISC-V entries.
25773 * doc/invoke.texi: Add RISC-V options section.
25774 * doc/md.texi: Add RISC-V constraints section.
25775 * configure: Regenerated.
25777 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25780 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25781 false values to be constant vectors with all 0 or all 1 bits set.
25782 (vcondu<mode><mode>): Likewise.
25783 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25785 (fpmask_comparison_operator): Update comment.
25786 (vecint_comparison_operator): New predicate.
25787 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25788 vector conditionals when the true and false values are constant
25789 vectors with all 0 bits or all 1 bits set.
25791 2017-02-06 Martin Sebor <msebor@redhat.com>
25793 PR tree-optimization/79376
25794 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25796 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25798 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25799 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25800 to simplify split condition.
25802 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25804 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25805 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25808 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25810 PR rtl-optimization/68664
25811 * target.def (can_speculate_insn): New hook.
25812 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25813 * doc/tm.texi: Regenerate.
25814 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25815 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25816 (rs6000_sched_can_speculate_insn): New function.
25818 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25820 PR tree-optimization/79284
25821 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25822 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25823 vectorizable_mask_load_store, vectorizable_operation,
25824 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25825 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25826 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25827 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25828 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25829 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25830 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25831 is_gimple_assign (stmt). Replace another such test with
25832 is_gimple_assign (stmt).
25834 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25837 * config/avr/avr.c (rtl-iter.h): Include it.
25838 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25839 (avr_legitimate_combined_insn): ...and implementation.
25841 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
25843 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
25844 * config/s390/s390.c (s390_const_operand_ok)
25845 (s390_canonicalize_comparison, s390_extract_part)
25846 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
25847 (s390_contiguous_bitmask_p, s390_rtx_costs)
25848 (legitimize_pic_address): Likewise.
25849 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
25850 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
25851 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
25852 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
25853 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
25855 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25857 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
25858 REGNO($0) == REGNO($1).
25860 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25862 * config/s390/linux.h(SIZE_TYPE): Add comment.
25864 2017-02-06 Julian Brown <julian@codesourcery.com>
25865 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25866 Virendra Pathak <virendra.pathak@broadcom.com>
25868 * config/aarch64/aarch64-cores.def: Change the scheduler
25870 * config/aarch64/aarch64.md: Include thunderx2t99.md.
25871 * config/aarch64/thunderx2t99.md: New file.
25873 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25875 * doc/standards.texi (Go Language): Update link to language
25878 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25880 * tree-eh.c (lower_resx): Sanitize profile.
25881 (cleanup_empty_eh_move_lp): Likewise.
25883 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
25886 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
25888 * cfgloopmanip.h (loop_version): Update prototype.
25889 * modulo-sched.c (sms_schedule): Update call of loop_version.
25890 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
25891 * tree-parloops.c (gen_parallel_loop): Likewise.
25892 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
25893 * tree-ssa-loop-split.c (split_loop): Likewise.
25894 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
25895 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25897 2017-02-05 Martin Liska <mliska@suse.cz>
25900 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
25902 (print_operand_address): Initialize a struct to zero.
25904 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25906 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
25907 garbage collector only in textual form.
25909 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
25911 * doc/extend.texi (x86 specific memory model extensions for
25912 transactional memory): Simplify a phrase.
25914 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
25917 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
25918 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
25919 (atomic_storedi_1): Likewise.
25921 2017-02-04 Jakub Jelinek <jakub@redhat.com>
25923 PR tree-optimization/79338
25924 * tree-parloops.c (gather_scalar_reductions): Don't call
25925 vect_analyze_loop_form for loop->inner before destroying loop's
25928 2017-02-03 Martin Sebor <msebor@redhat.com>
25930 PR tree-optimization/79327
25931 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
25932 when precision has resulted in leading zeros.
25933 (format_integer): Adjust the likely counter to assume an unknown
25934 argument that may be zero is non-zero.
25936 2017-02-03 Jason Merrill <jason@redhat.com>
25939 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
25940 avoid copying non-taken branch.
25942 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25944 PR tree-optimization/79340
25945 * tree-vect-loop.c (vectorizable_reduction): Release
25946 vec_defs elements after safe_splicing them into other vectors.
25949 PR tree-optimization/79327
25950 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
25951 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
25953 (format_integer): Use wide_int_to_tree instead of build_int_cst
25954 + to_?hwi. If argmin is NULL, just set argmin and argmax to
25955 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
25956 of shortest and longest sequence.
25958 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
25960 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
25961 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
25963 2017-02-03 Walter Lee <walt@tilera.com>
25966 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
25967 after initial stackframe link reg save.
25968 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
25970 2017-02-03 Jakub Jelinek <jakub@redhat.com>
25973 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
25974 wu for stxssp alternative.
25976 2017-02-03 Martin Sebor <msebor@redhat.com>
25978 PR tree-optimization/79352
25979 * gimple-fold.c (get_range_strlen): Add argument.
25980 (get_range_strlen): Change return type to bool.
25981 (get_maxval_strlen): Pass in a dummy argument.
25982 * gimple-fold.h (get_range_strlen): Change return type to bool.
25983 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
25984 * tree.h (array_at_struct_end_p): Add argument.
25985 * tree.c (array_at_struct_end_p): Handle it.
25987 2017-02-03 Martin Liska <mliska@suse.cz>
25990 * multiple_target.c (create_dispatcher_calls): Redirect edge
25991 from a caller of a dispatcher.
25992 (expand_target_clones): Make the clones local.
25993 (ipa_target_clone): Do both target clones and resolvers.
25994 (ipa_dispatcher_calls): Remove the pass.
25995 (pass_dispatcher_calls::gate): Likewise.
25996 (make_pass_dispatcher_calls): Likewise.
25997 * passes.def (pass_target_clone): Put as very first IPA early
26000 2017-02-03 Martin Liska <mliska@suse.cz>
26002 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26003 in case of a function with ifunc attribute.
26005 2017-02-03 Martin Liska <mliska@suse.cz>
26007 * cgraph.c (cgraph_node::dump): Dump function version info.
26008 * symtab.c (symtab_node::dump_base): Add missing new line.
26010 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26012 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26013 (ifcombine_ifandif): Use it.
26015 2017-02-03 Martin Liska <mliska@suse.cz>
26017 * doc/invoke.texi: Document default value for
26018 use-after-scope-direct-emission-threshold.
26020 2017-02-03 Martin Liska <mliska@suse.cz>
26022 PR tree-optimization/79339
26023 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26024 (format_floating): Likewise.
26026 2017-02-03 Martin Liska <mliska@suse.cz>
26029 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26031 (ipa_node_params_t::remove): Likewise.
26032 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26033 initialization from removed ipa_node_params_t::insert.
26034 (ipa_node_params::~ipa_node_params): Move from removed
26035 ipa_node_params_t::release.
26036 * symbol-summary.h (symbol_summary::m_released): New member.
26037 Do not release a summary twice. Do not allow to call finalizer
26038 for types of a summary that live in GGC memory.
26040 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26042 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26045 2017-02-02 Martin Sebor <msebor@redhat.com>
26047 PR middle-end/79275
26048 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26049 (format_string): Tighten up the range of output for non-constant
26050 strings and correct the expected range for wide non-constant strings.
26052 2017-02-02 Martin Sebor <msebor@redhat.com>
26054 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26056 PR middle-end/32003
26057 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26059 (-fdump-tree-@var): Add to index and document how to come up
26060 with pass-specific option and dump file names.
26061 (-fdump-passes): Clarify where to look for output.
26063 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26065 PR middle-end/77445
26066 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26067 statistics of the analyzed path; allow threading for speed when
26068 any of BBs along the path are optimized for speed.
26070 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26072 PR middle-end/78468
26073 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26074 settings of the virtual registers.
26077 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26079 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26080 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26083 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26085 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26086 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26088 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26090 * config/s390/s390.md: Add missing comments with the expanded
26092 * config/s390/vector.md: Likewise.
26093 * config/s390/vx-builtins.md: Likewise.
26095 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26098 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26099 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26100 conditions on a single line.
26102 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26104 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26105 __S390_VX__ to __VX__.
26107 2017-02-01 Andrew Pinski <apinski@cavium.com>
26109 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26110 stmt_info to record_stmt_cost.
26111 (vect_get_known_peeling_cost): Pass stmt_info if known to
26113 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26114 cpu_vector_cost field into
26115 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26116 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26117 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26118 splitting of scalar_stmt_cost and vec_stmt_cost.
26119 (thunderx_vector_cost): Likewise.
26120 (cortexa57_vector_cost): LIkewise.
26121 (exynosm1_vector_cost): Likewise.
26122 (xgene1_vector_cost): Likewise.
26123 (thunderx2t99_vector_cost): Improve after the splitting of the two
26125 (aarch64_builtin_vectorization_cost): Update for the splitting of
26126 scalar_stmt_cost and vec_stmt_cost.
26128 2017-02-01 Torvald Riegel <triegel@redhat.com>
26129 Richard Henderson <rth@redhat.com>
26131 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26132 conditional on existance of a fast atomic load.
26133 * optabs-query.c (can_atomic_load_p): New function.
26134 * optabs-query.h (can_atomic_load_p): Declare it.
26135 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26136 no fast atomic load is available for the particular size of access.
26137 (expand_atomic_compare_and_swap): Likewise.
26138 (expand_atomic_load): Likewise.
26139 (expand_atomic_store): Likewise.
26140 (expand_atomic_fetch_op): Likewise.
26141 * testsuite/lib/target-supports.exp
26142 (check_effective_target_sync_int_128): Remove x86 because it provides
26143 no fast atomic load.
26144 (check_effective_target_sync_int_128_runtime): Likewise.
26146 2017-02-01 Richard Biener <rguenther@suse.de>
26148 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26149 (graphite_transform_loops): Provide opt-info for optimized nests.
26150 * tree-parloop.c (parallelize_loops): Provide opt-info for
26151 parallelized loops.
26153 2017-02-01 Richard Biener <rguenther@suse.de>
26155 PR middle-end/79315
26156 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26157 was not set before.
26159 2017-02-01 Richard Biener <rguenther@suse.de>
26161 PR tree-optimization/71824
26162 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26163 Verify the loops are valid in the merged SESE region.
26164 (scop_detection::can_represent_loop_1): Check analyzing the
26165 evolution of the number of iterations in the region succeeds.
26167 2017-01-31 Ian Lance Taylor <iant@golang.org>
26169 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26170 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26172 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26174 PR preprocessor/79210
26175 * input.c (get_substring_ranges_for_loc): Replace line_width
26176 assertion with error-handling.
26178 2017-01-31 Richard Biener <rguenther@suse.de>
26180 PR tree-optimization/77318
26181 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26182 (create_pw_aff_from_tree): Take loop parameter.
26183 (add_condition_to_pbb): Pass loop of the condition to
26184 create_pw_aff_from_tree.
26186 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26188 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26189 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26191 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26195 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26197 (convert_int_to_float128): Likewise.
26198 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26199 (convert_int_to_float128): Likewise.
26200 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26201 (UNSPEC_IEEE128_CONVERT): Likewise.
26202 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26203 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26204 Use local variables for IBM extended format.
26205 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26206 (fix_trunc<mode>si2_fprs): Likewise.
26207 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26208 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26209 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26210 to know that we can now have integers of all sizes in vector
26212 (fix<uns>_<mode>di2_hw): Likewise.
26213 (float<uns>_<mode>si2_hw): Likewise.
26214 (fix_<mode>si2_hw): Likewise.
26215 (fixuns_<mode>si2_hw): Likewise.
26216 (float<uns>_<mode>di2_hw): Likewise.
26217 (float_<mode>di2_hw): Likewise.
26218 (float_<mode>si2_hw): Likewise.
26219 (floatuns_<mode>di2_hw): Likewise.
26220 (floatuns_<mode>si2_hw): Likewise.
26221 (xscvqp<su>wz_<mode>): Delete, no longer used.
26222 (xscvqp<su>dz_<mode>): Likewise.
26223 (xscv<su>dqp_<mode>): Likewise.
26224 (ieee128_mfvsrd_64bit): Likewise.
26225 (ieee128_mfvsrd_32bit): Likewise.
26226 (ieee128_mfvsrwz): Likewise.
26227 (ieee128_mtvsrw): Likewise.
26228 (ieee128_mtvsrd_64bit): Likewise.
26229 (ieee128_mtvsrd_32bit): Likewise.
26231 2017-01-31 Martin Liska <mliska@suse.cz>
26234 * ipa-prop.c (ipa_free_all_node_params): Call release method
26235 instead of ~sumbol_summary to not to trigger double times
26238 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26240 PR tree-optimization/71691
26241 * bitmap.h (class auto_bitmap): New.
26242 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26243 is_maybe_undefined instead of ssa_undefined_value_p.
26245 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26247 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26248 __S390_ARCH_LEVEL__ to __ARCH__.
26250 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26252 PR tree-optimization/79267
26253 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26254 if should_remove_lhs_p is true.
26256 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26259 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26260 (add_alignment_attribute): New.
26261 (base_type_die): Add alignment attribute.
26262 (subrange_type_die): Likewise.
26263 (modified_type_die): Likewise.
26264 (gen_array_type_die): Likewise.
26265 (gen_descr_array_type_die: Likewise.
26266 (gen_enumeration_type_die): Likewise.
26267 (gen_subprogram_die): Likewise.
26268 (gen_variable_die): Likewise.
26269 (gen_field_die): Likewise.
26270 (gen_ptr_to_mbr_type_die): Likewise.
26271 (gen_struct_or_union_type_die): Likewise.
26272 (gen_subroutine_type_die): Likewise.
26273 (gen_typedef_die): Likewise.
26274 (base_type_cmp): Compare alignment attribute.
26276 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26279 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26280 (setb_unsigned) New pattern for setb with CCUNS.
26281 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26282 subfc./subfe sequence to avoid overflow problems. Generate a
26283 shorter sequence with cmpld/setb for power9.
26284 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26285 for generating subfc. instruction.
26286 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26287 now uses this instruction.
26289 2017-01-30 Ian Lance Taylor <iant@google.com>
26292 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26293 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26295 2017-01-30 Martin Sebor <msebor@redhat.com>
26297 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26298 Move constant to the right of a relational operator.
26299 (get_mpfr_format_length, format_character, format_string): Ditto.
26300 (should_warn_p, maybe_warn): Same.
26302 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26304 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26307 * asan.c (get_translation_unit_decl): Remove function.
26308 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26310 2017-01-30 Martin Liska <mliska@suse.cz>
26312 PR gcov-profile/79259
26313 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26314 -fprofile-generate.
26316 2017-01-30 Martin Liska <mliska@suse.cz>
26319 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26320 Initialize variables with NULL value.
26322 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26325 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26327 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26329 2017-01-30 Richard Biener <rguenther@suse.de>
26331 PR tree-optimization/79276
26332 * tree-vrp.c (process_assert_insertions): Properly adjust common
26333 when removing a duplicate.
26335 * gcc.dg/torture/pr79276.c: New testcase.
26337 2017-01-30 Richard Biener <rguenther@suse.de>
26339 PR tree-optimization/79256
26340 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26341 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26343 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26345 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26348 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26349 ("*r<noxa>sbg_<mode>_sll_bitmask")
26350 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26351 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26352 Use contiguous_bitmask_nowrap_operand.
26354 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26357 * config/rs6000/altivec.h (vec_xl): Revise #define.
26358 (vec_xst): Likewise.
26360 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26362 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26364 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26366 PR rtl-optimization/79194
26367 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26368 traps before call to bypass_conditional_jumps.
26370 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26372 PR tree-optimization/71374
26373 * lra-constraints.c (check_conflict_input_operands): New.
26374 (match_reload): Use it.
26376 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26379 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26380 account to calculate conflict_set.
26382 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26384 PR rtl-optimization/78559
26385 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26386 other_insn in combine.
26388 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26390 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26391 uint16_type_node for BT_UINT16.
26393 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26395 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26396 "RTL Tests" to menu.
26397 (GIMPLE Tests): New node.
26398 (RTL Tests): New node.
26400 2017-01-27 Richard Biener <rguenther@suse.de>
26402 PR tree-optimization/79245
26403 * tree-loop-distribution.c (distribute_loop): Apply cost
26404 modeling also to detected patterns.
26406 2017-01-27 Richard Biener <rguenther@suse.de>
26408 PR tree-optimization/71433
26409 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26410 (compare_assert_loc): New function.
26411 (process_assert_insertions): Sort and optimize assert locations
26412 to remove duplicates and push down identical assertions on
26413 edges to their destination block.
26415 2017-01-27 Richard Biener <rguenther@suse.de>
26417 PR tree-optimization/79244
26418 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26419 out SSA names even if abnormal.
26421 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26423 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26424 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26425 instead of MPFR_RNDN.
26427 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26430 * arm.c (arm_option_override): Don't call build_target_option_node
26431 until after doing all option overrides.
26432 (arm_valid_target_attribute_tree): Likewise.
26434 2017-01-27 Martin Liska <mliska@suse.cz>
26436 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26437 for {cd}tors and C++ {cd}tors.
26439 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26441 * config/s390/s390.md ("*setmem_long_and")
26442 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26444 2017-01-26 Martin Sebor <msebor@redhat.com>
26446 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26449 2017-01-26 Martin Sebor <msebor@redhat.com>
26451 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26452 HAVE_DFmode before using XFmode or DFmode.
26453 (parse_directive): Avoid using the z length modifier to avoid
26454 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26456 PR middle-end/78703
26457 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26458 to accept adjustment as an array.
26459 (get_int_range): New function.
26460 (struct directive): Make width and prec arrays.
26461 (directive::set_width, directive::set_precision): Call get_int_range.
26462 (format_integer, format_floating): Handle width and precision ranges.
26463 (format_string, parse_directive): Same.
26465 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26468 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26469 comdat_type_p set, just clone them, but keep the children in the
26473 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26474 which have direct callers with -fvar-tracking-assignments enabled
26476 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26477 inside of type units.
26479 2017-01-26 Martin Sebor <msebor@redhat.com>
26481 PR middle-end/78703
26482 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26484 (struct format_result): Replace number_chars, number_chars_min,
26485 and number_chars_max with a single member of struct result_range.
26487 (format_result::operator+=): Adjust.
26488 (struct fmtresult): Remove bounded. Handle likely and unlikely
26490 (fmtresult::adjust_for_width_or_precision): New function.
26491 (fmtresult:type_max_digits): New function.
26492 (bytes_remaining): Handle likely and unlikely counters.
26493 (min_bytes_remaining): Remove.
26494 (format_percent): Simplify.
26495 (format_integer, format_floating): Set likely and unlikely counters.
26496 (get_string_length, format_character, format_string): Same.
26497 (format_plain, should_warn_p): New function.
26498 (maybe_warn): Call should_warn_p. Update diagnostic messages
26499 and handle those for all directives, including plain strings.
26500 (format_directive): Handle likely and unlikely counters.
26501 Remove unnecessary quoting from diagnostics. Add an informational
26503 (add_bytes): Remove.
26504 (pass_sprintf_length::compute_format_length): Simplify.
26505 (try_substitute_return_value): Handle likely and unlikely counters.
26507 2017-01-26 Carl Love <cel@us.ibm.com>
26509 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26510 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26512 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26515 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26516 endianess for subregs into account.
26517 * lra-constraints.c (lra_constraints): Do risky transformations
26518 always on the first iteration.
26519 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26520 last_call_used_reg_set.
26521 (process_bb_lives): Define and use last_call_used_reg_set.
26522 * lra.c (lra): Always continue after lra_constraints on the first
26525 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26527 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26529 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26531 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26533 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26534 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26535 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26536 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26537 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26538 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26539 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26540 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26541 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26543 2017-01-26 Marek Polacek <polacek@redhat.com>
26546 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26547 for the third operand.
26549 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26551 PR middle-end/79236
26552 * omp-low.c (struct omp_context): Add simt_stmt field.
26553 (scan_omp_for): Return omp_context *.
26554 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26555 context to the _simt_ SIMD stmt.
26556 (lower_omp_for): For combined SIMD with sibling _simt_
26557 SIMD, make sure to use the same decls in _looptemp_
26558 clauses as in the sibling.
26560 2017-01-26 David Sherwood <david.sherwood@arm.com>
26562 PR middle-end/79212
26563 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26566 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26569 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26570 emit fld b; fld a; if possible.
26572 * brig-builtins.def: Update copyright years.
26573 * config/arm/arm_acle_builtins.def: Update copyright years.
26575 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26578 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26579 constraint instead of o for the stxsd instruction.
26581 2017-01-25 Carl Love <cel@us.ibm.com>
26583 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26584 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26586 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26588 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26590 2017-01-25 Richard Biener <rguenther@suse.de>
26592 PR tree-optimization/69264
26593 * target.def (vector_alignment_reachable): Improve documentation.
26594 * doc/tm.texi: Regenerate.
26595 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26597 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26598 earlier changes with respect to TYPE_USER_ALIGN.
26599 (vector_alignment_reachable_p): Likewise. Improve dumping.
26601 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26604 * config/arm/arm.md (xordi3): Force constant operand into a register
26607 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26609 * doc/invoke.texi (-fstore-merging): Correct default optimization
26610 levels at which it is enabled.
26611 (-O): Move -fstore-merging from list to...
26614 2017-01-25 Richard Biener <rguenther@suse.de>
26617 * omp-expand.c: Include debug.h.
26618 (expand_omp_taskreg): Make sure to generate early debug before
26619 outlining anything from a function.
26620 (expand_omp_target): Likewise.
26621 (grid_expand_target_grid_body): Likewise.
26623 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26626 * asan.c (get_translation_unit_decl): New function.
26627 (asan_add_global): Extract modules file name from globals
26628 TRANSLATION_UNIT_DECL name.
26630 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26633 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26634 for long calls with APCS frame and VFP.
26636 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26638 * cfg.c (original_copy_tables_initialized_p): New function.
26639 * cfg.h (original_copy_tables_initialized_p): New decl.
26640 * cfgrtl.c (relink_block_chain): Guard the call to
26641 free_original_copy_tables with a call to
26642 original_copy_tables_initialized_p.
26643 * cgraph.h (symtab_node::native_rtl_p): New decl.
26644 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26645 (symtab_node::needed_p): Don't assert for early assembly output
26646 for __RTL functions.
26647 (cgraph_node::finalize_function): Set "force_output" for __RTL
26649 (cgraph_node::analyze): Bail out early for __RTL functions.
26650 (analyze_functions): Update assertion to support __RTL functions.
26651 (cgraph_node::expand): Bail out early for __RTL functions.
26652 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26654 * function.h (struct function): Update comment for field
26656 * gimple-expr.c: Include "tree-pass.h".
26657 (gimple_has_body_p): Return false for __RTL functions.
26658 * Makefile.in (OBJS): Add run-rtl-passes.o.
26659 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26661 (gcc::pass_manager::get_clean_slate): New accessor.
26662 * passes.c: Include "insn-addr.h".
26663 (should_skip_pass_p): Add logging. Update logic for running
26664 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26665 property-provider override so it is only done for gimple passes.
26667 (skip_pass): New function.
26668 (execute_one_pass): Call skip_pass when skipping passes.
26669 * read-md.c (md_reader::read_char): Support filtering
26670 the input to a subset of line numbers.
26671 (md_reader::md_reader): Initialize fields
26672 m_first_line and m_last_line.
26673 (md_reader::read_file_fragment): New function.
26674 * read-md.h (md_reader::read_file_fragment): New decl.
26675 (md_reader::m_first_line): New field.
26676 (md_reader::m_last_line): New field.
26677 * read-rtl-function.c (function_reader::create_function): Only
26678 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26679 curr_properties. Set DECL_INITIAL to a dummy block.
26680 (read_rtl_function_body_from_file_range): New function.
26681 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26683 * run-rtl-passes.c: New file.
26684 * run-rtl-passes.h: New file.
26686 2017-01-24 Jeff Law <law@redhat.com>
26688 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26691 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26693 PR tree-optimization/79159
26694 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26695 (record_nonwrapping_iv): Improve boundary using above function if no
26696 value range information.
26698 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26699 Martin Jambor <mjambor@suse.cz>
26701 * brig-builtins.def: New file.
26702 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26703 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26704 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26705 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26706 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26707 * builtin-types.def (BT_INT8): New.
26708 (BT_INT16): Likewise.
26709 (BT_UINT8): Likewise.
26710 (BT_UINT16): Likewise.
26711 (BT_FN_ULONG): Likewise.
26712 (BT_FN_UINT_INT): Likewise.
26713 (BT_FN_UINT_ULONG): Likewise.
26714 (BT_FN_UINT_LONG): Likewise.
26715 (BT_FN_UINT_PTR): Likewise.
26716 (BT_FN_ULONG_PTR): Likewise.
26717 (BT_FN_INT8_FLOAT): Likewise.
26718 (BT_FN_INT16_FLOAT): Likewise.
26719 (BT_FN_UINT32_FLOAT): Likewise.
26720 (BT_FN_UINT16_FLOAT): Likewise.
26721 (BT_FN_UINT8_FLOAT): Likewise.
26722 (BT_FN_UINT64_FLOAT): Likewise.
26723 (BT_FN_UINT16_UINT32): Likewise.
26724 (BT_FN_UINT32_UINT16): Likewise.
26725 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26726 (BT_FN_INT_PTR_INT): Likewise.
26727 (BT_FN_UINT_PTR_UINT): Likewise.
26728 (BT_FN_LONG_PTR_LONG): Likewise.
26729 (BT_FN_ULONG_PTR_ULONG): Likewise.
26730 (BT_FN_VOID_UINT64_UINT64): Likewise.
26731 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26732 (BT_FN_INT8_INT8_INT8): Likewise.
26733 (BT_FN_INT16_INT16_INT16): Likewise.
26734 (BT_FN_INT_INT_INT): Likewise.
26735 (BT_FN_UINT_FLOAT_UINT): Likewise.
26736 (BT_FN_FLOAT_UINT_UINT): Likewise.
26737 (BT_FN_ULONG_UINT_UINT): Likewise.
26738 (BT_FN_ULONG_UINT_PTR): Likewise.
26739 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26740 (BT_FN_UINT_UINT_UINT): Likewise.
26741 (BT_FN_VOID_UINT_PTR): Likewise.
26742 (BT_FN_UINT_UINT_PTR: Likewise.
26743 (BT_FN_UINT32_UINT64_PTR): Likewise.
26744 (BT_FN_INT_INT_UINT_UINT): Likewise.
26745 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26746 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26747 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26748 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26749 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26750 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26751 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26752 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26753 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26754 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26755 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26756 * doc/frontends.texi: List BRIG FE.
26757 * doc/install.texi (Testing): Add BRIG tesring requirements.
26758 * doc/invoke.texi (Overall Options): Mention BRIG.
26759 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26761 2017-01-24 Richard Biener <rguenther@suse.de>
26763 PR translation/79208
26764 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26766 2017-01-24 Martin Jambor <mjambor@suse.cz>
26769 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26770 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26771 and known_contexts.
26773 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26775 PR middle-end/79123
26776 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26777 casts from signed to unsigned really don't have a range.
26779 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26781 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26782 GMP_RNDx for compatiblity.
26784 2017-01-24 Martin Liska <mliska@suse.cz>
26787 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26788 that would prevent us to call alloca with -1 as argument.
26790 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26792 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26793 Avoid -Wformat-security warning.
26795 2017-01-23 Andrew Pinski <apinski@cavium.com>
26797 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26800 2017-01-23 Martin Sebor <msebor@redhat.com>
26802 PR middle-end/78703
26803 * gimple-ssa-sprintf.c (warn_level): New global.
26804 (format_integer): Use it here and throughout the rest of the file.
26805 Use the same switch to compute sign as base.
26806 (maybe_warn): New function.
26807 (format_directive): Factor out warnings into maybe_warn.
26808 Add debugging output. Use warn_level.
26809 (add_bytes): Use warn_level.
26810 (pass_sprintf_length::compute_format_length): Add debugging output.
26811 (try_substitute_return_value): Same.
26812 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26814 PR middle-end/78703
26815 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26816 (struct fmtresult, format_integer, format_floating): Adjust.
26817 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26818 (get_string_length, format_string,format_directive): Same.
26819 (pass_sprintf_length::compute_format_length): Same.
26820 (try_substitute_return_value): Simplify slightly.
26822 PR middle-end/78703
26823 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26824 (fmtresult::operator+=): Outlined.
26825 (struct fmtresult): Add ctors.
26826 (struct conversion_spec): Rename...
26827 (struct directive): ...to this. Add and remove data members.
26828 (directive::set_width, directive::set_precision): New functions.
26829 (format_percent): Use fmtresult ctor.
26830 (get_width_and_precision): Remove.
26831 (format_integer): Make naming changes. Avoid computing width and
26833 (format_floating): Same. Adjust indentation.
26834 (format_character, format_none): New functions.
26835 (format_string): Moved character handling to format_character.
26836 (format_directive): Remove arguments, change return type.
26837 (parse_directive): New function.
26838 (pass_sprintf_length::compute_format_length): Move directive
26839 parsing to parse_directive.
26841 2017-01-23 Jakub Jelinek <jakub@redhat.com>
26843 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
26844 (assign_assembler_name_if_needed): ... this.
26845 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
26846 (assign_assembler_name_if_needed): ... this.
26847 (free_lang_data_in_cgraph): Adjust callers.
26848 * cgraphunit.c (cgraph_node::analyze): Likewise.
26849 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
26852 2017-01-23 Richard Biener <rguenther@suse.de>
26854 PR tree-optimization/79088
26855 PR tree-optimization/79188
26856 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
26857 resetting loop bounds after last path deletion. Reset loop
26858 bounds of the target loop, make code match the comments.
26859 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
26860 Make sure loops need no fixups.
26862 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26864 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
26865 exponent support with double type for first argument.
26866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
26867 type returned by __builtin_vec_extract_sig,
26868 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
26869 functions from "vector int" to "vector unsigned int" or from
26870 "vector long long int" to "vector unsigned long long int".
26871 Changed type returned by __builtin_vec_extract_exp,
26872 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
26873 functions from "vector int" to "vector unsigned int" or from
26874 "vector long long int" to "vector unsigned long long int".
26875 Changed return type of __builtin_vec_test_data_class,
26876 __builtin_vec_test_data_class_sp, and
26877 __builtin_vec_test_data_class_dp from "vector int" to
26878 "vector bool int" or from "vector long long int" to "vector bool
26879 long long int" and changed second argument type from "unsigned
26880 int" to "int". Added new overloaded function forms "vector float
26881 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
26882 "vector float __builtin_vec_insert_exp_sp (vector float, vector
26883 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
26884 double, vector unsigned long long int)" and "vector double
26885 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
26886 long int)". Changed return type of
26887 __builtin_scalar_test_data_class and
26888 __builtin_scalar_test_data_class_sp and
26889 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
26890 int" and changed second argument from "unsigned int" to "int".
26891 Changed type returned by __builtin_scalar_test_neg,
26892 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
26893 from "int" to "bool int". Added new overloaded function form
26894 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
26895 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
26896 exponent double-precision with floating point first argument.
26897 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
26898 documentation of scalar_test_data_class, scalar_test_neg,
26899 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
26900 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
26901 vec_test_data_class built-in functions to reflect refinements in
26902 their type signatures.
26904 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
26906 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
26908 (aarch64_elf_asm_destructor): Likewise.
26910 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
26912 PR rtl-optimization/78634
26913 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
26914 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
26915 * ifcvt.c (noce_try_cmove): Add missing cost check.
26917 PR rtl-optimization/71724
26918 * combine.c (if_then_else_cond): Look for situations where it is
26919 beneficial to undo the work of one of the recursive calls.
26921 2017-01-23 Bin Cheng <bin.cheng@arm.com>
26923 PR tree-optimization/70754
26924 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
26925 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
26926 combined stmt before it if not NULL.
26927 (combine_chains): Process refs reversely and compute dominance point
26930 2017-01-23 Martin Liska <mliska@suse.cz>
26932 PR tree-optimization/79196
26933 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
26934 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
26936 (strlen_optimize_stmt): Call the renamed function.
26938 2017-01-23 Michael Matz <matz@suse.de>
26940 PR tree-optimization/78384
26941 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
26943 2017-01-23 Richard Biener <rguenther@suse.de>
26945 PR tree-optimization/79186
26946 * tree-vrp.c (register_new_assert_for): Make sure we've seen
26947 both incoming edges before moving an assert.
26949 2017-01-23 Martin Jambor <mjambor@suse.cz>
26951 * ipa-prop.c (load_from_param_1): Removed.
26952 (load_from_unmodified_param): Bits from load_from_param_1 put back
26954 (load_from_param): Removed.
26955 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
26956 with stmt. Reverted back to use of load_from_unmodified_param.
26958 2017-01-23 Martin Jambor <mjambor@suse.cz>
26961 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
26962 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
26963 field a pointer to garbage collected vector, mark lattices and
26964 ipcp_orig_node with GTY((skip)).
26965 (ipa_get_param_count): Adjust to descriptors being a pointer.
26966 (ipa_get_param): Likewise.
26967 (ipa_get_type): Likewise.
26968 (ipa_get_param_move_cost): Likewise.
26969 (ipa_set_param_used): Likewise.
26970 (ipa_get_controlled_uses): Likewise.
26971 (ipa_set_controlled_uses): Likewise.
26972 (ipa_is_param_used): Likewise.
26973 (ipa_node_params_t): Move into garbage collector. New methods insert
26975 (ipa_node_params_sum): Annotate wth GTY(()).
26976 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
26978 (ipa_load_from_parm_agg): Adjust declaration.
26979 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
26980 * ipa-profile.c (ipa_profile): Likewise.
26981 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
26982 (ipa_populate_param_decls): Make descriptors parameter garbage
26984 (ipa_dump_param): Adjust to descriptors being a pointer.
26985 (ipa_alloc_node_params): Likewise.
26986 (ipa_initialize_node_params): Likewise.
26987 (load_from_param_1): Make descriptors parameter garbage collected.
26988 (load_from_unmodified_param): Likewise.
26989 (load_from_param): Likewise.
26990 (ipa_load_from_parm_agg): Likewise.
26991 (ipa_node_params::~ipa_node_params): Removed.
26992 (ipa_free_all_node_params): Remove call to delete operator.
26993 (ipa_node_params_t::insert): New.
26994 (ipa_node_params_t::remove): Likewise.
26995 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
26996 copy known_csts and known_contexts vectors.
26997 (ipa_read_node_info): Adjust to descriptors being a pointer.
26998 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27000 (ipcp_transform_function): Make descriptors variable garbage
27003 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27005 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27006 * config/i386/avx512dqintrin.h: Ditto.
27007 * config/i386/avx512fintrin.h: Ditto.
27008 * config/i386/i386.c: Handle new builtins.
27009 * config/i386/i386-builtin.def: Add new builtins.
27010 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27011 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27013 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27014 Martin Liska <mliska@suse.cz>
27016 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27017 * asan.c (asan_expand_poison_ifn): Support stores and use
27018 appropriate ASAN report function.
27019 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27020 * internal-fn.def (ASAN_POISON_USE): Declare.
27021 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27022 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27023 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27024 ASAN_POISON calls w/o LHS.
27025 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27026 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27027 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27028 * gimplify.c (asan_poison_variables): Add attribute
27029 use_after_scope_memory to variables that really needs to live
27031 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27032 having the attribute.
27034 2017-01-23 Martin Liska <mliska@suse.cz>
27036 * asan.c (create_asan_shadow_var): New function.
27037 (asan_expand_poison_ifn): Likewise.
27038 * asan.h (asan_expand_poison_ifn): New declaration.
27039 * internal-fn.c (expand_ASAN_POISON): Likewise.
27040 * internal-fn.def (ASAN_POISON): New builtin.
27041 * sanopt.c (pass_sanopt::execute): Expand
27042 asan_expand_poison_ifn.
27043 * tree-inline.c (copy_decl_for_dup_finish): Make function
27045 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27046 * tree-ssa.c (is_asan_mark_p): New function.
27047 (execute_update_addresses_taken): Rewrite local variables
27048 (identified just by use-after-scope as addressable) into SSA.
27050 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27052 * doc/install.texi (Specific): opensource.apple.com uses https
27053 now. Remove trailing slash.
27055 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27057 * README.Portability: Remove note on an Irix compatibility issue.
27059 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27061 * gcov.c (INCLUDE_ALGORITHM): Define.
27062 (INCLUDE_VECTOR): Define.
27063 No longer include <vector> and <algorithm> directly.
27065 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27067 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27069 * doc/invoke.texi (Code Gen Options): Ditto.
27071 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27074 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27076 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27078 rtl-optimization/79125
27079 * cprop.c (local_cprop_pass): Handle cases where we make an
27080 unconditional trap.
27082 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27086 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27087 read from, for big endian.
27089 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27091 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27092 register pauth builtins for LP64 only.
27094 2017-01-20 Marek Polacek <polacek@redhat.com>
27097 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27100 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27102 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27104 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27105 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27106 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27108 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27111 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27112 in big-endian mode when they are not a single duplicated value.
27114 2017-01-20 Richard Biener <rguenther@suse.de>
27116 * BASE-VER: Bump to 7.0.1.
27118 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27120 * omp-low.c (omplow_simd_context): New struct. Use it...
27121 (lower_rec_simd_input_clauses): ...here and...
27122 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27123 references to idx, lane, max_vf, is_simt.
27125 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27127 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27130 2017-01-20 Martin Jambor <mjambor@suse.cz>
27132 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27133 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27135 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27136 (GTFILES): Rename hsa.c to hsa-common.c.
27137 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27138 * hsa-dump.c: Likewise.
27139 * hsa-gen.c: Likewise.
27140 * hsa-regalloc.c: Likewise.
27141 * ipa-hsa.c: Likewise.
27142 * omp-expand.c: Likewise.
27143 * omp-low.c: Likewise.
27144 * toplev.c: Likewise.
27146 2017-01-20 Marek Polacek <polacek@redhat.com>
27149 * doc/invoke.texi: Document -Wduplicated-branches.
27150 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27151 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27152 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27153 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27154 return 0 only when not OEP_LEXICOGRAPHIC.
27155 (fold_build_cleanup_point_expr): Use the expression
27156 location when building CLEANUP_POINT_EXPR.
27157 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27158 * tree.c (add_expr): Handle error_mark_node.
27160 2017-01-20 Martin Liska <mliska@suse.cz>
27163 * tree-profile.c (init_ic_make_global_vars): Do not call
27165 (gimple_init_gcov_profiler): Likewise.
27167 2017-01-20 Martin Liska <mliska@suse.cz>
27170 * cgraph.h (maybe_create_reference): Remove argument and
27172 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27174 * ipa-cp.c (create_specialized_node): Likewise.
27175 * symtab.c (symtab_node::maybe_create_reference): Handle
27176 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27178 2017-01-20 Martin Liska <mliska@suse.cz>
27180 * read-rtl-function.c (function_reader::create_function): Use
27181 build_decl instread of build_decl_stat.
27183 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27185 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27186 * config/i386/avx512dqintrin.h: Ditto.
27187 * config/i386/avx512fintrin.h: Ditto.
27188 * config/i386/i386-builtin-types.def: Add new types.
27189 * config/i386/i386.c: Handle new types.
27190 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27191 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27192 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27193 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27194 (__builtin_ia32_kshiftridi): New.
27195 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27197 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27201 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27202 define to rs6000_init_stack_protect_guard.
27203 (rs6000_init_stack_protect_guard): New function.
27205 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27206 Yunqiang Su <yunqiang.su@imgtec.com>
27208 * config.gcc (supported_defaults): Add madd4.
27209 (with_madd4): Add validation.
27210 (all_defaults): Add madd4.
27211 * config/mips/mips.opt (mmadd4): New option.
27212 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27214 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27216 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27217 (ISA_HAS_FUSED_MADD4): Likewise.
27218 * doc/invoke.texi (-mmadd4): Document the new option.
27219 * doc/install.texi (--with-madd4): Document the new option.
27221 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27223 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27224 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27225 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27226 (aarch64_init_pauth_hint_builtins): New.
27227 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27228 (aarch64_expand_builtin): Expand new builtins.
27230 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27232 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27233 * combine-stack-adj.c (no_unhandled_cfa): Handle
27234 REG_CFA_TOGGLE_RA_MANGLE.
27235 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27236 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27237 info for return address signing.
27238 (aarch64_expand_epilogue): Likewise.
27240 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27242 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27243 * config/aarch64/aarch64-protos.h
27244 (aarch64_return_address_signing_enabled): New declaration.
27245 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27247 (aarch64_expand_prologue): Sign return address before it's pushed onto
27249 (aarch64_expand_epilogue): Authenticate return address fetched from
27251 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27252 (aarch64_attributes): New function attributes for "sign-return-address".
27253 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27254 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27255 ("*do_return"): Generate combined instructions according to key index.
27256 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27257 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27259 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27260 * config/aarch64/aarch64.opt (msign-return-address=): New.
27261 * doc/extend.texi (AArch64 Function Attributes): Documents
27262 "sign-return-address=".
27263 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27265 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27267 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27268 overall option summary.
27270 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27272 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27273 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27274 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27275 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27277 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27279 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27280 -mpower9-minmax by default for -mcpu=power9.
27281 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27282 128-bit floating point.
27284 2017-01-20 Alan Modra <amodra@gmail.com>
27286 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27287 optimizing for size.
27289 2017-01-20 Alan Modra <amodra@gmail.com>
27292 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27293 for strcmp and strncmp from corresponding builtin decl.
27295 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27297 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27298 instead of i386/rtems-64.h.
27299 * config/i386/rtems-64.h: Remove.
27301 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27305 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27307 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27309 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27311 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27312 Change int to HOST_WIDE_INT.
27313 * config/aarch64/aarch64-protos.h
27314 (aarch64_simd_gen_const_vector_dup): Likewise.
27315 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27317 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27319 * langhooks-def.h (lhd_type_for_size): New decl.
27320 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27321 * langhooks.c (lhd_type_for_size): New function, taken from
27324 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27326 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27327 define_bypass for CR latency.
27328 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27329 (power9-alu2): Add define_bypass for CR latency.
27331 (power9-mul): Update insn latency.
27332 (power9-mul-compare): Update insn latency, bypass latency and remove
27335 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27337 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27339 * config/aarch64/aarch64.md
27340 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27341 aarch64_nopcrelative_literal_loads.
27342 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27344 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27346 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27347 TARGET_LOONGSON_3A.
27348 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27350 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27353 * config.gcc (supported_defaults): Add lxc1-sxc1.
27354 (with_lxc1_sxc1): Add validation.
27355 (all_defaults): Add lxc1-sxc1.
27356 * config/mips/mips.opt (mlxc1-sxc1): New option.
27357 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27359 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27360 __mips_no_lxc1_sxc1.
27361 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27362 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27363 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27365 2017-01-19 Richard Biener <rguenther@suse.de>
27367 PR tree-optimization/72488
27368 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27369 sure to restore SSA info.
27370 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27372 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27374 PR rtl-optimization/79121
27375 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27376 of the inner type when shifting an extended value.
27378 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27381 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27382 interposable aliases.
27384 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27387 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27388 Use the evmergelohi instruction.
27389 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27390 (mov_sitf_e500_subreg8_2_be): Likewise.
27391 (mov_sitf_e500_subreg12_2_le): Likewise.
27392 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27393 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27394 (mov_sitf_e500_subreg8_2_le): Likewise.
27395 (mov_sitf_e500_subreg12_2_be): Likewise.
27397 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27399 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27400 attribute from vecsimple to vecperm.
27401 (altivec_vbpermq2): Likewise.
27403 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27406 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27408 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27409 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27410 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27411 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27412 case where N arg is SIZE_MAX.
27413 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27414 (cmpstrsi): Add pattern.
27416 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27419 __builtin_vec_revb builtins.
27420 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27421 built-in functions to support generation of the ISA 3.0 XXBR<x>
27422 vector byte reverse instructions.
27423 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27424 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27425 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27426 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27427 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27428 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27429 (P9V_BUILTIN_VEC_REVB): Likewise.
27430 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27431 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27432 (p9_xxbrq_v16qi): Likewise.
27433 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27434 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27435 (p9_xxbrh_v8hi): Likewise.
27436 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27437 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27438 vec_revb built-in functions.
27440 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27442 PR rtl-optimization/78952
27443 * config/i386/i386.md (any_extract): New code iterator.
27444 (*insvqi_2): Use any_extract for source operand.
27445 (*insvqi_3): Use any_shiftrt for source operand.
27447 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27449 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27451 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27453 2017-01-18 Matthias Klose <doko@ubuntu.com>
27455 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27457 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27459 * config/rs6000/altivec.h (vec_bperm): Change #define.
27460 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27461 (altivec_vbpermq2): New define_insn.
27462 (altivec_vbpermd): Likewise.
27463 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27464 function interface.
27465 (VBPERMD): Likewise.
27466 (VBPERM): New polymorphic function interface.
27467 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27468 Add entries for P9V_BUILTIN_VEC_VBPERM.
27469 * doc/extend.texi: Add interfaces for vec_bperm.
27471 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27473 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27474 first letter of error messages.
27475 (s390_resolve_overloaded_builtin): Likewise.
27476 * config/s390/s390.c (s390_expand_builtin): Likewise.
27477 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27478 (s390_valid_target_attribute_inner_p): Likewise.
27479 * config/s390/s390.md ("tabort"): Likewise.
27481 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27483 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27484 (ISA_AVOID_DIV_HILO): New macro.
27485 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27486 (ISA_HAS_DDIV): Likewise.
27488 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27490 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27492 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27494 * doc/invoke.texi (fabi-version): Spelling fix.
27496 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27499 * doc/invoke.texi (fabi-version): Mention mangling fix for
27502 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27505 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27507 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27510 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27513 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27515 (stack_protect_test): Ditto.
27516 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27517 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27519 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27520 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27521 -mstack-protector-guard-offset=.
27522 (RS/6000 and PowerPC Options): Ditto.
27524 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27526 * config/i386/i386.h (MASK_CLASS_P): New define.
27527 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27528 there are no registers from different register sets also when
27529 mask registers are used. Update function comment.
27530 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27531 to (*k/*r) and (*k/*km) alternatives.
27533 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27535 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27536 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27537 (EH_RETURN_HANDLER_RTX): New define.
27538 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27539 Force frame pointer in EH return functions.
27540 (aarch64_expand_epilogue): Add barrier for eh_return.
27541 (aarch64_final_eh_return_addr): Remove.
27542 (aarch64_eh_return_handler_rtx): New function.
27543 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27545 (aarch64_eh_return_handler_rtx): New prototype.
27547 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27549 * config/rs6000/altivec.h (vec_rlmi): New #define.
27550 (vec_vrlnm): Likewise.
27551 (vec_rlnm): Likewise.
27552 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27553 (UNSPEC_VRLNM): Likewise.
27554 (VIlong): New mode iterator.
27555 (altivec_vrl<VI_char>mi): New define_insn.
27556 (altivec_vrl<VI_char>nm): Likewise.
27557 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27559 (VRLDNM): Likewise.
27560 (RLNM): New polymorphic function entry.
27561 (VRLWMI): New monomorphic function entry.
27562 (VRLDMI): Likewise.
27563 (RLMI): New polymorphic function entry.
27564 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27565 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27566 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27569 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27572 * dwarf2out.c (field_byte_offset): Restore the
27573 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27574 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27575 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27578 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27581 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27583 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27586 * dwarf2out.c (add_data_member_location_attribute): For constant
27587 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27588 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27589 DW_AT_byte_size attributes.
27591 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27593 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27594 after forcing to constant memory when the code model is medium.
27596 2017-01-17 Julia Koval <julia.koval@intel.com>
27599 * config/i386/avx512fintrin.h
27600 (_mm512_i32gather_ps): Change __addr type to void const*.
27601 (_mm512_mask_i32gather_ps): Ditto.
27602 (_mm512_i32gather_pd): Ditto.
27603 (_mm512_mask_i32gather_pd): Ditto.
27604 (_mm512_i64gather_ps): Ditto.
27605 (_mm512_mask_i64gather_ps): Ditto.
27606 (_mm512_i64gather_pd): Ditto.
27607 (_mm512_mask_i64gather_pd): Ditto.
27608 (_mm512_i32gather_epi32): Ditto.
27609 (_mm512_mask_i32gather_epi32): Ditto.
27610 (_mm512_i32gather_epi64): Ditto.
27611 (_mm512_mask_i32gather_epi64): Ditto.
27612 (_mm512_i64gather_epi32): Ditto.
27613 (_mm512_mask_i64gather_epi32): Ditto.
27614 (_mm512_i64gather_epi64): Ditto.
27615 (_mm512_mask_i64gather_epi64): Ditto.
27616 (_mm512_i32scatter_ps): Change __addr type to void*.
27617 (_mm512_mask_i32scatter_ps): Ditto.
27618 (_mm512_i32scatter_pd): Ditto.
27619 (_mm512_mask_i32scatter_pd): Ditto.
27620 (_mm512_i64scatter_ps): Ditto.
27621 (_mm512_mask_i64scatter_ps): Ditto.
27622 (_mm512_i64scatter_pd): Ditto.
27623 (_mm512_mask_i64scatter_pd): Ditto.
27624 (_mm512_i32scatter_epi32): Ditto.
27625 (_mm512_mask_i32scatter_epi32): Ditto.
27626 (_mm512_i32scatter_epi64): Ditto.
27627 (_mm512_mask_i32scatter_epi64): Ditto.
27628 (_mm512_i64scatter_epi32): Ditto.
27629 (_mm512_mask_i64scatter_epi32): Ditto.
27630 (_mm512_i64scatter_epi64): Ditto.
27631 (_mm512_mask_i64scatter_epi64): Ditto.
27632 * config/i386/avx512pfintrin.h
27633 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27634 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27635 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27636 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27637 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27638 (_mm512_prefetch_i32scatter_ps): Ditto.
27639 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27640 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27641 (_mm512_prefetch_i64scatter_pd): Ditto.
27642 (_mm512_prefetch_i64scatter_ps): Ditto.
27643 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27644 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27645 * config/i386/avx512vlintrin.h
27646 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27647 (_mm_mmask_i32gather_ps): Ditto.
27648 (_mm256_mmask_i32gather_pd): Ditto.
27649 (_mm_mmask_i32gather_pd): Ditto.
27650 (_mm256_mmask_i64gather_ps): Ditto.
27651 (_mm_mmask_i64gather_ps): Ditto.
27652 (_mm256_mmask_i64gather_pd): Ditto.
27653 (_mm_mmask_i64gather_pd): Ditto.
27654 (_mm256_mmask_i32gather_epi32): Ditto.
27655 (_mm_mmask_i32gather_epi32): Ditto.
27656 (_mm256_mmask_i32gather_epi64): Ditto.
27657 (_mm_mmask_i32gather_epi64): Ditto.
27658 (_mm256_mmask_i64gather_epi32): Ditto.
27659 (_mm_mmask_i64gather_epi32): Ditto.
27660 (_mm256_mmask_i64gather_epi64): Ditto.
27661 (_mm_mmask_i64gather_epi64): Ditto.
27662 (_mm256_i32scatter_ps): Change __addr type to void*.
27663 (_mm256_mask_i32scatter_ps): Ditto.
27664 (_mm_i32scatter_ps): Ditto.
27665 (_mm_mask_i32scatter_ps): Ditto.
27666 (_mm256_i32scatter_pd): Ditto.
27667 (_mm256_mask_i32scatter_pd): Ditto.
27668 (_mm_i32scatter_pd): Ditto.
27669 (_mm_mask_i32scatter_pd): Ditto.
27670 (_mm256_i64scatter_ps): Ditto.
27671 (_mm256_mask_i64scatter_ps): Ditto.
27672 (_mm_i64scatter_ps): Ditto.
27673 (_mm_mask_i64scatter_ps): Ditto.
27674 (_mm256_i64scatter_pd): Ditto.
27675 (_mm256_mask_i64scatter_pd): Ditto.
27676 (_mm_i64scatter_pd): Ditto.
27677 (_mm_mask_i64scatter_pd): Ditto.
27678 (_mm256_i32scatter_epi32): Ditto.
27679 (_mm256_mask_i32scatter_epi32): Ditto.
27680 (_mm_i32scatter_epi32): Ditto.
27681 (_mm_mask_i32scatter_epi32): Ditto.
27682 (_mm256_i32scatter_epi64): Ditto.
27683 (_mm256_mask_i32scatter_epi64): Ditto.
27684 (_mm_i32scatter_epi64): Ditto.
27685 (_mm_mask_i32scatter_epi64): Ditto.
27686 (_mm256_i64scatter_epi32): Ditto.
27687 (_mm256_mask_i64scatter_epi32): Ditto.
27688 (_mm_i64scatter_epi32): Ditto.
27689 (_mm_mask_i64scatter_epi32): Ditto.
27690 (_mm256_i64scatter_epi64): Ditto.
27691 (_mm256_mask_i64scatter_epi64): Ditto.
27692 (_mm_i64scatter_epi64): Ditto.
27693 (_mm_mask_i64scatter_epi64): Ditto.
27694 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27695 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27696 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27697 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27698 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27699 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27700 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27701 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27702 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27703 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27704 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27705 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27706 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27707 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27708 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27709 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27710 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27711 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27712 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27713 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27714 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27715 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27716 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27717 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27718 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27719 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27720 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27721 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27722 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27723 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27724 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27725 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27726 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27727 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27728 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27729 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27730 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27731 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27732 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27733 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27734 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27735 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27736 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27737 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27738 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27739 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27740 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27741 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27742 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27743 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27744 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27745 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27746 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27747 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27748 definitions accordingly.
27750 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27751 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27754 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27757 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27760 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27761 hard regs for inner regno.
27763 2017-01-17 Martin Liska <mliska@suse.cz>
27766 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27767 assumption and add comment.
27769 2017-01-17 Nathan Sidwell <nathan@acm.org>
27771 * ipa-visibility.c (localize_node): New function, broken out of ...
27772 (function_and_variable_visibility): ... here. Call it.
27774 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27776 PR middle-end/77445
27777 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27778 correctly set frequency of oudgoing edge.
27779 (duplicate_thread_path): Fix profile updating.
27781 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27784 * configure.ac: Add GCC_BASE_VER.
27785 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27786 version from BASE-VER file.
27787 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27788 (gcc.o): Depend on $(BASEVER).
27789 * common.opt (dumpfullversion): New option.
27790 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27791 * doc/invoke.texi: Document -dumpfullversion.
27792 * doc/install.texi: Document --with-gcc-major-version-only.
27793 * configure: Regenerated.
27795 2017-01-17 Richard Biener <rguenther@suse.de>
27797 PR tree-optimization/71433
27798 * tree-vrp.c (register_new_assert_for): Merge same asserts
27799 on all incoming edges.
27800 (process_assert_insertions_for): Handle insertions at the
27803 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27805 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27806 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27808 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27811 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27814 2017-01-17 Alan Modra <amodra@gmail.com>
27817 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27818 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27819 symbolic stack limit when pic.
27821 2017-01-16 Martin Sebor <msebor@redhat.com>
27823 PR tree-optimization/78608
27824 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27826 2017-01-16 Jeff Law <law@redhat.com>
27829 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27830 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27831 for several include directories that may be relative to sysroot.
27832 * config/i386/x-mingw32 (gplus_includedir): Define.
27833 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27834 (native_system_includedir): Likewise.
27835 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27836 override if TARGET_SYSTEM_ROOT is defined.
27837 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27839 PR tree-optimization/79090
27840 PR tree-optimization/33562
27841 PR tree-optimization/61912
27842 PR tree-optimization/77485
27843 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
27844 and computed trims into the dump file.
27846 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27848 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
27850 2017-01-16 Jakub Jelinek <jakub@redhat.com>
27853 * gimplify.c (gimplify_init_constructor): If want_value and
27854 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
27858 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
27859 sequence. Formatting fixes.
27860 (doloop_optimize): Formatting fixes.
27863 * gcc.c (debug_level_greater_than_spec_func): New function.
27864 (static_spec_functions): Add debug-level-gt spec function.
27865 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
27867 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27868 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
27869 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
27870 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
27871 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
27872 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
27874 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
27876 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
27877 QImode fixups to general and mask registers only.
27879 2017-01-16 Carl Love <cel@us.ibm.com>
27881 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
27882 for built-in functions
27883 vector signed char vec_nabs (vector signed char)
27884 vector signed short vec_nabs (vector signed short)
27885 vector signed int vec_nabs (vector signed int)
27886 vector signed long long vec_nabs (vector signed long long)
27887 vector float vec_nabs (vector float)
27888 vector double vec_nabs (vector double)
27889 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
27891 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
27892 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
27893 * doc/extend.texi: Update the documentation file for the new built-in
27896 2017-01-16 Martin Sebor <msebor@redhat.com>
27898 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
27901 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27903 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
27904 UNSPEC_VSX__XXSPLTD to require special splat handling.
27906 2017-01-16 David Malcolm <dmalcolm@redhat.com>
27909 * system.h: Poison strndup.
27911 2017-01-16 Alan Modra <amodra@gmail.com>
27914 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
27917 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
27919 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
27921 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
27923 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
27924 call recog here. Assert that INSN_CODE (insn) is non-negative.
27926 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27929 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
27931 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
27932 in the currently scheduled RTL fragment.
27934 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27936 PR rtl-optimization/78751
27937 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
27940 2017-01-14 Jeff Law <law@redhat.com>
27942 PR tree-optimization/79090
27943 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
27944 variable length stores.
27945 (compute_trims): Delete dead assignment to *trim_tail.
27946 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
27949 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
27951 PR rtl-optimization/78626
27952 PR rtl-optimization/78727
27953 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
27954 of a block, and split such blocks after everything else is finished.
27956 2017-01-14 Alan Modra <amodra@gmail.com>
27959 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
27960 target legitimate_combined_insn.
27961 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
27962 (rs6000_legitimate_combined_insn): New function.
27963 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
27965 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
27966 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
27967 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
27969 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
27971 * doc/frontends.texi (G++ and GCC): Remove references to Java.
27973 2017-01-13 Jeff Law <law@redhat.com>
27975 PR tree-optimization/33562
27976 PR tree-optimization/61912
27977 PR tree-optimization/77485
27978 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
27980 (delete_dead_assignment): Likewise.
27981 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
27982 statement to delete_dead_call and delete_dead_assignment.
27984 2017-01-13 David Malcolm <dmalcolm@redhat.com>
27987 * substring-locations.c (format_warning_va): Strengthen case 1 so
27988 that both endpoints of the substring must be within the format
27989 range for just the substring to be printed.
27991 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
27993 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
27994 * config/i386/i386.c (ix86_target_string): Add missing options
27995 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
27996 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
27997 flags_other and ix86_target_other to flags2_other. Display unknown
27999 (ix86_valid_target_attribute_inner_p): Add missing options and
28000 reorder options by implied ISAs, as in ix86_target_string.
28002 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28004 * hash-table.h (hash_table::too_empty_p): New function.
28005 (hash_table::expand): Use it.
28006 (hash_table::traverse): Likewise.
28007 (hash_table::empty_slot): Use sizeof (value_type) instead of
28008 sizeof (PTR) to convert bytes to elements. Shrink the table
28009 if the current size is excessive for the current number of
28012 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28014 * ira-costs.c (record_reg_classes): Break from the inner loop
28015 early once alt_fail is known to be true. Update outer loop
28016 handling accordingly.
28018 2017-01-13 Jeff Law <law@redhat.com>
28020 * tree-ssa-dse.c (decrement_count): New function.
28021 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28022 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28023 when we know the partially dead statement is a mem* function.
28025 PR tree-optimization/61912
28026 PR tree-optimization/77485
28027 * tree-ssa-dse.c: Include expr.h.
28028 (maybe_trim_constructor_store): New function.
28029 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28031 PR tree-optimization/33562
28032 PR tree-optimization/61912
28033 PR tree-optimization/77485
28034 * doc/invoke.texi: Document new dse-max-object-size param.
28035 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28036 * tree-ssa-dse.c: Include params.h.
28037 (dse_store_status): New enum.
28038 (initialize_ao_ref_for_dse): New, partially extracted from
28040 (valid_ao_ref_for_dse, normalize_ref): New.
28041 (setup_live_bytes_from_ref, compute_trims): Likewise.
28042 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28043 (maybe_trim_partially_dead_store): Likewise.
28044 (maybe_trim_complex_store): Likewise.
28045 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28046 Track what bytes live from the original store. Return tri-state
28047 for dead, partially dead or live.
28048 (dse_dom_walker): Add constructor, destructor and new private members.
28049 (delete_dead_call, delete_dead_assignment): New extracted from
28051 (dse_optimize_stmt): Make a member of dse_dom_walker.
28052 Use initialize_ao_ref_for_dse.
28054 PR tree-optimization/33562
28055 PR tree-optimization/61912
28056 PR tree-optimization/77485
28057 * sbitmap.h (bitmap_count_bits): Prototype.
28058 (bitmap_clear_range, bitmap_set_range): Likewise.
28059 * sbitmap.c (bitmap_clear_range): New function.
28060 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28062 2017-01-13 Martin Liska <mliska@suse.cz>
28065 * function.c (set_cfun): Add new argument force.
28066 * function.h (set_cfun): Likewise.
28067 * ipa-inline-transform.c (inline_call): Use the function when
28068 strict alising from is dropped for function we inline to.
28070 2017-01-13 Richard Biener <rguenther@suse.de>
28072 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28073 for dumping GIMPLE INTEGER_CSTs.
28075 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28077 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28078 to 201112L since C++17.
28080 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28083 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28084 if -fsanitize=kernel-address is present.
28086 2017-01-13 Richard Biener <rguenther@suse.de>
28088 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28089 as _Literal ( type ) number in case usual suffixes do not
28090 preserve all information.
28092 2017-01-13 Richard Biener <rguenther@suse.de>
28094 PR tree-optimization/77283
28095 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28096 and ssa-iterators.h.
28097 (is_feasible_trace): Implement a cost model based on joiner
28100 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28103 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28104 char or short to __float128/_Float128 directly.
28106 2017-01-12 Martin Sebor <msebor@redhat.com>
28108 to -Wformat-overflow.
28109 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28110 (min_bytes_remaining): Same.
28111 (get_string_length): Same.
28112 (format_string): Same.
28113 (format_directive): Same.
28115 (pass_sprintf_length::handle_gimple_call): Same.
28117 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28119 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28120 info.nowrite calls with no lhs that can't throw. Return bool
28121 whether gsi_remove has been called or not.
28122 (pass_sprintf_length::handle_gimple_call): Return bool whether
28123 try_substitute_return_value called gsi_remove. Formatting fix.
28124 (pass_sprintf_length::execute): Don't use gsi_remove if
28125 handle_gimple_call returned true.
28128 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28129 be removed due to side-effects, don't remove following barrier nor
28130 turn the successor edge into fallthru edge.
28132 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28135 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28136 element-reversing loads and stores as not swappable.
28138 2017-01-12 Nathan Sidwell <nathan@acm.org>
28139 Nicolai Stange <nicstange@gmail.com>
28141 * combine.c (try_combine): Don't ignore result of overlap checking
28142 loop. Combine overlap & asm check into single loop.
28144 2017-01-12 Richard Biener <rguenther@suse.de>
28146 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28147 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28149 2017-01-12 Richard Biener <rguenther@suse.de>
28151 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28152 and TS_TARGET_OPTION directly derive from TS_BASE.
28153 * tree-core.h (tree_optimization_option): Derive from tree_base.
28154 (tree_target_option): Likewise.
28156 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28158 * config/i386/i386.c (memory_address_length): Increase len
28159 only when rip_relative_addr_p returns false.
28161 2017-01-11 Julia Koval <julia.koval@intel.com>
28163 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28164 (OPTION_MASK_ISA_SGX_SET): New.
28165 (ix86_handle_option): Handle OPT_msgx.
28166 * config.gcc: Added sgxintrin.h.
28167 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28168 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28169 * config/i386/i386.c (ix86_target_string): Add -msgx.
28171 (ix86_option_override_internal): Handle new options.
28172 (ix86_valid_target_attribute_inner_p): Add sgx.
28173 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28174 * config/i386/i386.opt: Add msgx.
28175 * config/i386/sgxintrin.h: New file.
28176 * config/i386/x86intrin.h: Add sgxintrin.h.
28178 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28181 * fold-const.c (maybe_nonzero_address): Return 1 for function
28183 (tree_single_nonzero_warnv_p): Don't handle function local objects
28187 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28190 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28193 * opts.c: Include "spellcheck.h"
28194 (struct string_fragment): New struct.
28195 (struct edit_distance_traits<const string_fragment &>): New
28197 (get_closest_sanitizer_option): New function.
28198 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28200 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28202 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28204 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28205 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28206 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28207 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28208 for initial die_offset if dwarf_split_debug_info.
28209 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28210 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28212 (output_skeleton_debug_sections): Formatting fix. Use
28213 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28214 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28216 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28218 * config/arm/cortex-a53.md: Add bypasses for
28219 cortex_a53_r2f_cvt.
28220 (cortex_a53_r2f): Only use for transfers.
28221 (cortex_a53_f2r): Likewise.
28222 (cortex_a53_r2f_cvt): Add reservation for conversions.
28223 (cortex_a53_f2r_cvt): Likewise.
28225 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28227 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28228 to all inlined functions, change static to extern.
28230 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28233 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28235 (arm_assemble_integer): Likewise.
28237 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28239 * config.gcc: Use new awk script to check CPU, FPU and architecture
28240 parameters for --with-... options.
28241 * config/arm/parsecpu.awk: New file
28242 * config/arm/arm-cpus.in: New file.
28243 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28245 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28247 * config/arm/t-arm: Update dependency rules.
28248 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28249 of processing .def files.
28250 * config/arm/genopt.sh: Deleted.
28251 * config/arm/gentune.sh: Deleted.
28252 * config/arm/arm-cores.def: Deleted.
28253 * config/arm/arm-arches.def: Deleted.
28254 * config/arm/arm-fpus.def: Deleted.
28255 * config/arm/arm-tune.md: Regenerated.
28256 * config/arm/arm-tables.opt: Regenerated.
28257 * config/arm/arm-cpu.h: New generated file.
28258 * config/arm/arm-cpu-data.h: New generated file.
28259 * config/arm/arm-cpu-cdata.h: New generated file.
28261 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28264 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28266 (input_varpool_node): Unpack dynamically_initialized bit.
28268 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28270 PR rtl-optimization/79032
28271 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28272 the alignment of the adjusted memory reference against that of MODE,
28273 instead of the alignment of the original memory reference.
28275 2017-01-11 Martin Jambor <mjambor@suse.cz>
28277 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28279 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28280 decorated functions.
28282 2017-01-11 Richard Biener <rguenther@suse.de>
28284 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28285 set range/nonnull info for PHI results. Do not set it on
28286 stmts marked for removal.
28288 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28290 * expr.c (store_field): In the bitfield case, fetch the return value
28291 from the registers before applying a single big-endian adjustment.
28292 Always do a final load for a BLKmode value not larger than a word.
28294 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28297 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28298 that we correctly handle column numbers greater than
28299 LINE_MAP_MAX_COLUMN_NUMBER.
28301 2017-01-10 Martin Sebor <msebor@redhat.com>
28303 PR middle-end/78245
28304 * gimple-ssa-sprintf.c (get_destination_size): Call
28305 {init,fini}object_sizes.
28306 * tree-object-size.c (addr_object_size): Adjust.
28307 (pass_through_call): Adjust.
28308 (pass_object_sizes::execute): Adjust.
28309 * tree-object-size.h (fini_object_sizes): Declare.
28311 2017-01-10 Martin Sebor <msebor@redhat.com>
28313 PR tree-optimization/78775
28314 * builtins.c (get_size_range): Move...
28315 * calls.c: ...to here.
28316 (alloc_max_size): Accept zero argument.
28317 (operand_signed_p): Remove.
28318 (maybe_warn_alloc_args_overflow): Call get_size_range.
28319 * calls.h (get_size_range): Declare.
28321 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28323 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28324 from TI's devices.csv file as of September 2016.
28325 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28327 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28329 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28330 * doc/invoke.texi: Likewise.
28331 * doc/md.texi: Likewise.
28332 * doc/objc.texi: Likewise.
28334 2017-01-10 Joshua Conner <joshconner@google.com>
28336 * config/arm/fuchsia-elf.h: New file.
28337 * config/fuchsia.h: New file.
28338 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28339 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28341 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28343 2016-01-10 Richard Biener <rguenther@suse.de>
28345 PR tree-optimization/79034
28346 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28347 Propagate out degenerate PHIs in the joiner.
28349 2017-01-10 Martin Liska <mliska@suse.cz>
28351 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28352 (sort_congruence_classes_by_decl_uid): Likewise.
28353 (sort_congruence_class_groups_by_decl_uid): Likewise.
28354 (sem_item_optimizer::merge_classes): Sort class, groups in these
28355 classes and members in the groups by DECL_UID of declarations.
28356 This would make merge operations stable.
28358 2017-01-10 Martin Liska <mliska@suse.cz>
28360 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28361 usage of m_classes_vec.
28362 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28363 (sem_item_optimizer::get_group_by_hash): Likewise.
28364 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28365 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28366 (sem_item_optimizer::verify_classes): Likewise.
28367 (sem_item_optimizer::process_cong_reduction): Likewise.
28368 (sem_item_optimizer::dump_cong_classes): Likewise.
28369 (sem_item_optimizer::merge_classes): Likewise.
28370 * ipa-icf.h (congruence_class_hash): Rename from
28371 congruence_class_group_hash. Remove declaration of m_classes_vec.
28373 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28375 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28376 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28377 * config.gcc: Add avx512vpopcntdqintrin.h.
28378 * config/i386/avx512vpopcntdqintrin.h: New.
28379 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28380 * config/i386/i386-builtin-types.def: Add new types.
28381 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28382 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28383 __builtin_ia32_vpopcountq_v8di_mask): New.
28384 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28385 __AVX512VPOPCNTDQ__.
28386 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28387 (PTA_AVX512VPOPCNTDQ): Define.
28388 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28389 TARGET_AVX512VPOPCNTDQ_P): Define.
28390 * config/i386/i386.opt: Add mavx512vpopcntdq.
28391 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28392 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28394 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28396 PR middle-end/77484
28397 * predict.def (PRED_CALL): Set to 67.
28399 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28401 * expr.c (store_field): In the bitfield case, if the value comes from
28402 a function call and is of an aggregate type returned in registers, do
28403 not modify the field mode; extract the value in all cases if the mode
28404 is BLKmode and the size is not larger than a word.
28406 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28409 * config/i386/cpuid.h: Fix undefined behavior.
28411 2017-01-04 Jeff Law <law@redhat.com>
28413 PR tree-optimization/79007
28414 PR tree-optimization/67955
28415 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28416 conservative for pt.null when flag_non_call_exceptions is on.
28418 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28420 PR translation/79019
28421 PR translation/79020
28422 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28423 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28424 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28426 * config/avr/avr.opt (maccumulate-args): Likewise.
28427 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28428 * common.opt (freport-bug): Likewise.
28429 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28430 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28431 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28432 translatable string.
28433 * config/i386/i386.c (function_value_32): Likewise.
28434 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28435 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28437 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28438 * common/config/msp430/msp430-common.c (msp430_handle_option):
28440 * symtab.c (symtab_node::verify_base): Likewise.
28441 * opts.c (set_debug_level): Likewise.
28442 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28443 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28444 missing whitespace to translatable strings.
28445 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28446 * config/sh/superh.h: Likewise.
28447 * config/i386/xopintrin.h: Likewise.
28448 * config/i386/znver1.md: Likewise.
28449 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28450 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28451 * double-int.h (struct double_int): Likewise.
28452 * double-int.c (div_and_round_double): Likewise.
28453 * wide-int.cc: Likewise.
28454 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28455 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28456 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28457 (crossjumps_occurred): ... this.
28458 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28461 PR tree-optimization/78899
28462 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28463 returning bool return struct loop *, NULL for failure and the new
28465 (versionable_outer_loop_p): Don't version outer loop if it has
28466 dont_vectorized bit set.
28467 (tree_if_conversion): When versioning outer loop, ensure
28468 tree_if_conversion is performed also on the inner loop of the
28469 non-vectorizable outer loop copy.
28470 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28471 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28472 prevent vectorization of it.
28473 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28474 the outer loop vectorization of the non-scalar version is attempted
28475 before vectorization of the inner loop in scalar version. If
28476 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28477 vectorization of its inner loop.
28478 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28479 has 2 inner loops, rename also on edges from bb whose single pred
28480 is outer_loop->header. Fix typo in function comment.
28482 2017-01-09 Martin Sebor <msebor@redhat.com>
28485 * asan.c (asan_emit_stack_protection): Increase local buffer size
28486 to avoid snprintf truncation warning.
28488 2017-01-09 Andrew Pinski <apinski@cavium.com>
28490 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28491 to reference thunderx2t99 for the tuning structure
28492 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28494 (thunderx2t99_extra_costs): This.
28495 * config/aarch64/aarch64-tune.md: Regenerate.
28496 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28497 (vulcan_addrcost_table): This.
28498 (vulcan_regmove_cost): Rename to ...
28499 (thunderx2t99_regmove_cost): This.
28500 (vulcan_vector_cost): Rename to ...
28501 (thunderx2t99_vector_cost): this.
28502 (vulcan_branch_cost): Rename to ...
28503 (thunderx2t99_branch_cost): This.
28504 (vulcan_tunings): Rename to ...
28505 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28506 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28508 2017-01-09 Martin Jambor <mjambor@suse.cz>
28512 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28513 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28514 (propagate_vr_accross_jump_function): Use the above function for all
28515 value range computations for pass-through jump functions and type
28516 converasion from explicit value range values.
28517 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28518 parameters from TYPE_ARG_TYPES.
28519 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28520 (ipa_write_node_info): Stream type of the actual argument.
28521 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28523 2017-01-09 Martin Liska <mliska@suse.cz>
28526 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28527 (lookup_compiler): Do not show error message with have_E.
28529 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28531 PR tree-optimization/78938
28532 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28533 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28534 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28535 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28538 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28540 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28543 2017-01-09 Richard Biener <rguenther@suse.de>
28545 PR tree-optimization/78997
28546 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28547 name condition properly.
28549 2017-01-09 Richard Biener <rguenther@suse.de>
28552 * dwarf2out.c (is_cxx): New overload with context.
28553 (is_naming_typedef_decl): Use it.
28555 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28557 * invoke.texi (Option Summary): Correct spacing in option lists
28558 and add line breaks to fix over-long lines.
28560 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28562 PR middle-end/17660
28564 * extend.texi (Common Variable Attributes): Add xref to GCC
28565 Internals manual to explain mode attribute keywords.
28567 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28570 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28571 and Preprocessor Options.
28572 (Options for Linking): Document -pthread here....
28573 (RS/6000 and PowerPC Options): ...not here.
28574 (Solaris 2 Options): ...or here.
28575 * doc/cppopts.texi: Document -pthread.
28577 2017-01-08 Martin Sebor <msebor@redhat.com>
28579 PR middle-end/77708
28580 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28581 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28582 New member functions.
28583 (format_directive): Used them.
28585 (pass_sprintf_length::handle_gimple_call): Same.
28586 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28587 to avoid truncation for any argument.
28588 (extract_affine_mul): Same.
28589 * tree.c (get_file_function_name): Same.
28591 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28593 PR middle-end/77484
28594 * predict.def (PRED_INDIR_CALL): Set to 86.
28596 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28598 PR preprocessor/54124
28599 * doc/cppopts.texi: Reformat -d subtable to list the full name
28600 of the options. Add cross-reference to the docs for the general
28601 compiler -d options.
28602 * doc/invoke.texi (Developer Options): Add cross-reference to the
28603 preprocessor-specific -d option documentation.
28605 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28607 PR preprocessor/13498
28608 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28609 redudant material, and reflect new command-line options.
28610 (System Headers): Likewise.
28612 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28614 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28615 -isystem, and -idirafter. Copy-edit.
28616 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28617 default for -ftrack-macro-expansion. Delete obsolete and
28618 badly-formatted implementation details about -fdebug-cpp output.
28619 * doc/cppwarnopts.texi: Copy-edit.
28621 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28624 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28625 that the transition from a max line width >= 1<<10 to narrower
28626 lines works correctly.
28628 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28630 * doc/options.texi (PerFunction): New.
28631 * opt-functions.awk (switch_flags): Map both Optimization and
28632 PerFunction to CL_OPTIMIZATION.
28633 * opth-gen.awk: Test for PerFunction flag along with
28635 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28636 it only when the latter is present. Skip those that don't in
28637 the hash function generator.
28638 * common.opt (fvar-tracking): Mark as PerFunction instead of
28640 (fvar-tracking-assignments): Likewise.
28641 (fvar-tracking-assignments-toggle): Likewise.
28642 (fvar-tracking-uninit): Likewise.
28644 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28646 PR translation/79018
28647 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28650 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28653 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28654 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28655 TARGET_LONG_JUMP_TABLE_OFFSETS.
28656 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28657 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28658 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28659 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28660 * config/m68k/m68k.md (tablejump expander): Likewise.
28661 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28662 TARGET_LONG_JUMP_TABLE_OFFSETS.
28663 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28664 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28666 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28667 David Holsgrove <david.holsgrove@xilinx.com>
28669 * common/config/microblaze/microblaze-common.c
28670 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28671 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28673 * config/microblaze/microblaze.c (microblaze_must_save_register)
28674 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28676 (microblaze_eh_return): New function.
28677 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28678 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28679 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28680 * config/microblaze/microblaze.md (eh_return): New pattern.
28682 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28684 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28685 GCC_DIAGNOSTIC_STRINGIFY): Define.
28687 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28689 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28691 * config/arm/arm.md (<mcrr>): New.
28693 * config/arm/arm.c (arm_arch5te): New.
28694 (arm_option_override): Set arm_arch5te.
28695 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28697 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28698 (arm_mcrr_qualifiers): ... this. New.
28699 (MRRC_QUALIFIERS): Define to...
28700 (arm_mrrc_qualifiers): ... this. New.
28701 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28703 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28704 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28705 (MRRCI, mrrc, MRRC): New.
28706 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28707 VUNSPEC_MRRC2): New.
28709 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28711 * config/arm/arm.md (<mcr>): New.
28713 * config/arm/arm.c (arm_coproc_builtin_available): Add
28714 support for mcr, mrc, mcr2 and mrc2.
28715 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28716 (arm_mcr_qualifiers): ... this. New.
28717 (MRC_QUALIFIERS): Define to ...
28718 (arm_mrc_qualifiers): ... this. New.
28719 (MCR_QUALIFIERS): Define to ...
28720 (arm_mcr_qualifiers): ... this. New.
28721 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28723 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28724 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28725 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28726 VUNSPEC_MRC2): New.
28728 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28730 * config/arm/arm.md (*ldc): New.
28734 * config/arm/arm.c (arm_coproc_builtin_available): Add
28735 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28736 (arm_coproc_ldc_stc_legitimate_address): New.
28737 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28738 'qualifier_const_pointer'.
28739 (LDC_QUALIFIERS): Define to...
28740 (arm_ldc_qualifiers): ... this. New.
28741 (STC_QUALIFIERS): Define to...
28742 (arm_stc_qualifiers): ... this. New.
28743 * config/arm/arm-protos.h
28744 (arm_coproc_ldc_stc_legitimate_address): New.
28745 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28746 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28747 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28748 stc2, stcl, stc2l): New.
28749 * config/arm/constraints.md (Uz): New.
28750 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28751 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28752 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28753 VUNSPEC_STC2L): New.
28755 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28757 * config/arm/arm.md (<cdp>): New.
28758 * config/arm/arm.c (neon_const_bounds): Rename this ...
28759 (arm_const_bounds): ... this.
28760 (arm_coproc_builtin_available): New.
28761 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28762 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28763 (CDP_QUALIFIERS): Define to...
28764 (arm_cdp_qualifiers): ... this. New.
28766 (arm_expand_builtin_args): Add case for 6 arguments.
28767 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28768 (arm_const_bounds): ... this.
28769 (arm_coproc_builtin_available): New.
28770 * config/arm/arm_acle.h (__arm_cdp): New.
28772 * config/arm/arm_acle_builtins.def (cdp): New.
28774 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28775 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28776 'arm_const_bounds'.
28777 * config/arm/types.md (coproc): New.
28778 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28779 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28780 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28781 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28783 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28785 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28786 (UBINOP_QUALIFIERS): New.
28788 (acle_builtin_data): New. Change comment.
28789 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28790 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28791 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28792 arm_acle_builtins.def.
28793 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28794 (arm_init_acle_builtins): New.
28795 (CRC32_BUILTIN): Remove.
28796 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28797 crc32cb, crc32ch and crc32cw.
28798 (arm_init_crc32_builtins): Remove.
28799 (arm_init_builtins): Use arm_init_acle_builtins rather
28800 than arm_init_crc32_builtins.
28801 (arm_expand_acle_builtin): New.
28802 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28803 * config/arm/arm_acle_builtins.def: New.
28805 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28807 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28808 (arm_builtin_datum): ... this.
28809 (arm_init_neon_builtin): Rename to ...
28810 (arm_init_builtin): ... this. Add a new parameters PREFIX
28811 and USE_SIG_IN_NAME.
28812 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28813 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28814 'arm_builtin_datum'.
28815 (arm_init_vfp_builtins): Likewise.
28816 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28817 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28818 (arm_expand_neon_args): Rename to ...
28819 (arm_expand_builtin_args): ... this. Rename builtin_arg
28820 enum values and differentiate between ARG_BUILTIN_MEMORY
28821 and ARG_BUILTIN_NEON_MEMORY.
28822 (arm_expand_neon_builtin_1): Rename to ...
28823 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28824 values, arm_expand_builtin_args and add bool parameter NEON.
28825 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28826 (arm_expand_vfp_builtin): Likewise.
28827 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28829 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28831 PR middle-end/77484
28832 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28833 * predict.c (tree_estimate_probability_bb): Reverse direction of
28834 polymorphic call predictor.
28836 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28838 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28839 (determine_pass_name_match): ...this new function and...
28840 (should_skip_pass_p): ...this new function.
28842 2017-01-06 Nathan Sidwell <nathan@acm.org>
28844 * ipa-visibility.c (function_and_variable_visibility): Reformat
28845 comments and long lines. Remove extrneous if.
28846 * symtab.c (symtab_node::make_decl_local): Fix code format.
28847 (symtab_node::set_section_for_node): Fix comment typo.
28849 2017-01-06 Martin Liska <mliska@suse.cz>
28852 * lra-constraints.c: Rename invariant to lra_invariant.
28853 * predict.c (set_even_probabilities): Initialize e to NULL.
28855 2017-01-05 Martin Sebor <msebor@redhat.com>
28857 PR tree-optimization/78910
28858 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
28859 (format_integer): Correct off-by-one error in the handling
28860 of precision with negative numbers in signed conversions..
28862 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
28864 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
28866 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28868 PR tree-optimization/71016
28869 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
28870 factor_out_conditional_conversion. Formatting fix.
28871 (factor_out_conditional_conversion): Add cond_stmt argument.
28872 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
28873 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
28876 2017-01-05 David Malcolm <dmalcolm@redhat.com>
28878 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
28879 read-rtl-function.o, and selftest-rtl.o.
28880 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
28881 (selftest::aarch64_test_loading_full_dump): New function.
28882 (selftest::aarch64_run_selftests): New function.
28883 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
28884 selftest::aarch64_run_selftests.
28885 * config/i386/i386.c
28886 (selftest::ix86_test_loading_dump_fragment_1): New function.
28887 (selftest::ix86_test_loading_call_insn): New function.
28888 (selftest::ix86_test_loading_full_dump): New function.
28889 (selftest::ix86_test_loading_unspec): New function.
28890 (selftest::ix86_run_selftests): Call the new functions.
28891 * emit-rtl.c (maybe_set_max_label_num): New function.
28892 * emit-rtl.h (maybe_set_max_label_num): New decl.
28893 * function.c (instantiate_decls): Guard call to
28894 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
28895 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
28897 * gensupport.c (gen_reader::gen_reader): Pass "false"
28898 for new "compact" param of rtx_reader.
28899 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
28900 rather than an empty string for NULL strings.
28901 * read-md.c: Potentially include config.h rather than bconfig.h.
28902 Wrap include of errors.h with #ifdef GENERATOR_FILE.
28903 (have_error): New global, copied from errors.c.
28904 (md_reader::read_name): Rename to...
28905 (md_reader::read_name_1): ...this, adding "out_loc" param,
28906 and converting "missing name or number" to returning false, rather
28908 (md_reader::read_name): Reimplement in terms of read_name_1.
28909 (md_reader::read_name_or_nil): New function.
28910 (md_reader::read_string): Handle "(nil)" by returning NULL.
28911 (md_reader::md_reader): Add new param "compact".
28912 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
28913 (md_reader::read_file): New method.
28914 * read-md.h (md_reader::md_reader): Add new param "compact".
28915 (md_reader::read_file): New method.
28916 (md_reader::is_compact): New accessor.
28917 (md_reader::read_name): Convert return type from void to file_location.
28918 (md_reader::read_name_or_nil): New decl.
28919 (md_reader::read_name_1): New decl.
28920 (md_reader::m_compact): New field.
28921 (noop_reader::noop_reader): Pass "false" for new "compact" param
28923 (rtx_reader::rtx_reader): Add new "compact" param.
28924 (rtx_reader::read_rtx_operand): Make virtual and convert return
28925 type from void to rtx.
28926 (rtx_reader::read_until): New decl.
28927 (rtx_reader::handle_any_trailing_information): New virtual function.
28928 (rtx_reader::postprocess): New virtual function.
28929 (rtx_reader::finalize_string): New virtual function.
28930 (rtx_reader::m_in_call_function_usage): New field.
28931 (rtx_reader::m_reuse_rtx_by_id): New field.
28932 * read-rtl-function.c: New file.
28933 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
28934 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
28935 (selftest::verify_three_block_rtl_cfg): New decl.
28936 * read-rtl-function.h: New file.
28937 * read-rtl.c: Potentially include config.h rather than bconfig.h.
28938 For host, include function.h, memmodel.h, and emit-rtl.h.
28939 (one_time_initialization): New function.
28940 (struct compact_insn_name): New struct.
28941 (compact_insn_names): New array.
28942 (find_code): Handle insn codes in compact dumps.
28943 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
28944 (bind_subst_iter_and_attr): Likewise.
28945 (add_condition_to_string): Likewise.
28946 (add_condition_to_rtx): Likewise.
28947 (apply_attribute_uses): Likewise.
28948 (add_current_iterators): Likewise.
28949 (apply_iterators): Likewise.
28950 (initialize_iterators): Guard usage of apply_subst_iterator with
28951 #ifdef GENERATOR_FILE.
28952 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
28953 (md_reader::read_mapping): Likewise.
28954 (add_define_attr_for_define_subst): Likewise.
28955 (add_define_subst_attr): Likewise.
28956 (read_subst_mapping): Likewise.
28957 (check_code_iterator): Likewise.
28958 (rtx_reader::read_rtx): Likewise. Move one-time initialization
28960 (one_time_initialization): New function.
28961 (rtx_reader::read_until): New method.
28962 (read_flags): New function.
28963 (parse_reg_note_name): New function.
28964 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
28965 Handle reuse_rtx ids.
28966 Wrap iterator lookup within #ifdef GENERATOR_FILE.
28967 Add parsing support for RTL dumps, mirroring the special-cases in
28968 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
28969 values, and calling handle_any_trailing_information.
28970 (rtx_reader::read_rtx_operand): Convert return type from void
28971 to rtx, returning return_rtx. Handle case 'e'. Call
28972 finalize_string on XSTR and XTMPL fields.
28973 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
28974 "(nil)" values were omitted. Call the postprocess vfunc on the
28976 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
28977 class ctor. Initialize m_in_call_function_usage. Call
28978 one_time_initialization.
28979 * rtl-tests.c (selftest::test_uncond_jump): Call
28980 set_new_first_and_last_insn.
28981 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
28982 * selftest-rtl.c: New file.
28983 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
28984 (selftest::get_insn_by_uid): New decl.
28985 * selftest-run-tests.c (selftest::run_tests): Call
28986 read_rtl_function_c_tests.
28987 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
28988 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
28991 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
28993 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
28994 operands in a special way. Assert that pos+len <= mode precision.
28996 2017-01-05 Jakub Jelinek <jakub@redhat.com>
28998 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
28999 3 argument Alias with unlimited for the negative form.
29000 (fno-vect-cost-model): Removed.
29002 2017-01-05 Martin Liska <mliska@suse.cz>
29004 * hsa-gen.c (gen_hsa_divmod): New function.
29005 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29007 2017-01-05 Martin Liska <mliska@suse.cz>
29010 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29013 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29015 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29016 small constant length operands.
29018 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29020 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29021 between loop iterations.
29023 2017-01-05 Martin Liska <mliska@suse.cz>
29026 * gimplify.c (gimplify_decl_expr): Compare to
29027 asan_poisoned_variables instread of checking flags.
29028 (gimplify_target_expr): Likewise.
29029 (gimplify_expr): Likewise.
29030 (gimplify_function_tree): Conditionally initialize
29031 asan_poisoned_variables.
29033 2017-01-04 Jeff Law <law@redhat.com>
29035 PR tree-optimizatin/78812
29036 * rtl.h (contains_mem_rtx_p): Prototype.
29037 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29038 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29039 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29040 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29041 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29043 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29045 * input.c (assert_char_at_range): Default-initialize actual_range.
29047 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29049 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29050 to match the caller.
29052 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29054 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29055 insns after final jump in test to emit dummy move.
29057 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29059 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29060 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29062 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29064 * multiple_target.c (create_dispatcher_calls): Init e_next.
29065 * tree-ssa-loop-split.c (split_loop): Init border.
29066 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29069 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29074 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29075 (altivec_register_operand): Do not return true if the operand
29076 contains a SUBREG mixing SImode and SFmode.
29077 (vsx_register_operand): Likewise.
29078 (vsx_reg_sfsubreg_ok): New predicate.
29079 (vfloat_operand): Do not return true if the operand contains a
29080 SUBREG mixing SImode and SFmode.
29081 (vint_operand): Likewise.
29082 (vlogical_operand): Likewise.
29083 (gpc_reg_operand): Likewise.
29084 (int_reg_operand): Likewise.
29085 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29086 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29087 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29089 (rs6000_emit_move_si_sf_subreg): New helper function.
29090 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29091 fixup SUBREGs involving SImode and SFmode.
29092 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29093 numbers for the new peephole2 optimization.
29094 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29095 the GLIBC math library that do AND/IOR/XOR operations on single
29096 precision floating point.
29097 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29098 target macros to say whether we need to avoid SUBREGs mixing
29100 (TARGET_ALLOW_SF_SUBREG): Likewise.
29101 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29102 (UNSPEC_SI_FROM_SF): Likewise.
29103 (iorxor): Change spacing.
29104 (and_ior_xor): New iterator for AND, IOR, and XOR.
29105 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29106 (movdi_from_sf_zero_ext): Likewise.
29107 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29108 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29109 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29110 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29111 (fms<mode>4): Likewise.
29112 (fnma<mode>4): Likewise.
29113 (fnms<mode>4): Likewise.
29114 (nfma<mode>4): Likewise.
29115 (nfms<mode>4): Likewise.
29117 2017-01-04 Marek Polacek <polacek@redhat.com>
29120 * doc/invoke.texi: Document -Wpointer-compare.
29122 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29124 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29127 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29128 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29131 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29134 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29135 documentation of the powerpc_popcntb_ok attribute.
29136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29137 code to issue warning messages if a requested CPU configuration is
29138 not supported by the binary (assembler and loader) toolchain.
29139 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29140 made to define a built-in function that has been disabled.
29141 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29142 made to define a built-in function that has been disabled.
29143 (altivec_init_builtins): Add comment explaining why definition
29144 of the DST built-in functions is not preceded by an assertion
29145 check. Add assertions to prevent ICE if attempts are made to
29146 define an altivec predicate or an abs* built-in function that has
29148 (htm_init_builtins): Add comment explaining why definition of the
29149 htm built-in functions is not preceded by an assertion check.
29151 2017-01-04 Jeff Law <law@redhat.com>
29153 PR tree-optimizatin/67955
29154 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29155 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29156 the points-to solution does not include pt_null. Use DECL_PT_UID
29159 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29161 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29162 Use gen_int_mode instead of gen_lopwart for const_int operands.
29164 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29166 PR tree-optimization/71563
29167 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29168 out of range value - has low bits known to be zero.
29170 2017-01-04 Alan Modra <amodra@gmail.com>
29172 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29173 * configure: Regenerate.
29174 * config.in: Regenerate.
29176 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29179 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29180 a substring of the message, but strcmp with the whole message. Ifdef
29181 ENABLE_NLS, translate the message first using dgettext.
29183 2017-01-03 Jeff Law <law@redhat.com>
29185 PR tree-optimizatin/78856
29186 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29187 (mark_threaded_blocks): Remove code to truncate thread paths that
29188 cross multiple loop headers. Instead invalidate the cached loop
29189 iteration information and handle case of a thread path walking
29190 into an irreducible region.
29192 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29195 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29196 assertions. Add support for doing the signbit if the IEEE 128-bit
29197 floating point value is in a GPR.
29198 * config/rs6000/rs6000.md (Fsignbit): Delete.
29199 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29200 Update the length attribute if the value is in a GPR.
29201 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29202 the sign or zero extension instruction, since the value is always 0/1.
29203 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29206 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29207 extracting SImode to a GPR register so that we can generate a
29208 store, limit the vector to be in a traditional Altivec register
29209 for the vextuwrx instruction.
29211 2017-01-03 Ian Lance Taylor <iant@google.com>
29213 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29215 2017-01-03 Martin Sebor <msebor@redhat.com>
29217 PR tree-optimization/78696
29218 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29219 precision. Use MPFR for %f for greater fidelity. Correct handling
29221 (pass_sprintf_length::compute_format_length): Set width and precision
29222 specified by asrerisk to void_node for vararg functions.
29223 (try_substitute_return_value): Adjust dump output.
29225 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29227 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29229 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29231 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29232 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29233 -mlra/-mno-lra was passed to the compiler.
29235 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29237 PR rtl-optimization/65618
29238 * emit-rtl.c (try_split): Move initialization of "before" and
29239 "after" to just before the call to emit_insn_after_setloc.
29241 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29243 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29245 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29247 * dwarf2out.c (gen_enumeration_type_die): When
29248 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29250 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29252 PR tree-optimization/78965
29253 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29254 Change first argument from const call_info & to call_info &. For %n
29255 set info.nowrite to false.
29257 PR middle-end/78901
29258 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29259 possibly throwing calls.
29261 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29262 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29263 and fns handling, rather than in a separate case SSA_NAME.
29265 2017-01-02 Jeff Law <law@redhat.com>
29267 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29268 fixes for first_period and second_period variables.
29270 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29273 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29274 (*insvqi_1): New insn pattern.
29275 (*insvqi_1_mem_rex64): Ditto.
29276 (*insvqi_2): Ditto.
29277 (*insvqi_3): Rename from *insvqi.
29279 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29281 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29283 * doc/cfg.texi (Edges): Remove reference to Java.
29284 (Maintaining the CFG): Ditto.
29286 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29288 PR middle-end/77674
29289 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29290 transparent aliases.
29292 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29294 PR middle-end/77484
29295 * predict.def (PRED_CALL): Update hitrate.
29296 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29297 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29298 into direct/indirect/polymorphic variants.
29300 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29302 Update copyright years.
29304 * gcc.c (process_command): Update copyright notice dates.
29305 * gcov-dump.c (print_version): Ditto.
29306 * gcov.c (print_version): Ditto.
29307 * gcov-tool.c (print_version): Ditto.
29308 * gengtype.c (create_file): Ditto.
29309 * doc/cpp.texi: Bump @copying's copyright year.
29310 * doc/cppinternals.texi: Ditto.
29311 * doc/gcc.texi: Ditto.
29312 * doc/gccint.texi: Ditto.
29313 * doc/gcov.texi: Ditto.
29314 * doc/install.texi: Ditto.
29315 * doc/invoke.texi: Ditto.
29317 Copyright (C) 2017 Free Software Foundation, Inc.
29319 Copying and distribution of this file, with or without modification,
29320 are permitted in any medium without royalty provided the copyright
29321 notice and this notice are preserved.